IoT, Code, Security and Server Stuff

Views are my own and not my employer's.

Personal Development Blog...

Coding for fun since 1996, Learn by doing and sharing.

Buy a domain name, then create your own server (get $25 free credit)

View all of my posts.

CDN

I moved my domain to UpCloud (on the other side of the world) from Vultr (Sydney) and could not be happier with the performance.

by

I moved my domain to UpCloud (on the other side of the world) from Vultr (Sydney) and could not be happier with the performance. Here is what I did to set up a complete Ubuntu 18.04 system (NGINX, PHP, MySQL, WordPress etc). This is not a paid review (just me documenting my steps over 2 days).

Background (CPanel hosts)

In 1999 I hosted my first domain (www.fearby.com) on a host in Seattle (for $10 USD a month), the host used CPanel and all was good.  After a decade I was using the domain more for online development and the website was now too slow (I think I was on dial-up or ADSL 1 at the time). I moved my domain to an Australian host (for $25 a month).

After 8 years the domain host was sold and performance remained mediocre. After another year the new host was sold again and performance was terrible.

I started receiving Resource Limit Is Reached warnings (basically this was a plot by the new CPanel host to say “Pay us more and this message will go away”).

Page load times were near 30 seconds.

cpenal_usage_exceeded

The straw that broke the camel’s back was their demand of $150/year for a dodgy SSL certificate.

I needed to move to a self-managed server where I was in control.

Buying a Domain Name

Buy a domain name from Namecheap here.

Domain names for just 88 cents!

Self Managed Server

I found a good web IDE ( http://www.c9.io/ ) that allowed me to connect to a cloud VM.  C9 allowed me to open many files and terminal windows and reconnect to them later. Don’t get excited, though, as AWS have purchased C9 and it’s not the same.

C9 IDE

C9 IDE

I spun up a Digital Ocean Server at the closest data centre in Singapore. Here was my setup guide creating a Digital Ocean VM, connecting to it with C9 and configuring it. I moved my email to G Suite and moved my WordPress to Digital Ocean (other guides here and here).

I was happy since I could now send emails via CLI/code, set up free SSL certs, add second domain email to G Suite and Secure G Suite. No more usage limit errors either.

Self-managing servers requires more work but it is more rewarding (flexible, faster and cheaper).  Page load times were now near 20 seconds (10-second improvement).

Latency Issue

Over 6 months, performance on Digital Ocean (in Singapore) from Australia started to drop (mentioned here).  I tried upgrading the memory but that did not help (latency was king).

Moved the website to Australia

I moved my domain to Vultr in Australia (guide here and here). All was good for a year until traffic growth started to increase.

Blog Growth

I tried upgrading the memory on Vultr and I setup PHP child workers, setup Cloudflare.

GT Metrix scores were about a “B” and Google Page Speed Scores were in the lower 40’s. Page loads were about 14 seconds (5-second improvement).

Tweaking WordPress

I set up an image compression plugin in WordPress then set up a cloud image compression and CDN Plugin from the same vendor.  Page Speed info here.

GT Metrix scores were now occasionally an “A” and Page Speed Scores were in the lower 20’s. Page loads were about 3-5 seconds (10-second improvement).

Mixed bag from Vultr (more optimisation and performance improvements were needed).

This screenshot is showing poor www.gtmetrix.com scores , pool google page speed index scores and upgrading from 1GB to 2GB memory on my server.

Google Chrome Developer Console Audit Results on Vultr hosted website were not very good (I stopped checking as nothing helped).

This is a screenshot showing poor site performance (screenshot taken in Google Dev tools audit feature)

The problem was the Vultr server (400km away in Sydney) was offline (my issue) and everything above (adding more memory, adding 2x CDN’s (EWWW and Cloudflare), adding PHP Child workers etc) did not seem to help???

Enter UpCloud…

Recently, a friend sent a link to a blog article about a host called “UpCloud” who promised “Faster than SSD” performance.  This can’t be right: “Faster than SSD”? I was intrigued. I wanted to check it out as I thought nothing was faster than SSD (well, maybe RAM).

I signed up for a trial and ran a disk IO test (read the review here) and I was shocked. It’s fast. Very fast.

Summary: UpCloud was twice as fast (Disk IO and CPU) as Vultr (+ an optional $4/m firewall and $3/m for 1x backup).

This is a screenshot showing Vultr.com servers getting half the read and write disk io performance compared to upcloud.com.

fyi: Labels above are K Bytes per second. iozone loops through all file sizes from 4 KB to 16,348 KB and measures through the reads per second. To be honest, the meaning of the numbers doesn’t interest me, I just want to compare apples to apples.

This is am image showing iozone results breakdown chart (kbytes per sec on vertical axis, file size in horizontal axis and transfer size on third access)

(image snip from http://www.iozone.org/ which explains the numbers)

I might have to copy my website on UpCloud and see how fast it is.

Where to Deploy and Pricing

UpCloud Pricing: https://www.upcloud.com/pricing/

UpCloud Pricing

UpCloud does not have a data centre in Australia yet so why choose UpCloud?

Most of my site’s visitors are based in the US and UpCloud have disk IO twice as fast as Vultr (win-win?).  I could deploy to Chicago?

This image sows most of my visitors are in the US

My site’s traffic is growing and I need to ensure the site is fast enough in the future.

This image shows that most of my sites visitors are hitting my site on week days.

Advertisement:



Creating an UpCloud VM

I used a friend’s referral code and signed up to create my first VM.

FYI: use my Referral code and get $25 free credit.  Sign up only takes 2 minutes.

https://www.upcloud.com/register/?promo=D84793

When you click the link above you will receive 25$ to try out serves for 3 days. You can exit his trail and deposit $10 into UpCloud.

Trial Limitations

The trial mode restrictions are as following:

* Cloud servers can only be accessed using SSH, RDP, HTTP or HTTPS protocols
* Cloud servers are not allowed to send outgoing e-mails or to create outbound SSH/RDP connections
* The internet connection is restricted to 100 Mbps (compared to 500 Mbps for non-trial accounts)
* After your 72 hours free trial, your services will be deleted unless you make a one-time deposit of $10

UpCloud Links

The UpCloud support page is located here: https://www.upcloud.com/support/

More UpCloud links to read:

Signing up to UpCloud

Navigate to https://upcloud.com/signup and add your username, password and email address and click signup.

New UpCloud Signup Page

Add your address and payment details and click proceed (you don’t need to pay anything ($1 may be charged and instantly refunded to verify the card)

Add address and payment details

That’s it, check yout email.

Signup Done

Look for the UpCloud email and click https://my.upcloud.com/

Check Email

Now login

Login to UpCloud

Now I can see a dashboard 🙂

UpCloud Dashboard

I was happy to see 24/7 support is available.

This image shows the www.upcloud.com live chat

I opted in for the new dashboard

UpCloud new new dashboard

Deploy My First UpCloud Server

This is how I deployed a server.

Note: If you are going to deploy a server consider using my referral code and get $25 credit for free.

Under the “deploy a server” widget I named the server and chose a location (I think I was supposed to use an FQDN name -e.g., “fearby.com”). The deployment worked though. I clicked continue, then more options were made available:

  1. Enter a short server description.
  2. Choose a location (Frankfurt, Helsinki, Amsterdam, Singapore, London and Chicago)
  3. Choose the number of CPU’s and amount of memory
  4. Specify disk number/names and type (MaxIOPS or HDD).
  5. Choose an Operating System
  6. Select a Timezone
  7. Define SSH Keys for access
  8. Allowed login methods
  9. Choose hardware adapter types
  10. Where the send the login password

Deploy Server

FYI: How to generate a new SSH Key (on OSX or Ubuntu)

Output

Did the key export? (yes)

> /temp# ls /temp/ -al
> drwxr-xr-x 2 root root 4096 Jun 9 15:33 .
> drwxr-xr-x 27 root root 4096 Jun 8 14:25 ..
> -rw——- 1 user user 1766 Jun 9 15:33 example_rsa
> -rw-r–r– 1 user user 396 Jun 9 15:33 example_rsa.pub

“example_rsa” is the private key and “example_rsa.pub “is the public key.

  • The public key needs to be added to the server to allow access.
  • The private key needs to be added to any local ssh program used for remote access.

Initialisation script (after deployment)

I was pleased to see an initialization script section that calls actions after the server is deployed. I configured the initialisation script to pull down a few GB of backups from my Vultr website in Sydney (files now removed).

This was my Initialisation script:

Confirm and Deploy

I clicked “Confirm and deploy” but I had an alert that said trial mode can only deploy servers up to 1024MB of memory.

This image shows I cant deploy servers with 2/GB in trial modeExiting UpCloud Trial Mode

I opened the dashboard and clicked My Account then Billing, I could see the $25 referral credit but I guess I can’t use that in Trial.

I exited trial mode by depositing $10 (USD).

View Billing Details

Make a manual 1-time deposit of $10 to exit trial mode.

Deposit $10 to exit the trial

FYI: Server prices are listed below (or view prices here).

UpCloud Pricing

Now I can go back and deploy the server with the same settings above (1x CPU, 2GB Memory, Ubuntu 18.04, MaxIOPS Storage etc)

Advertisement:



Deployment takes a few minutes and depending on how you specified a password may be emailed to you.

UpCloud Server Deployed

The server is now deployed; now I can connect to it with my SSH program (vSSH).  Simply add the server’s IP, username, password and the SSH private key (generated above) to your ssh program of choice.

fyi: The public key contents start with “ssh-rsa”.

This image shows me connecting to my sever via ssh

I noticed that the initialisation script downloaded my 2+GB of files already. Nice.

UpCloud Billing Breakdown

I can now see on the UpCloud billing page in my dashboard that credit is deducted daily (68c); at this rate, I have 49 days credit left?

Billing Breakdown

I can manually deposit funds or set up automatic payments at any time 🙂

UpCloud Backup Options

You do not need to setup backups but in case you want to roll back (if things stuff up), it is a good idea. Backups are an additional charge.

I have set up automatic daily backups with an auto deletion after 2 days

To view backup scheduled click on your deployed server then click backup

List of UpCloud Backups

Note: Backups are charged at $0.056 for every GB stored – so $5.60 for every 100GB per month (half that for 50GB etc)

You can take manual backups at any time (and only be charged for the hour)

UpCloud Firewall Options

I set up a firewall at UpCloud to only allow the minimum number of ports (UpCloud DNS, HTTP, HTTPS and My IP to port 22).  The firewall feature is charged at $0.0056 an hour ($4.03 a month)

I love the ability to set firewall rules on incoming, destination and outgoing ports.

To view your firewall click on your deployed server then click firewall

UpCloud firewall

Update: I modified my firewall to allow inbound ICMP (IPv4/IPv6) and UDP (IPv4/IPv6) packets.

(Note: Old firewall screenshot)

Firewall Rules Allow port 80, 443 and DNS

Because my internet provider has a dynamic IP, I set up a VPN with a static IP and whitelisted it for backdoor access.

Local Ubuntu ufw Firewall

I duplicated the rules in my local ufw (2nd level) firewall (and blocked mail)

UpCloud Download Speeds

I pulled down a 1.8GB Ubuntu 18.08 Desktop ISO 3 times from gigenet.com and the file downloaded in 32 seconds (57MB/sec). Nice.

Install Common Ubuntu Packages

I installed common Ubuntu packages.

Timezone

I checked the server’s time (I thought this was auto set before I deployed)?

I reset the time to Australia/Sydney.

Now the timezone is set 🙂

Shell History

I increased the shell history.

SSH Login

I created a ~/.ssh/authorized_keys file and added my SSH public key to allow password-less logins.

I added my pubic ssh key, then exited the ssh session and logged back in. I can now log in without a password.

Install NGINX

nginx/1.14.0 is now installed.

A quick GT Metrix test.

This image shows awesome static nginx performance ratings of of 99%

Install MySQL

Run these commands to install and secure MySQL.

Securing the MySQL server deployment.
> Would you like to setup VALIDATE PASSWORD plugin?: n
> New password: **********************************************
> Re-enter new password: **********************************************
> Remove anonymous users? (Press y|Y for Yes, any other key for No) : y
> Disallow root login remotely? (Press y|Y for Yes, any other key for No) : y
> Remove test database and access to it? (Press y|Y for Yes, any other key for No) : y
> Reload privilege tables now? (Press y|Y for Yes, any other key for No) : y
> Success.

I disabled the validate password plugin because I hate it.

MySQL Ver 14.14 Distrib 5.7.22 is now installed.

Set MySQL root login password type

Set MySQL root user to authenticate via “mysql_native_password”. Run the “mysql” command.

Now let’s set the root password authentication method to “mysql_native_password”

Check authentication method.

Now we need to flush permissions.

Done.

Advertisement:



Install PHP

Install PHP 7.2

PHP 7.2.5, Zend Engine v3.2.0 with Zend OPcache v7.2.5-1 is now installed. Do update PHP frequently.

I made the following changes in /etc/php/7.2/fpm/php.ini

> cgi.fix_pathinfo=0
> max_input_vars = 1000
> memory_limit = 1024M
> max_file_uploads = 20M
> post_max_size = 20M

Install PHP Modules

sudo apt-get install php-pear php7.2-curl php7.2-dev php7.2-mbstring php7.2-zip php7.2-mysql php7.2-xml

Install PHP FPM

Configure PHP FPM config.

Edit /etc/php/7.2/fpm/php.ini

> cgi.fix_pathinfo=0
> max_input_vars = 1000
> memory_limit = 1024M
> max_file_uploads = 20M
> post_max_size = 20M

Reload php sudo service.

Install PHP Modules

Configuring NGINX

If you are not comfortable editing NGINX config files read here, here and here.

I made a new “www root” folder, set permissions and created a default html file.

I edited the “root” key in “/etc/nginx/sites-enabled/default” file and set the root a new location (e.g., “/www-root”)

I added these performance tweaks to /etc/nginx/nginx.conf

> worker_cpu_affinity auto;
> worker_rlimit_nofile 100000

I add the following lines to “http {” section in /etc/nginx/nginx.conf

Check NGINX Status

Install Open SSL that supports TLS 1.3

This is a work in progress. The steps work just fine for me on Ubuntu 16.04. but not Ubuntu 18.04.?

Installing Adminer MySQL GUI

I will use the PHP based Adminer MySQL GUI to export and import my blog from one server to another. All I needed to do is install it on both servers (simple 1 file download)

Use Adminer to Export My Blog (on Vultr)

On the original server open Adminer (http) and..

  1. Login with the MySQL root account
  2. Open your database
  3. Choose “Save” as the output
  4. Click on Export

This image shows the export of the wordpress adminer page

Save the “.sql” file.

I used Adminer on the UpCloud server to Import My Blog

FYI: Depending on the size of your database backup you may need to temporarily increase your upload and post sizes limits in PHP and NGINX before you can import your database.

Edit /etc/php/7.2/fpm/php.ini
> max_file_uploads = 100M
> post_max_size =100M

And Edit: /etc/nginx/nginx.conf
> client_max_body_size 100M;

Don’t forget to reload NGINX config and restart NGINX and PHP. Take note of the maximum allowed file size in the screenshot below. I temporarily increased my upload limits to 100MB in order to restore my 87MB blog.

Now I could open Adminer on my UpCloud server.

  1. Create a new database
  2. Click on the database and click Import
  3. Choose the SQL file
  4. Click Execute to import it

Import MuSQL backup with Adminer

Don’t forget to create a user and assign permissions (as required – check your wp-config.php file).

Import MySQL Database

Tip: Don’t forget to lower the maximum upload file size and max post size after you import your database,

Cloudflare DNS

I use Cloudflare to manage DNS, so I need to tell it about my new server.

You can get your server’s IP details from the UpCloud dashboard.

Find IP

At Cloudflare update your DNS details to point to the server’s new IPv4 (“A Record”) and IPv6 (“AAAA Record”).

Cloudflare DNS

Advertisement:



Domain Error

I waited an hour and my website was suddenly unavailable.  At first, I thought this was Cloudflare forcing the redirection of my domain to HTTP (that was not yet set up).

DNS Not Replicated Yet

I chatted with UpCloud chat on their webpage and they kindly assisted me to diagnose all the common issues like DNS values, DNS replication, Cloudflare settings and the error was pinpointed to my NGINX installation.  All NGINX config settings were ok from what we could see?  I uninstalled NGINX and reinstalled it (and that fixed it). Thanks UpCloud Support 🙂

Reinstalled NGINX

I reinstalled NGINX and reconfigured /etc/nginx/nginx.conf (I downloaded my SSL cert from my old server just in case).

Here is my /etc/nginx/nginx.conf file.

Here is my /etc/nginx/sites-available/default file (fyi, I have not fully re-setup TLS 1.3 yet so I commented out the settings)

SSL Labs SSL Certificate Check

All good thanks to the config above.

SSL Labs

Install WP-CLI

I don’t like setting up FTP to auto-update WordPress plugins. I use the WP-CLI tool to manage WordPress installations by the command line. Read my blog here on using WP-CLI.

Download WP-CLI

Move WP-CLI to the bin folder as “wp”

Test wp

Update WordPress Plugins

Now I can run “wp plugin update” to update all WordPress plugins

Update WordPress Core

WordPress core file can be updated with “wp core update“

Troubleshooting: Use the flag “–allow-root “if wp needs higher access (unsafe action though).

Install PHP Child Workers

I edited the following file to setup PHP child workers /etc/php/7.2/fpm/pool.d/www.conf

Changes

> pm = dynamic
> pm.max_children = 40
> pm.start_servers = 15
> pm.min_spare_servers = 5
> pm.max_spare_servers = 15
> pm.process_idle_timeout = 30s;
> pm.max_requests = 500;
> php_admin_value[error_log] = /var/log/www-fpm-php.www.log
> php_admin_value[memory_limit] = 512M

Advertisement:



Restart PHP

Test NGINX config, reload NGINX config and restart NGINX

Output (14 workers are ready)

Memory Tweak (set at your own risk)

vm.swappiness = 1

Setting swappiness to a value of 1 all but disables the swap file and tells the Operating System to aggressively use ram, a value of 10 is safer. Only set this if you have enough memory available (and free).

Possible swappiness settings:

> vm.swappiness = 0 Swap is disabled. In earlier versions, this meant that the kernel would swap only to avoid an out of memory condition when free memory will be below vm.min_free_kbytes limit, but in later versions, this is achieved by setting to 1.[2]> vm.swappiness = 1 Kernel version 3.5 and over, as well as Red Hat kernel version 2.6.32-303 and over: Minimum amount of swapping without disabling it entirely.
> vm.swappiness = 10 This value is sometimes recommended to improve performance when sufficient memory exists in a system.[3]
> vm.swappiness = 60 The default value.
> vm.swappiness = 100 The kernel will swap aggressively.

The “htop” tool is a handy memory monitoring tool to “top”

Also you can use good old “watch” command to show near-live memory usage (auto-refreshes every 2 seconds)

Script to auto-clear the memory/cache

As a habit, I am setting up a cronjob to check when free memory falls below 100MB, then cache is automatically cleared (freeing memory).

Script Contents: clearcache.sh

I set the cronjob to run every 15 mins, I added this to my cronjob.

Sample log output

I will check the log (/scripts/clearcache.log) in a few days and view the memory trends.

After 1/2 a day Ubuntu 18.04 is handling memory just fine, no externally triggered cache clears have happened 🙂

Free memory over time

I used https://crontab.guru/every-hour to set the right schedule in crontab.

I rebooted the VM.

Update: I now use Nixstats monitoring

Swap File

FYI: Here is a handy guide on viewing swap file usage here. I’m not using swap files so it is only an aside.

After the system rebooted I checked if the swappiness setting was active.

Yes, swappiness is set.

File System Tweaks – Write Back Cache (set at your own risk)

First, check your disk name and file system

Take note of your disk name (e.g vda1)

I used TuneFS to enable writing data to the disk before writing to the journal. tunefs is a great tool for setting file system parameters.

Warning (snip from here): “I set the mode to journal_data_writeback. This basically means that data may be written to the disk before the journal. The data consistency guarantees are the same as the ext3 file system. The downside is that if your system crashes before the journal gets written then you may loose new data — the old data may magically reappear.“

Warning this can corrupt your data. More information here.

I ran this command.

I edited my fstab to append the “writeback,noatime,nodiratime” flags for my volume after a reboot.

Edit FS Tab:

I added “writeback,noatime,nodiratime” flags to my disk options.

Updating Ubuntu Packages

Show updatable packages.

Update Packages.

Unattended Security Updates

Read more on Ubuntu 18.04 Unattended upgrades here, here and here.

Install Unattended Upgrades

Enable Unattended Upgrades.

Now I configure what packages not to auto update.

Edit /etc/apt/apt.conf.d/50unattended-upgrades

Find “Unattended-Upgrade::Package-Blacklist” and add packages that you don’t want automatically updated, you may want to manually update these (and monitor updates).

I prefer not to auto-update critical system apps (I will do this myself).

FYI: You can find installed packages by running this command:

Enable automatic updates by editing /etc/apt/apt.conf.d/20auto-upgrades

Edit the number at the end (the number is how many days to wait before updating) of each line.

> APT::Periodic::Update-Package-Lists “1”;
> APT::Periodic::Download-Upgradeable-Packages “1”;
> APT::Periodic::AutocleanInterval “7”;
> APT::Periodic::Unattended-Upgrade “1”;

Set to “0” to disable automatic updates.

The results of unattended-upgrades will be logged to /var/log/unattended-upgrades

Update packages now.

Almost done.

I Rebooted

Advertisement:



GT Metrix Score

I almost fell off my chair. It’s an amazing feeling hitting refresh in GT Metrix and getting sub-2-second score consistently (and that is with 17 assets loading and 361KB of HTML content)

0.9sec load times

WebPageTest.org Test Score

Nice. I am not sure why the effective use of CDN has an X rating as I have the EWWW CDN and Cloudflare. First Byte time is now a respectable “B”, This was always bad.

Update: I found out the longer you set cache delays in Cloudflare the higher the score.

Web Page Test

GT Metrix has a nice historical breakdown of load times (night and day).

Upcloud Site Speed in GTMetrix

Google Page Speed Insight Desktop Score

I benchmarked with https://developers.google.com/speed/pagespeed/insights/

This will help with future SEO rankings. It is well known that Google is pushing fast servers.

100% Desktop page speed score

Google Chrome 70 Dev Console Audit (Desktop)

100% Chrome Audit Score

This is amazing, I never expected to get this high score.  I know Google like (and are pushing) sub-1-second scores.

My site is loading so well it is time I restored some old features that were too slow on other servers

  • I disabled Lazy loading of images (this was not working on some Android devices)
  • I re-added the News Widget and news images.

GTMetrix and WebpageTest sores are still good (even after adding bloat)

Benchmarks are still good

My WordPress site is not really that small either

Large website

FYI: WordPress Plugins I use.

These are the plugins I use.

How I use these plugins to speed up my site.

  • I use EWWW Image Optimizer plugin to auto-compress my images and to provide a CDN for media asset deliver (pre-Cloudflare). Learn more about ExactDN and EWWW.io here.
  • I use Autoptimize plugin to optimise HTML/CSS/JS and ensure select assets are on my EWWW CDN. This plugin also removes WordPress Emojis, removed the use of Google Fonts, allows you to define pre-configured domains, Async Javascript-files etc.
  • I use BJ Lazy Load to prevent all images in a post from loading on load (and only as the user scrolls down the page).
  • GTmetrix for WordPress and Cloudflare plugins are for information only?
  • I use WP-Optimize to ensure my database is healthy and to disable comments/trackbacks and pingbacks.

Let’s Test UpCloud’s Disk IO in Chicago

Looks good to me, Read IO is a little bit lower than UpCloud’s Singapore data centre but still, it’s faster than Vultr.  I can’t wait for more data centres to become available around the world.

Why is UpCloud Disk IO so good?

I asked UpCloud on Twitter why the Disk IO was so good.

  • “MaxIOPS is UpCloud’s proprietary block-storage technology. MaxIOPS is physically redundant storage technology where all customer’s data is located in two separate physical devices at all times. UpCloud uses InfiniBand (!) network to connect storage backends to compute nodes, where customers’ cloud servers are running. All disks are enterprise-grade SSD’s. And using separate storage backends, it allows us to live migrate our customers’ cloud servers freely inside our infrastructure between compute nodes – whether it be due to hardware malfunction (compute node) or backend software updates (example CPU vulnerability and immediate patching).“

My Answers to Questions to support

Q1) What’s the difference between backups and snapshots (a Twitter user said Snapshots were a thing)

A1) Backups and snapshots are the same things with our infrastructure.

Q2) What are charges for backup of a 50GB drive?

A2) We charge $0.06 / GB of the disk being captured. But capture the whole disk, not just what was used. So for a 50GB drive, we charge $0.06 * 50 = $3/month. Even if 1GB were only used.

  • Support confirmed that each backup is charged (so 5 times manual backups are charged 5 times). Setting up a daily auto backup schedule for 2 weeks would create 14 billable backup charges.
  • I guess a 25GB server will be $1.50 a month

Q3) What are data charges if I go over my 2TB quota?

A3) Outgoing data charges are $0.056/GB after the pre-configured allowance.

Q4) What happens if my balance hits $0?

A4) You will get notification of low account balance 2 weeks in advance based on your current daily spend. When your balance reaches zero, your servers will be shut down. But they will still be charged for. You can automatically top-up if you want to assign a payment type from your Control Panel. You deposit into your balance when you want. We use a prepay model of payment, so you need to top up before using, not billing you after usage. We give you lots of chances to top-up.

Support Tips

  • One thing to note, when deleting servers (CPU, RAM) instances, you get the option to delete the storages separately via a pop-up window. Choose to delete permanently to delete the disk, to save credit. Any disk storage lying around even unattached to servers will be billed.
  • Charges are in USD.

I think it’s time to delete my domain from Vultr in Sydney.

Deleted my Vultr domain

I deleted my Vultr domain.

Delete Vultr Server

Done.

More Reading on UpCloud

https://www.upcloud.com/documentation/faq/

UpCloud Server Status

http://status.upcloud.com

Check out my new guide on Nixstats for awesome monitoring

What I would like

  1. Ability to name individual manual backups (tag with why I backed up).
  2. Ability to push user defined data from my VM to the dashboard
  3. Cheaper scheduled backups
  4. Sydney data centres (one day)

Advertisement:



Update: Post UpCloud Launch Tweaks (Awesome)

I had a look at https://www.webpagetest.org/ results to see where else I can optimise webpage delivery.

Optimisation Options

Disable dasjhicons.min.css (for unauthenticated WordPress users).

Find functions.php in the www root

Edit functions.php

Add the following

// Remove dashicons in frontend for unauthenticated users
add_action( 'wp_enqueue_scripts', 'bs_dequeue_dashicons' );
function bs_dequeue_dashicons() {
    if ( ! is_user_logged_in() ) {
        wp_deregister_style( 'dashicons' );
    }
}

HTTP2 Push

I added http2 to my listening servers

I tested a http2 push page by defining this in /etc/nginx/sites-available/default 

Once I tested that push (demo here) was working I then defined two files to push that were being sent from my server

I used the WordPress Plugin Autoptimize to remove Google font usage (this removed a number of files being loaded when my page loads).

I used the WordPress Plugin WP-Optimize plugin into to remove comments and disable pingbacks and trackbacks.

WordPress wp-config.php tweaks

Add 2FA Authentication to server logins.

I recently checked out YubiCo YubiKeys and I have secured my Linux servers with 2FA prompts at login. Read the guide here. I secured my WordPress too.

Tweaks Todo

  • Compress placeholder BJ Lazy Load Image (plugin is broken)
  • Solve 2x Google Analytics tracker redirects (done, switched to Matomo)

Conclusion

I love UpCloud’s fast servers, give them a go (use my link and get $25 free credit).

I love Cloudflare for providing a fast CDN.

I love ewww.io’s automatic Image Compression and Resizing plugin that automatically handles image optimisations and pre Cloudflare/first hit CDN caching.

Read my post about server monitoring with Nixstats here.

Let the results speak for themselves (sub <1 second load times).

Results

I hope this guide helps someone.

Please consider using my referral code and get $25 credit for free.

https://www.upcloud.com/register/?promo=D84793

 

2020 Update. I have stopped using Putty and WinSCP. I now use MobaXterm (a tabbed SSH client for Windows) as it is way faster than WinSCP and better than Putty. Read my review post of MobaXTerm here.

Ask a question or recommend an article

[contact-form-7 404 "Not Found"]

Revision History

v2.1 Newer GTMetrix scores

v2.0 New UpCloud UI Update and links to new guides.

v1.9 Spelling and grammar

v1.8 Trial mode gotcha (deposit money ASAP)

v1.7 Added RSA Private key info

v1.7 – Added new firewall rules info.

v1.6 – Added more bloat to the site, still good.

v1.5 Improving Accessibility

v1.4 Added Firewall Price

v1.3 Added wp-config and plugin usage descriptions.

v1.2 Added GTMetrix historical chart.

v1.1 Fixed free typos and added final conclusion images.

v1.0 Added final results

v0.9 added more tweaks (http2 push, removing unwanted files etc)

v0.81 Draft  – Added memory usage chart and added MaxIOPS info from UpCloud.

v0.8 Draft post.

I set the cronjob to run every 15 mins, I added this to my cronjob.

 

Sample log output

 

I will check the log (/scripts/clearcache.log) in a few days and view the memory trends.

After 1/2 a day Ubuntu 18.04 is handling memory just fine, no externally triggered cache clears have happened 🙂

Free memory over time

I used https://crontab.guru/every-hour to set the right schedule in crontab.

I rebooted the VM.

Update: I now use Nixstats monitoring

Swap File

FYI: Here is a handy guide on viewing swap file usage here. I’m not using swap files so it is only an aside.

After the system rebooted I checked if the swappiness setting was active.

 

Yes, swappiness is set.

File System Tweaks – Write Back Cache (set at your own risk)

First, check your disk name and file system

 

Take note of your disk name (e.g vda1)

I used TuneFS to enable writing data to the disk before writing to the journal. tunefs is a great tool for setting file system parameters.

Warning (snip from here): “I set the mode to journal_data_writeback. This basically means that data may be written to the disk before the journal. The data consistency guarantees are the same as the ext3 file system. The downside is that if your system crashes before the journal gets written then you may loose new data — the old data may magically reappear.“

Warning this can corrupt your data. More information here.

I ran this command.

 

I edited my fstab to append the “writeback,noatime,nodiratime” flags for my volume after a reboot.

Edit FS Tab:

 

I added “writeback,noatime,nodiratime” flags to my disk options.

 

Updating Ubuntu Packages

Show updatable packages.

 

Update Packages.

 

Unattended Security Updates

Read more on Ubuntu 18.04 Unattended upgrades here, here and here.

Install Unattended Upgrades

 

Enable Unattended Upgrades.

 

Now I configure what packages not to auto update.

Edit /etc/apt/apt.conf.d/50unattended-upgrades

Find “Unattended-Upgrade::Package-Blacklist” and add packages that you don’t want automatically updated, you may want to manually update these (and monitor updates).

I prefer not to auto-update critical system apps (I will do this myself).

 

FYI: You can find installed packages by running this command:

 

Enable automatic updates by editing /etc/apt/apt.conf.d/20auto-upgrades

Edit the number at the end (the number is how many days to wait before updating) of each line.

> APT::Periodic::Update-Package-Lists “1”;
> APT::Periodic::Download-Upgradeable-Packages “1”;
> APT::Periodic::AutocleanInterval “7”;
> APT::Periodic::Unattended-Upgrade “1”;

Set to “0” to disable automatic updates.

The results of unattended-upgrades will be logged to /var/log/unattended-upgrades

Update packages now.

 

Almost done.

I Rebooted

Advertisement:



GT Metrix Score

I almost fell off my chair. It’s an amazing feeling hitting refresh in GT Metrix and getting sub-2-second score consistently (and that is with 17 assets loading and 361KB of HTML content)

0.9sec load times

WebPageTest.org Test Score

Nice. I am not sure why the effective use of CDN has an X rating as I have the EWWW CDN and Cloudflare. First Byte time is now a respectable “B”, This was always bad.

Update: I found out the longer you set cache delays in Cloudflare the higher the score.

Web Page Test

GT Metrix has a nice historical breakdown of load times (night and day).

Upcloud Site Speed in GTMetrix

Google Page Speed Insight Desktop Score

I benchmarked with https://developers.google.com/speed/pagespeed/insights/

This will help with future SEO rankings. It is well known that Google is pushing fast servers.

100% Desktop page speed score

Google Chrome 70 Dev Console Audit (Desktop)

100% Chrome Audit Score

This is amazing, I never expected to get this high score.  I know Google like (and are pushing) sub-1-second scores.

My site is loading so well it is time I restored some old features that were too slow on other servers

  • I disabled Lazy loading of images (this was not working on some Android devices)
  • I re-added the News Widget and news images.

GTMetrix and WebpageTest sores are still good (even after adding bloat)

Benchmarks are still good

My WordPress site is not really that small either

Large website

FYI: WordPress Plugins I use.

These are the plugins I use.

How I use these plugins to speed up my site.

  • I use EWWW Image Optimizer plugin to auto-compress my images and to provide a CDN for media asset deliver (pre-Cloudflare). Learn more about ExactDN and EWWW.io here.
  • I use Autoptimize plugin to optimise HTML/CSS/JS and ensure select assets are on my EWWW CDN. This plugin also removes WordPress Emojis, removed the use of Google Fonts, allows you to define pre-configured domains, Async Javascript-files etc.
  • I use BJ Lazy Load to prevent all images in a post from loading on load (and only as the user scrolls down the page).
  • GTmetrix for WordPress and Cloudflare plugins are for information only?
  • I use WP-Optimize to ensure my database is healthy and to disable comments/trackbacks and pingbacks.

Let’s Test UpCloud’s Disk IO in Chicago

Looks good to me, Read IO is a little bit lower than UpCloud’s Singapore data centre but still, it’s faster than Vultr.  I can’t wait for more data centres to become available around the world.

Why is UpCloud Disk IO so good?

I asked UpCloud on Twitter why the Disk IO was so good.

  • “MaxIOPS is UpCloud’s proprietary block-storage technology. MaxIOPS is physically redundant storage technology where all customer’s data is located in two separate physical devices at all times. UpCloud uses InfiniBand (!) network to connect storage backends to compute nodes, where customers’ cloud servers are running. All disks are enterprise-grade SSD’s. And using separate storage backends, it allows us to live migrate our customers’ cloud servers freely inside our infrastructure between compute nodes – whether it be due to hardware malfunction (compute node) or backend software updates (example CPU vulnerability and immediate patching).“

My Answers to Questions to support

Q1) What’s the difference between backups and snapshots (a Twitter user said Snapshots were a thing)

A1) Backups and snapshots are the same things with our infrastructure.

Q2) What are charges for backup of a 50GB drive?

A2) We charge $0.06 / GB of the disk being captured. But capture the whole disk, not just what was used. So for a 50GB drive, we charge $0.06 * 50 = $3/month. Even if 1GB were only used.

  • Support confirmed that each backup is charged (so 5 times manual backups are charged 5 times). Setting up a daily auto backup schedule for 2 weeks would create 14 billable backup charges.
  • I guess a 25GB server will be $1.50 a month

Q3) What are data charges if I go over my 2TB quota?

A3) Outgoing data charges are $0.056/GB after the pre-configured allowance.

Q4) What happens if my balance hits $0?

A4) You will get notification of low account balance 2 weeks in advance based on your current daily spend. When your balance reaches zero, your servers will be shut down. But they will still be charged for. You can automatically top-up if you want to assign a payment type from your Control Panel. You deposit into your balance when you want. We use a prepay model of payment, so you need to top up before using, not billing you after usage. We give you lots of chances to top-up.

Support Tips

  • One thing to note, when deleting servers (CPU, RAM) instances, you get the option to delete the storages separately via a pop-up window. Choose to delete permanently to delete the disk, to save credit. Any disk storage lying around even unattached to servers will be billed.
  • Charges are in USD.

I think it’s time to delete my domain from Vultr in Sydney.

Deleted my Vultr domain

I deleted my Vultr domain.

Delete Vultr Server

Done.

 

Check out my new guide on Nixstats for awesome monitoring

What I would like

  1. Ability to name individual manual backups (tag with why I backed up).
  2. Ability to push user defined data from my VM to the dashboard
  3. Cheaper scheduled backups
  4. Sydney data centres (one day)

 

Update: Post UpCloud Launch Tweaks (Awesome)

I had a look at https://www.webpagetest.org/ results to see where else I can optimise webpage delivery.

Optimisation Options

HTTP2 Push

I added http2 to my listening servers I tested a http2 push page by defining this in /etc/nginx/sites-available/default 

Once I tested that push (demo here) was working I then defined two files to push that were being sent from my server

2FA Authentication at login

I recently checked out YubiCo YubiKeys and I have secured my Linux servers with 2FA prompts at login. Read the guide here. I secured my WordPress aswel.

Performance

I used the WordPress Plugin Autoptimize to remove Google font usage (this removed a number of files being loaded when my page loads).

I used the WordPress Plugin WP-Optimize plugin into to remove comments and disable pingbacks and trackbacks.

Results

Conclusion

I love UpCloud’s fast servers, give them a go (use my link and get $25 free credit).

I love Cloudflare for providing a fast CDN.

I love ewww.io’s automatic Image Compression and Resizing plugin that automatically handles image optimisations and pre Cloudflare/first hit CDN caching.

Read my post about server monitoring with Nixstats here.

Let the results speak for themselves (sub <1 second load times).

More Reading on UpCloud

https://www.upcloud.com/documentation/faq/

UpCloud Server Status

http://status.upcloud.com

I hope this guide helps someone.

Free Credit

Please consider using my referral code and get $25 credit for free.

https://www.upcloud.com/register/?promo=D84793

2020 Update. I have stopped using Putty and WinSCP. I now use MobaXterm (a tabbed SSH client for Windows) as it is way faster than WinSCP and better than Putty. Read my review post of MobaXTerm here.

Ask a question or recommend an article

[contact-form-7 404 "Not Found"]

Revision History

v2.2 Converting to Blocks

v2.1 Newer GTMetrix scores

v2.0 New UpCloud UI Update and links to new guides.

v1.9 Spelling and grammar

v1.8 Trial mode gotcha (deposit money ASAP)

v1.7 Added RSA Private key info

v1.7 – Added new firewall rules info.

v1.6 – Added more bloat to the site, still good.

v1.5 Improving Accessibility

v1.4 Added Firewall Price

v1.3 Added wp-config and plugin usage descriptions.

v1.2 Added GTMetrix historical chart.

v1.1 Fixed free typos and added final conclusion images.

v1.0 Added final results

v0.9 added more tweaks (http2 push, removing unwanted files etc)

v0.81 Draft  – Added memory usage chart and added MaxIOPS info from UpCloud.

v0.8 Draft post.

Setting up DNSSEC on a Namecheap domain hosted on UpCloud using CloudFlare

by

This is how I setup DNSSEC on a Namecheap domain hosted on UpCloud using CloudFlare to setup DNS security extensions

Advertisement:



If you have not read my previous posts I have now moved my blog to the awesome UpCloud host (signup using this link to get $25 free UpCloud VM credit). I compared Digital Ocean, Vultr and UpCloud Disk IO here and UpCloud came out on top by a long way (read the blog post here). Here is my blog post on moving from Vultr to UpCloud.

About DNSSEC

Wikipedia has a great write-up on DNSSEC also read the ICANN page on DNSSEC.

Snip “DNSSEC is a set of extensions to DNS which provide to DNS clients (resolvers) origin authentication of DNS data, authenticated denial of existence, and data integrity, but not availability or confidentiality.”

https://daniellbenway.net has a great video explaining DNSSEC

Handy DNSSEC Links

Let’s view my DNSSEC status now

https://dnssec-analyzer.verisignlabs.com/ can help you check your sites DNSSEC status.

DNSSEC Analyzer - https://dnssec-analyzer.verisignlabs.com/

Prerequisites

This guide assumes you have already purchased a domain and set it up with say UpCloud hosting (read my setup guide here).

Buy a domain name from Namecheap here.

Domain names for just 88 cents!

Read my old guide here that I created while setting up Cloudflare on the Vultr host to see how to setup Cloudflare.

Setting up DNSSEC

First I logged into My Namecheap account, selected my domain, selected Advanced DNS and enabled DNSSEC.

Screenshot of Namecheap Advanced DNS page

I can see a number of values for DNSSEC KeyType/Algorithm/Digest Type and Digest. Below are the options in the dropdowns for Algorithm and Digest Type.

DNSSEC Algorithms

DNSSEC Algorithms (RSA, MD5 etc)

DNSSEC Digest Types

DNSSEC Digest Types (SHA etc)

I contacted NameCheap support and they said I needed to contact my UpCloud hosts to get relevant DNSSEC values.

My domain was purchased at NameCheap but by domain routers by Cloudflare DNS.

Namecheap DNS Nameservers pointing to cloudflare

Advertisement:



By chance, I logged into my Cloudflare account and noticed they have a DNSSEC section under DNS. Nice.

Screenshot of Cloudflare menu, DNS highlighted.

I enabled DNSSEC

Enable Cloudflare DNSSEC records

Cloudflare offers all the relevant DNSSEC values.

Screenshot of Cloudflare DNSSEC generated Values

I entered these values into Namecheap under Advanced DNS on my domain.

Screenshot fo adding a DNS record at Namecheap

After 5 mins re-ran the DNSSEC Analyzr tool.

Screenshot of http://dnssec-debugger.verisignlabs.com/ Results

Hmmm, Cloudflare seem to think something is wrong 🙁

Screenshot of Cloudflare saying DNSSEC is not configured

I ran a DNS DS Lookup on my site. Everything appears ok.

Screenshot of https://mxtoolbox.com/SuperTool.aspx?action=mx

I re-added the record in Namecheap and waited for 15 mins and this time Cloudflare was happy. Maybe I just needed to wait for DNS replication a little longer?

Screenshot of cloudflare showing DNSSEC is all ok.

I tested my DNS serves with DNS Root Canary

DNS test with https://rootcanary.org/

I tested my site’s DNSSEC with https://zonemaster.iis.se/

Screenshot of https://zonemaster.iis.se/

Done

Skipping Cloudflare

I found that I can simply skip cloudflare by enabling Premium DNS at Namecheap

Then enabling DNSSEC

Easy (totally independant of Cloudflare)

I hope this guide helps someone.

Advertisement:



Please consider using my referral code and get $25 UpCloud VM credit for free.

https://www.upcloud.com/register/?promo=D84793

Ask a question or recommend an article

[contact-form-7 404 "Not Found"]

Revision History

V1.3 Namecheap DNSSEC

V1.2 ICANN DNSSEC link

V1.1 https://daniellbenway.net explainer video.

v1.0 Initial Post

Measuring VM performance (CPU, Disk, Latency, Concurrent Users etc) on Ubuntu and comparing Vultr, Digital Ocean and UpCloud – Part 3 of 4

by

How can you measure VM performance (CPU, Disk, Latency, Concurrent Users etc) on Ubuntu and comparing Vultr, Digital Ocean and UpCloud – Part 3 of 4

Advertisement:



Read Part 1, Part 2, Part 3 or Part 4

I used these commands to generate bonnie++ reports from the data in part 2

echo "<h1>Bonnie Results</h1>" > /www-data/bonnie.html
echo "<h2>Vultr (Sydney)</h2>" >> /www-data/bonnie.html
echo "1.97,1.97,servername,1,1528177870,4G,,656,99,308954,68,113706,33,1200,92,188671,30,10237,251,16,,,,,+++++,+++,+++++,+++,+++++,+++,+++++,+++,+++++,+++,+++++,+++,26067us,119ms,179ms,29139us,26069us,16118us,1463us,703us,880us,263us,119us,593us" | bon_csv2html >> /www-data/bonnie.html
echo "<h2>Digital Ocean (London)</h2>" >> /www-data/bonnie.html
echo "1.97,1.97,servername,1,1528186398,4G,,699,99,778636,74,610414,60,1556,99,1405337,59,+++++,+++,16,,,,,+++++,+++,+++++,+++,+++++,+++,+++++,+++,+++++,+++,+++++,+++,17678us,10099us,17014us,7027us,3067us,2366us,1243us,376us,611us,108us,59us,181us" | bon_csv2html >> /www-data/bonnie.html
echo "<h2>UpCloud (Singapore)</h2>" >> /www-data/bonnie.html
echo "1.97,1.97,servername,1,1528226703,4G,,1014,99,407179,24,366622,32,2137,99,451886,17,+++++,+++,16,,,,,+++++,+++,+++++,+++,+++++,+++,+++++,+++,+++++,+++,+++++,+++,11297us,54232us,16443us,4949us,44883us,1595us,264us,340us,561us,138us,66us,327us" | bon_csv2html >> /www-data/bonnie.html

Image of results here

Bonnie Results

Network Performace

IMHO Network Latency is the biggest impact on server performance, Read my old post on scalability on a budget here. I am in Australia an having a server in Singapore was too far away and latency was terrible.

Here is a non-scientific example of pinging a Vultr, Digital Ocean and UpCloud server in three different locations (and Google).

Ping Test

Test Ping Results

Latency matters, run a https://www.webpagetest.org/ scan over your site to see why.

Adding https added almost 0.7 seconds to https communications in the past on Digital Ocean (a few thousand kilometres away). The longer the latency the longer HTTPS handshakes take.

SSL

Deploying a server to Singapore (in my experience) is bad if your visitors are in Australia. But deploying to other regions may be lower in cost though. It’s a trade off.

Server Location

Deploy servers as close as you can to your customers is the best tip for performance.

Deploy serves close to your customers

Also, consider setting up Image Optimization and Image CDN plugins (guide here) in WordPress and using Cloudflare (guide here)

Benchmarking with SysBench

Install CPU Benchmark

sudo apt-get install sysbench

CPU Benchmark (Vultr/Sydney)

Result

sysbench 0.4.12:  multi-threaded system evaluation benchmark

Running the test with following options:
Number of threads: 1

Doing CPU performance benchmark

Threads started!
Done.

Maximum prime number checked in CPU test: 20000


Test execution summary:
    total time:                          39.1700s
    total number of events:              10000
    total time taken by event execution: 39.1586
    per-request statistics:
         min:                                  2.90ms
         avg:                                  3.92ms
         max:                                 20.44ms
         approx.  95 percentile:               7.43ms

Threads fairness:
    events (avg/stddev):           10000.0000/0.00
    execution time (avg/stddev):   39.1586/0.00

39.15 seconds

CPU Benchmark (Digital Ocean/London)

sysbench --test=cpu --cpu-max-prime=20000 run

Result

sysbench 0.4.12:  multi-threaded system evaluation benchmark

Running the test with following options:
Number of threads: 1

Doing CPU performance benchmark

Threads started!
Done.

Maximum prime number checked in CPU test: 20000


Test execution summary:
    total time:                          33.4382s
    total number of events:              10000
    total time taken by event execution: 33.4352
    per-request statistics:
         min:                                  3.24ms
         avg:                                  3.34ms
         max:                                  6.45ms
         approx.  95 percentile:               3.45ms

Threads fairness:
    events (avg/stddev):           10000.0000/0.00
    execution time (avg/stddev):   33.4352/0.00

33.43 sec

CPU Benchmark (UpCloud/Singapore)

sysbench --test=cpu --cpu-max-prime=20000 run

Result

sysbench 0.4.12:  multi-threaded system evaluation benchmark

Running the test with following options:
Number of threads: 1



Doing CPU performance benchmark

Threads started!
Done.

Maximum prime number checked in CPU test: 20000


Test execution summary:
    total time:                          23.7809s
    total number of events:              10000
    total time taken by event execution: 23.7780
    per-request statistics:
         min:                                  2.35ms
         avg:                                  2.38ms
         max:                                  6.92ms
         approx.  95 percentile:               2.46ms

Threads fairness:
    events (avg/stddev):           10000.0000/0.00
    execution time (avg/stddev):   23.7780/0.00

23.77 sec

Surprisingly, 1st place in prime generation goes to UpCloud, then Digital Ocean then Vultr.  UpCloud has some good processors.

Processors:

  • UpCLoud (Singapore): Intel(R) Xeon(R) CPU E5-2687W v4 @ 3.00GHz
  • Digital Ocean (London): Intel(R) Xeon(R) CPU E5-2630L v2 @ 2.40GHz
  • Vultr (Sydney): Virtual CPU a7769a6388d5 (Masked/Hidden CPU @ 2.40GHz)

(Lower is better)

prime benchmark results

(oops, typo in the chart should say Vultr)

Benchmark the file IO

Confirm free space

df -h /

Install Sysbench

sudo apt-get install sysbench

I had 10GB free on all servers (Vultr, Digitial Ocean and UpCloud) so I created a 10GB test file.

sysbench --test=fileio --file-total-size=10G prepare
sysbench 0.4.12:  multi-threaded system evaluation benchmark

128 files, 81920Kb each, 10240Mb total
Creating files for the test...

Now I can run the benchmark and use the pre-created text file.

sysbench --test=fileio --file-total-size=10G --file-test-mode=rndrw --init-rng=on --max-time=300 --max-requests=0 run

SysBench description from the Ubuntu manpage.

“SysBench is a modular, cross-platform and multi-threaded benchmark tool for evaluating OS parameters that are important for a system running a database under intensive load. The idea of this benchmark suite is to quickly get an impression about system performance without setting up complex database benchmarks or even without installing a database at all.”

SysBench Results (Vultr/Sydney)

sysbench 0.4.12:  multi-threaded system evaluation benchmark

Running the test with following options:
Number of threads: 1
Initializing random number generator from timer.


Extra file open flags: 0
128 files, 80Mb each
10Gb total file size
Block size 16Kb
Number of random requests for random IO: 0
Read/Write ratio for combined random IO test: 1.50
Periodic FSYNC enabled, calling fsync() each 100 requests.
Calling fsync() at the end of test, Enabled.
Using synchronous I/O mode
Doing random r/w test
Threads started!
Time limit exceeded, exiting...
Done.

Operations performed:  385920 Read, 257280 Write, 823266 Other = 1466466 Total
Read 5.8887Gb  Written 3.9258Gb  Total transferred 9.8145Gb  (33.5Mb/sec)
 2143.98 Requests/sec executed

Test execution summary:
    total time:                          300.0026s
    total number of events:              643200
    total time taken by event execution: 182.4249
    per-request statistics:
         min:                                  0.01ms
         avg:                                  0.28ms
         max:                                 18.12ms
         approx.  95 percentile:               0.55ms

Threads fairness:
    events (avg/stddev):           643200.0000/0.00
    execution time (avg/stddev):   182.4249/0.00

SysBench Results (Digital Ocean/London)

sysbench 0.4.12:  multi-threaded system evaluation benchmark

Running the test with following options:
Number of threads: 1
Initializing random number generator from timer.


Extra file open flags: 0
128 files, 80Mb each
10Gb total file size
Block size 16Kb
Number of random requests for random IO: 0
Read/Write ratio for combined random IO test: 1.50
Periodic FSYNC enabled, calling fsync() each 100 requests.
Calling fsync() at the end of test, Enabled.
Using synchronous I/O mode
Doing random r/w test
Threads started!
Time limit exceeded, exiting...
Done.

Operations performed:  944280 Read, 629520 Write, 2014432 Other = 3588232 Total
Read 14.409Gb  Written 9.6057Gb  Total transferred 24.014Gb  (81.968Mb/sec)
 5245.96 Requests/sec executed

Test execution summary:
    total time:                          300.0024s
    total number of events:              1573800
    total time taken by event execution: 160.5558
    per-request statistics:
         min:                                  0.00ms
         avg:                                  0.10ms
         max:                                 18.62ms
         approx.  95 percentile:               0.34ms

Threads fairness:
    events (avg/stddev):           1573800.0000/0.00
    execution time (avg/stddev):   160.5558/0.00

SysBench Results (UpCloud/Singapore)

sysbench 0.4.12:  multi-threaded system evaluation benchmark

Running the test with following options:
Number of threads: 1
Initializing random number generator from timer.


Extra file open flags: 0
128 files, 80Mb each
10Gb total file size
Block size 16Kb
Number of random requests for random IO: 0
Read/Write ratio for combined random IO test: 1.50
Periodic FSYNC enabled, calling fsync() each 100 requests.
Calling fsync() at the end of test, Enabled.
Using synchronous I/O mode
Doing random r/w test
Threads started!
Time limit exceeded, exiting...
Done.

Operations performed:  994320 Read, 662880 Write, 2121090 Other = 3778290 Total
Read 15.172Gb  Written 10.115Gb  Total transferred 25.287Gb  (86.312Mb/sec)
 5523.97 Requests/sec executed

Test execution summary:
    total time:                          300.0016s
    total number of events:              1657200
    total time taken by event execution: 107.4434
    per-request statistics:
         min:                                  0.00ms
         avg:                                  0.06ms
         max:                                 15.43ms
         approx.  95 percentile:               0.13ms

Threads fairness:
    events (avg/stddev):           1657200.0000/0.00
    execution time (avg/stddev):   107.4434/0.00

Comparison

Sysbench Results table

sysbench fileio results (text)

Read

  • Vultr (Sydney): 385,920
  • Digital Ocean (London): 944,280
  • UpCloud (Singapore): 944,320

Write

  • Vultr (Sydney): 823,266
  • Digital Ocean (London): 629,520
  • UpCloud (Singapore): 662,880

Other

  • Vultr (Sydney): 1,466,466
  • Digital Ocean (London): 3,588,232
  • UpCloud (Singapore): 2,121,090

Total Read Gb

  • Vultr (Sydney): 5.8887 Gb
  • Digital Ocean (London): 14.409 Gb
  • UpCloud (Singapore): 15.172 Gb

Total Written Gb

  • Vultr (Sydney): 3.9258 Gb
  • Digital Ocean (London): 9.6057 Gb
  • UpCloud (Singapore): 10.115 Gb

Total Transferred Gb

  • Vultr (Sydney): 9.8145 Gb
  • Digital Ocean (London): 24.014 Gb
  • UpCloud (Singapore): 25.287 Gb

Now I can remove test file io benchmark file

sysbench --test=fileio --file-total-size=2=10G cleanup
sysbench 0.4.12:  multi-threaded system evaluation benchmark

Removing test files...

Confirm the test file has been deleted

df -h /
Filesystem      Size  Used Avail Use% Mounted on
/dev/vda1        40G   16G   23G  41% /

Bonus: Benchmark MySQL (on my main server (Vultr) not on Digital Ocean and UpCLoud)

I tried to run a command

sysbench --test=oltp --oltp-table-size=1000000 --db-driver=mysql --mysql-db=test --mysql-user=root --mysql-password=#################################### prepare
sysbench 0.4.12:  multi-threaded system evaluation benchmark

FATAL: unable to connect to MySQL server, aborting...
FATAL: error 1049: Unknown database 'test'
FATAL: failed to connect to database server!

To fix the error I created a test table with Adminer (guide here).

Create Test Table

< Previous – Next >

Read Part 1, Part 2, Part 3 or Part 4

Measuring VM performance (CPU, Disk, Latency, Concurrent Users etc) on Ubuntu and comparing Vultr, Digital Ocean and UpCloud – Part 2 of 4

by

How can you measure VM performance (CPU, Disk, Latency, Concurrent Users etc) on Ubuntu and comparing Vultr, Digital Ocean and UpCloud – Part 2 of 4

Advertisement:



Read Part 1, Part 2, Part 3 or Part 4

Measure Disk Performance with Bonnie++

Installing Bonnie++ on Ubuntu

apt-get install bonnie++

Read this. post on using Bonnie++

Benchmark disk IO with DD and Bonnie++

Starting Bonnie++

bonnie++ -d /tmp -r 2048 -u username

Bonnie++ Readme.

Disk io with bonnie++ on Vultr/Sydney

Writing a byte at a time...done
Writing intelligently...done
Rewriting...done
Reading a byte at a time...done
Reading intelligently...done
start 'em...done...done...done...done...done...
Create files in sequential order...done.
Stat files in sequential order...done.
Delete files in sequential order...done.
Create files in random order...done.
Stat files in random order...done.
Delete files in random order...done.
Version 1.97 ------Sequential Output------ --Sequential Input- --Random-
Concurrency 1 -Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--
Machine Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP /sec %CP
servername 4G 656 99 308954 68 113706 33 1200 92 188671 30 10237 251
Latency 26067us 119ms 179ms 29139us 26069us 16118us
Version 1.97 ------Sequential Create------ --------Random Create--------
servername -Create-- --Read--- -Delete-- -Create-- --Read--- -Delete--
files /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP
16 +++++ +++ +++++ +++ +++++ +++ +++++ +++ +++++ +++ +++++ +++
Latency 1463us 703us 880us 263us 119us 593us
1.97,1.97,servername,1,1528177870,4G,,656,99,308954,68,113706,33,1200,92,188671,30,10237,251,16,,,,,+++++,+++,+++++,+++,+++++,+++,+++++,+++,+++++,+++,+++++,+++,26067us,119ms,179ms,29139us,26069us,16118us,1463us,703us,880us,263us,119us,593us

Disk io with bonnie++ on Digital Ocean/London

Writing a byte at a time...done
Writing intelligently...done
Rewriting...done
Reading a byte at a time...done
Reading intelligently...done
start 'em...done...done...done...done...done...
Create files in sequential order...done.
Stat files in sequential order...done.
Delete files in sequential order...done.
Create files in random order...done.
Stat files in random order...done.
Delete files in random order...done.
Version 1.97 ------Sequential Output------ --Sequential Input- --Random-
Concurrency 1 -Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--
Machine Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP /sec %CP
servername 4G 699 99 778636 74 610414 60 1556 99 1405337 59 +++++ +++
Latency 17678us 10099us 17014us 7027us 3067us 2366us
Version 1.97 ------Sequential Create------ --------Random Create--------
servername -Create-- --Read--- -Delete-- -Create-- --Read--- -Delete--
files /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP
16 +++++ +++ +++++ +++ +++++ +++ +++++ +++ +++++ +++ +++++ +++
Latency 1243us 376us 611us 108us 59us 181us
1.97,1.97,servername,1,1528186398,4G,,699,99,778636,74,610414,60,1556,99,1405337,59,+++++,+++,16,,,,,+++++,+++,+++++,+++,+++++,+++,+++++,+++,+++++,+++,+++++,+++,17678us,10099us,17014us,7027us,3067us,2366us,1243us,376us,611us,108us,59us,181us

Disk io with bonnie++ on UpCloud/Singapore

Writing a byte at a time...done
Writing intelligently...done
Rewriting...done
Reading a byte at a time...done
Reading intelligently...done
start 'em...done...done...done...done...done...
Create files in sequential order...done.
Stat files in sequential order...done.
Delete files in sequential order...done.
Create files in random order...done.
Stat files in random order...done.
Delete files in random order...done.
Version 1.97 ------Sequential Output------ --Sequential Input- --Random-
Concurrency 1 -Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--
Machine Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP /sec %CP
servername 4G 1014 99 407179 24 366622 32 2137 99 451886 17 +++++ +++
Latency 11297us 54232us 16443us 4949us 44883us 1595us
Version 1.97 ------Sequential Create------ --------Random Create--------
servername -Create-- --Read--- -Delete-- -Create-- --Read--- -Delete--
files /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP
16 +++++ +++ +++++ +++ +++++ +++ +++++ +++ +++++ +++ +++++ +++
Latency 264us 340us 561us 138us 66us 327us
1.97,1.97,servername,1,1528226703,4G,,1014,99,407179,24,366622,32,2137,99,451886,17,+++++,+++,16,,,,,+++++,+++,+++++,+++,+++++,+++,+++++,+++,+++++,+++,+++++,+++,11297us,54232us,16443us,4949us,44883us,1595us,264us,340us,561us,138us,66us,327us

Now read this site on how to make sense of this data

 

< Previous – Next >

Read Part 1, Part 2, Part 3 or Part 4

Measuring VM performance (CPU, Disk, Latency, Concurrent Users etc) on Ubuntu and comparing Vultr, Digital Ocean and UpCloud – Part 1 of 4

by

How can you measure VM performance (CPU, Disk, Latency, Concurrent Users etc) on Ubuntu and comparing Vultr, Digital Ocean and UpCloud – Part 1 of 4. Update: I moved my domain to UpCloud.

Advertisement:



Update (June 2018): I moved my domain to UpCloud (they are that awesome). Use this link to signup and get $25 free credit. Read the steps I took to move my domain to UpCloud here.

Upcloud Site Speed in GTMetrix

Comparing Digital Ocean/Vultr and UpCloud Disk IO

I have a number of guides on moving away from CPanel, Setting up VM’s on AWS, Vultr or Digital Ocean (all in the search of extra performance) but how do you know when a server performance is ok apart from running GT Metrix and other external site benchmarking tools.

This post is split up as it was too long.

Read Part 1, Part 2, Part 3 or Part 4

Spoiler: It all depends on where your server is located and what you do with it (Tweaks will improve the performance).

P.S This is NOT a paid endorsement or conclusive test (just a quick benchmark/review).

What does your server do?

You need to know what your server does 24/7 and what resources the services need.

I use htop to view real-time and historical usage data for each process.

htop

Tweaking Advice

A friend gave me good advice re-tweaking a cheap host to get good performance

I am not a fan of just throwing more money at a host and expecting better performance. Host have unique features and cons., there is no shortage of hosts or host cons.

How can you run synthetic benchmarks to determine comparable performance metrics?

WARNING: Comparing synthetic benchmarks can be far removed from real-world speeds. Benchmark results below were from 3 different servers I have on 3 different hosts in three different locations (the only thing the same was the use of Ubuntu 16.04 $5/m servers). These results are not scientific and should not be used to compare host providers. Benchmark runs were one-off (not averages over multiple timezones/days).

Disk Performance

Speaking of disk performance I noticed this the other day on the RunCloud blog. Faster than SSD (UpCloud)?

UpCloud Faster-than-SSD Cloud Hosting Server (Promo Code Inside)

Runcloud is a server management console that can interface with your domains (read my old review here).  I don’t use Runcloud but it is great for those who need a GUI to help manage VM via a dashboard. However, I prefer to know what is going on under the hood. I have investigated webmin in the past though.

Let’s do a quick IO benchmark test between UpCloud, Digital Ocean and Vultr on similarly low end $5/m servers,

Good advice on command line benchmarking tools from a friend.


Installing iozone to test disk performance

I searched for a post on using iozone (Thanks thegeekstuff).  I will be reviewing the “Writer report” and “Reader report”. Read more about iozone here.

View the iozone page for how to break down results.

iozone results breakdown

(image snip from http://www.iozone.org/)

Install iozone on Ubuntu

sudo apt-get install iozone3

Run an iozone disk test and output the results to a spreadsheet.

iozone -a -b iozone.xls

Now let’s run a Read/Write test on Vultr/Digital Ocean and UpCloud. Multiple runs were not performed, this is not a scientific test (just a simple benchmark test (as is, ignoring sever load and local infrastructure/timezone load)).

iozone Benchmark results for Vultr “Read” (Sydney)

 “4” “8” “16” “32” “64” “128” “256” “512” “1024” “2048” “4096” “8192” “16384”
6421337303363612427406245647866421025
1282248149353656641359587082197413595811720614
2561884399269916138790453667079597167856870205687020
51231404883736016368473342625234610256263881650671425684095
1024161780819392073411938399976240487784614246308368058850836609617
2048192651025696784423683499761839370754596052896324354252449715854707314
409617016832151300420992050017004751325486984553892463647681492852162070354347346
8192206342423293463203763293728032214853232699362643136507063789200411060337150454350542
163841738553277836233976133679205369344231715013524291339358630040243552531345657426938452488861
327680000295289435371533574875376815547196133890280339499527352222542914
655360000405748936107893619967380007832753273591212360718817704262826659
1310720000355227018907425275167372733935276071753893323473623411111378601
2621440000379858613020211491429371282532288163757963371551025924852481061
5242880000275875624879233705741180732821183093675988319636733943302396842

iozone Benchmark results for Digital Ocean “Read” (London)

4 “8” “16” “32” “64” “128” “256”  “512”  “1024”  “2048” “4096”  “8192”  “16384”
644564786710039790061791040217812902017
12847174347082197854812497958961056714010779307
256484091170731328271916986843310148242106515981E+07
51247426166909408814039993042929638369100440891E+0710044089
102442490535917516620834375375999300377104549847E+0671131619946527
204838854316967792660354968456291040188398080369E+06790383693084977817519
40962506983595323162636116953144777437962250286E+068081580768397280815808240513
81923665114485046354793176141364627712061086086E+0665699835732541716603366334025479317
163843673501482858454161826187150661476162988726E+06643031059840336402750604615947918833405527
327680000469254261409296E+06629564252312246545707578110845134753702577
655360000631543058301316E+06644469562191256473838533859542481183679324
1310720000613000264614966E+06595806859834236387547613807839948883602079
2621440000645674663237276E+06650414663901766486151643396339551653654188
5242880000166733763814566E+06644570864487146421071598120041551853770740

iozone Benchmark results for UpCloud “Read” (Singapore)

 “4” “8” “16” “32” “64” “128” “256” “512” “1024” “2048” “4096” “8192” “16384”
6464210256421025108215241290201715972885
128488928164061389129573107793071420079414200794
25653206713879045107583228815202102450711281227712228612
512430525051154228844453823403670919528394979754017010235583
10244339202476263058212716163794681951146745106479979818391810230845
20484204968531948458008515816563624356663780055953632685108979403678229438
409645260135556581481794854045047301864575963458102806007355691953886209456281934
8192429829550190935927357603670267813416082655585563665275466553692679206564661264437634
163844282172584955863139196635840674195866570546423097553662265585756442970452703237847773901898
327680000582546054234086504198666538563653296426343526307637186053705971
655360000690807566231166493259660973863118056483610548967440359823561526
1310720000565018057189492465429539125334959115784844536740837334903582175
2621440000681462766912506189661590678660816455799913524791941212503637601
5242880000640476463092635673979575160962882456305103597868039119843767116

iozone Benchmark results for Vultr “Write” (Sydney)

 “4” “8” “16” “32” “64” “128” “256” “512” “1024” “2048” “4096” “8192” “16384”
64289322532815507625429630566551
128398921465304434078417212669577821147
256530031613985820398474937891956815414370025
512387576754083709019819085702295609421924123496091
102429723344852271608992348885407381734012031371072453601636
204840869763465569538313581345496571295458821154797520964207258493
409623615043380412157741245025820832809958137133991426992108310046821481431
819261111366667780628671521977982582429487594787009110463787911921023592453248
16384435454706149718313845499893495888068812778842885820591941120839610862672406590
327680000465196786067938881627294890917968147872369871329842843
655360000515057790172937568915601897235867197907562852002743856
13107200005010914804928131478708868802398053336846301117578633185
2621440000387126323185323656473258405744369599422554468992453563
5242880000325588380450392965451608303255355148386250432054416512

iozone Benchmark results for Digital Ocean “Write” (London)

 “4” “8” “16” “32” “64” “128” “256” “512” “1024” “2048” “4096” “8192” “16384”
64831569566551127944713639611392258
1286524881319723142102399089116631391561553
2561185399115232315343421598292182669517075891514860
51211665991296159139918916209801620980136192015897791672748
1024107919013212001584972191756215926121701108171812014629601643814
20481210394147017216217191550584179637816437531713598175958116491171488257
40969165131287575157471814065941742237173414816524181583280159934616610451533532
8192110974513187481178567154420115023401371492146674714995211479759156487812912921347609
163841106205128208413740371503649142939814614071496119157813215472891333431120337111988151501316
327680000127091414065891513114146822615583031552038151633614432801440360
655360000131932213279841311504141195512669881359645138644613470921368295
1310720000110065812293261227197131863112655521233306122774712378961233502
2621440000116716010640781155828118518510861521193673108087210626111141960
524288000097783511248161052757121918311289721140177109195411416351132063

iozone Benchmark results for UpCloud “Write” (Singapore)

 “8” “16” “32” “64” “128” “256” “512” “1024” “2048” “4096” “8192” “16384”
6411432231255511156243614525281279447
128145176414061361543594150465918525201749872
2561642294182980819708711855098180216719529472000242
51215374241854787180187322947961983258212452618957211417662
1024143413815534421609925193135920983752044438187241917683451892218
20481562145190177118172811848169196709712962402267786208149719157682007554
409616253721966378192474113420921950306207817519148731459656199515221028491326855
8192144406218083301956503192439721273002042328213563019864782062557206131913370161812049
163841667066182024818984952051339201253021110802119806149121720608751974254193478918158231921911
327680000205750614545372075621207089918697952052896189234718553821873440
655360000206712720776732088994217980920874712099108190472316425051832204
1310720000123466318249591304340177551412874811560379163199210856091675467
2621440000685774808487823824662524681762548308814946645663732176
5242880000547296517384503422521173538714518429528950529593512944

Here is my quick unscientific take on a one pass benchmark results above.

Vultr (Read) Vultr (Write) Digital Ocean (Write) UpCloud (Read) UpCloud (Write)

These results need some decoding.

Next >>

Read Part 1, Part 2, Part 3 or Part 4

Ask a question or recommend an article

[contact-form-7 404 "Not Found"]

Update (June 2018): I moved my domain to UpCloud (they are that awesome). Use this link to signup and get $25 free credit. Read the steps I took to move my domain to UpCloud here.

Upcloud Site Speed in GTMetrix

Revision History

v1.2 added the fact that I Moved to UpCloud.

v1.1 Re ran iozone -a -b iozone.xls on all servers.

v1.0 Initial post

Setting up a website to use Cloudflare on a VM hosted on Vultr and Namecheap

by

This guide will show how you can set up a website to use Cloudflare on a VM hosted on Vultr and Namecheap

Advertisement:



I have a number of guides on moving hasting away form CPanel, Setting up VM’s on AWS, Vultr or Digital Ocean along with installing and managing WordPress from the command line. This post will show how to let Cloudflare handle the DNS for the domain.

Update 2018: For the best performing VM host (UpCloud) read my guide on the awesome UpCloud VM hosts (get $25 free credit by signing up here).

Snip from here “Cloudflare’s enterprise-class web application firewall (WAF) protects your Internet property from common vulnerabilities like SQL injection attacks, cross-site scripting, and cross-site forgery requests with no changes to your existing infrastructure.”

Buy a Domain 

Buy a domain name from Namecheap here.

Domain names for just 88 cents!

Cloudflare Benefits (Free Plan)

  • DDoS Attack Protection (Huge network to absorb attacks DDoS attacks over 600Gbps are no problem for our 15 Tbps networks)
  • Global CDN
  • Shared SSL certificate (I disabled this and opted to use my own)
  • Access to audit logs
  • 3 page rules (maximum)

View paid plan options here.

Cloudflare CDN map

Cloudflare CDN says it can load assets up to 2x faster, 60% less bandwidth from your servers by delivering assets from 127 data centres.

Cloudflare Global Network

Setup

You will need to sign up at cloudflare.com

Cloudflare

After you create an account you will be prompted to add a siteAdd SiteCloudflare will pull your public DNS records to import.

Query DNS

You will be prompted to select a plan (I selected free)

Plan Select

Verify DNS settings to import.

DNS Import

You will now be asked to change your DNS nameservers with your domain reseller

DNS Nameservers

TIP: If you have an SSL cert (e.g Lets Encrypt) already setup head to the crypto section and select ” Full (Strict)” to prevent ERR_TOO_MANY_REDIRECTS errors.

Strict SSL

Cloudflare UI

I asked Twitter if they could kindly load my site so I could see if Cloudflare dashboard/stats were loading.

The Cloudflare CTO responded.  🙂

Confirm Cloudflare link to a domain from the OSX Comand line

Caching Rule

I set up the following caching rule to cache everything for 8 hours instead of WordPress pages

Page Rules

“fearby.com.com/wp-*” Cache level: Bypass

“fearby.com.com/wp-admin/post.php*” Cache level: Bypass

“fearby.com/*” Cache Everything, Edge Cache TTL: 8 Hours

Cache Results

Cache appears to be sitting at 50% after 12 hours.  having cache os dynamic pages out there is ok unless I need to fix a typo, then I need to login to Cloudflare and clear the cache manually (or wait 8 hours)

Performance after a few hours

DNS times in gtmetrix have now fallen to a sub 200ms (Y Slow is now a respectable A, it was a C before).  I just need to wait for caching and minification to kick in.

DNS Improved

webpagetest.org results are awesome

See here: https://www.webpagetest.org/result/180314_PB_7660dfbe65d56b94a60d7a604ca250b3/

  • Load Time: 1.80s
  • First Byte 0.176s
  • Start Render 1.200s

webpagetest

Google Page Speed Insights Report

Mobile: 78/100

Desktop: 87/100

Check with https://developers.google.com/speed/pagespeed/insights/

Update 24th March 2018 Attacked?

I noticed a spike in and traffic (incoming and threats) on the 24th of March 2018.

I logged into Cloudflare on my mobile device and turned on Under Attack Mode.

Under Attack Flow

Cloudflare was now adding a delay screen in the middle of my initial page load. Read more here.  A few hours after the Attach started it was over.

After the Attack

I looked at the bandwidth and found no increase in traffic from my initial host VM. Nice.

cloudflare-attack-001

Thanks, Cloudflare.

Cloudflare Pros

  • Enabling Attack mode was simple.
  • Soaked up an attack.
  • Free Tier
  • Many Reports
  • Option to force HTTPS over HTTP
  • Option to ban/challenge suspicious IP’s and set challenge timeframes.
  • Ability to setup IP firewall rules and Application Firewalls.
  • User-agent blocking
  • Lockdown URL’s to IP’s (pro feature)
  • Option to minify Javascript, CSS and HTML
  • Option to accelerate mobile links
  • Brotli compression on assets served.
  • Optio to enable BETA Rocket loader for Javascript performance tweaks.
  • Run Javascript service workers from the 120+ CDN’s
  • Page/URL rules o perform custom actions (redirects, skip cache, Encryption etc)
  • HTTP/2 on, IPV6 ON
  • Option to setup load balancing/failover
  • CTO of Cloudflare responded in Twitter 🙂
  • Option to enable rate limiting (charged at 10,000 hits for $0.05c)
  • Option to block countries (pro feature)
  • Option to install apps in Cloudflare like(Goole Analytics,

Cloudflare Cons

  • No more logging into NameCheap to perform DNS management (I now goto Cloudflare, Namecheap are awesome).
  • Cloudflare Support was slow/confusing (I ended up figuring out the redirect problem myself).
  • Some sort of verify Cloudflare Setup/DNS/CDN access would be nice. After I set this up my gtmetrix load times were the same and I was not sure if DNS needs to replicate? Changing minify settings in Cloudflare did not seem to happen.
  • WordPress draft posts are being cached even though page riles block wp-admin page caching.
  • Would be nice to have ad automatic Under Attack mode
  • Now all sub-domains were transferred in the setup ( id did not know for weeks)

Cloudflare status

Check out https://www.cloudflarestatus.com/ for status updates.

Don’t forget to install the CloudFlare Plugin for WordPress if you use WordPress.

More Reading

Check out my OWASP Zap and Kali Linux self-application Penetration testing posts.

I hope this guide helps someone.

Ask a question or recommend an article

[contact-form-7 404 "Not Found"]

Revision History

v1.8 host Command from the OSX CLI

v1.7 Subdomain error

v1.6 Cloudflare Attack

v1.5 WordPress Plugin

v1.4 More Reading

v1.3 added WAF snip

v1.2 Added Google Page Speed Insights and webpage rest results

v1.1 Added Y-Slow

v1.0 Initial post

Speeding up WordPress with the ewww.io ExactDN CDN and Image Compression Plugin

by

Below is my quick blog posts on using the EWWW IO ExactDN CDN plugin in WordPress to set up an ExactDN (Global Dis.cotribution Network (CDN)) to distribute images to my site’s visitors and shrink (optimize) images in posts.

Advertisement:



I have blogged before on speeding up WordPress that has involved moving servers away from CPanel domains to self-managed servers (e.g on Digital Ocean or Vultr), using Lazy Load image plugins like BJ Lazy Load, Optimize images automatically in WordPress with EWWW.io and scaling and moving servers closer to your customers.

For the best performing VM host (UpCloud) read my guide on the awesome UpCloud VM hosts (get $25 free credit by signing up here).

Today I am going to set up the https://www.ewww.io/resize/ ExactDN (CDN) delivery network. I am paying for this myself (and this is an unbiased review).

FYI: I use Ubuntu servers and not Windows.

Know your starting point. 

“If you fail to plan, you are planning to fail!” – Benjamin Franklin

In my original post about Speeding up WordPress, I used the webpagetest.org site to test my sites response times, I was getting an embarrassing 21 seconds load time and 6 second first-byte time. I have worked to speed up WordPress by moving WordPress to a self-managed server (away from CPanel), used the BJ Lazy Load plugin and the awesome ewww.io image optimize plugin, now I get about 4-5 seconds.

I am hoping adding a CDN will make things faster. My blog is delivering over two-thirds images, perfect for a CDN, this is why I am trying this out.

Image type cdn

TIP: Check where your customers/readers are located, and how many are New versus Returning customers? Do you need a CDN to deliver content (Images) that are closer to your customers/readers or do you need to move your web server somewhere else?  The more you know the more you can help them. Worst case you will be supporting a positive experience (and potentially turning a one time visitor into a returning visitor).

I looked at my Google Analytics data to see where my visitors are. Whether good or bad, they are all over the world (Hello)?

World

Other data is available in Google Analytics.  I can see the last few years growth is growing and I am getting more returning visitors, now is the time to ensure my site is ready for more traffic and returning visitors.

Data

Note: The fall in traffic in the Audience overview (right-hand side of the left image) is the unfinished month (not a reader fall off).

Personally, I set a goal to have a high page bounce rate of 90% be way lower (at present I am at about 80% and falling (good) and my page read time has gone from 40 seconds to 1 minute 40.  Every bit you can do will help create a positive experience and help your visitors. I can see from the data above the content is being read, I am building returning visitors and they are geographically spread out. A CDN will be great. After you know where your visitors are it is good to know the times of day that your visitors are hitting your site. Lucky for me it is spread evenly over a 24 hour period.

Data Quantity

FYI: My servers outgoing data (last 30 days), not huge but BJ LazyLoad and image optimization may be helping.

Outbound

Traffic Forecast

Do you know the forecasted growth of your website?

Site Growth

Measure Before Optimizing

Before I started to optimize my WordPress site (hosted on a shared CPanel server) I had the following Web Page Test score. I tested from Singapore as that’s was where my server was originally (and the closest to me).

TIP: Read more about Performance, First Byte, Start Render and Complete scores at my blog post here.

Web Page Performace Test Results (Before Optimizations):

  • Load Time: 23.672s
  • First Byte: 6.743s
  • Start Render: 11.8300s
  • Speed Index: 15024
  • Requests: 132/164 (Document complete v Fully Loaded)
  • Bytes: 3,346KB/3,454KB  (Document complete v Fully Loaded)

Quick Web Page Performace Score Card (Before Optimizations):

  • First Byte: F (I should have captured the subscores)
  • Keep Alive Enabled: F (I should have captured the subscores)
  • Compress Transfer: F (I should have captured the subscores)
  • Compress Images: A (I should have captured the subscores)
  • Cache Static Content: X (I should have captured the subscores)
  • Effective use of CDN: X (I should have captured the subscores)

My initial scores were bad across the range of tests (before optimisations). On the upside, I was manually compressing images with a tool on my desktop before uploading images and this showed an “A” but this scorecard overall was really bad.

Here are the results after Quick Optimizations (EWWW.io image compress, moved the server, reorganizing the site and Lazy Load Images)

Now I get these results after speeding up my site (after using the EWWW.io image resizing, reorganizing the site, minifying, lazy load images, moving servers etc.).

Web Page Performace Test Results (After Simple Optimizations):

  • Load Time: 8.823s (down 14.849s)
  • First Byte: 3.5533s (down 3.1897s)
  • Start Render: 5,4800s (down 6.35s)
  • Speed Index: 5594 (down 9430)
  • Requests: 73/76 (Document complete v Fully Loaded) (down 59/88)
  • Bytes: 848KB/855KB  (Document complete v Fully Loaded) (down 2,498KB/2,599KB)

Quick Web Page Performace Score Card (After Simple Optimizations):

  • First Byte: F (I should have captured the subscores)
  • Keep Alive Enabled: A (I should have captured the subscores)
  • Compress Transfer: A (I should have captured the subscores)
  • Compress Images: A (I should have captured the subscores)
  • Cache Static Content: B (I should have captured the subscores)
  • Effective Use of CDN: X  (I should have captured the subscores)

Tested my pagespeed tests

Even at 4 seconds web page “First Byte“, this is considered not good. My brain says I want sub 1 second, I doubt this is achievable with WordPress over thousands of miles away with SSL (read here about scalability).

I know https and non-geographically favourable servers add half a second to data. SSL will add processing overheads and latency period. If you only want speed don’t setup SSL but if you want SEO and security then setup SSL.

locationlocationlocation

WebPageTest.org test reveals there is no effective use of Content Delivery Networks (CDN) on fearby.com (that’s why I am about to install EWWW.io ExactDN).

ExactDN (Content Delivery Network)

I did try and set up a number of caching and CDN plugins in the past (e.g Max CDN, W3 Total Cache, WP Fastest Cache, Cache Enable, WP Rocket, WP Super Cache, etc.) but they either made results worse or were impossible to set up.

Now that EWWW.io has a CDN let’s give that a go.

What is EWWW.io’s ExactDN?

You can read more about EWWW.io’s two-pronged approach to delivering one plugin to A) “compress images” and B) “add a Content Delivery Network (CDN)” here: https://ewww.io/resize/

What is ExactDN

Ensure you read up about EWWW.io’s ExactDN here: https://ewww.io/resize/ . If you have not used EWWW.io check out my review of the EWWW.io image compression plugin here first.

Pre Signup

Purchasing and Installing ExactDN Exact DN

Login to your EWWW.io account (or signup then log in).

Signin

FYI: If you have used the Optimise images automatically in WordPress plugin from EWWW.io plugin before, then you will see past purchases here.

Signed in

Now you can go back to the EWWW.io ExactDN product page (https://ewww.io/resize/) and purchase a subscription (Make sure you are logged in with an EWWW.io account before you purchase ExactDN).

Pre purchase

I purchased an ExactDN monthly subscription for $9.00 monthly (with a $1 signup fee for the cloud compression service).

Order

Purchase confirmation screen.

Order

Post-purchase, I was advised to find my “Site Address (URL)” in WordPress and add it to EWWW.io Manage Sites screen.

Add site to the plugin

I now noticed I had a CDN for my domain ( fearby-com.exactdn.com ). Nice.

CDN Site Created

EWWW.io said to tick the CDN option in the Image Resize area of the EWWW.io plugin. But before I do that I  will update WordPress Core and WordPress Plugins before enabling the ExactDN as they are out of date.

Update

TIP: I update WordPress Core and WordPress plugins via command line (my guide here).

Backup WordPress (just in case)

It is always a good idea to backup your website, I logged into my Ubuntu server and checked the size of my site before backing it up.

I copied the website folder (from “/www/” to “/www-backup”)

I confirmed the copied folder size (same, good)

I dumped all databases with the MySQL dump command

While I was there I compress the SQL backup files (text files)

Checking the CDN Status

I checked the DNS replication for fearby-com.exactdn.com with this DNS checker. It’s setup and ready to go across the globe 🙂

A quick CNAME check reveals its upstream provider is fearbycom-8ba8.kxcdn.com. Keycdn.com have been around since 2012. It is great that EWWW.io has paired with keycdn and included their image compression magic in a single WordPress plugin.

Configuring the EWWW.io plugin in WordPress.

Before you enable the CDN below do check the https://www.whatsmydns.net/ and see if your CDN has replicated around the world (Australia can be a bit slow at DNS replication from time to time). Do wait at least 10 minutes before proceeding.

Click the settings in your EWWW Image Optimizer Plugin.

Configure

Now click the Resize Settings tab and click Enable CDN in the EWWW Image Optimizer plugin screen.

Enable Exact DN

Don’t forget to click Save Changes

Save

 

Testing the CDN

https://www.webpagetest.org NOW reports that my site is using a CDN 🙂

FYI: I have always used webpage test from Singapore and I have done the same here to compare apples to apples.  Singapore is not a good test location in Australia (my server is in Australia) and it is 200ms away and the added layer of SSL adds latency to the connection (read here). This is a real-world test though (worst case).

CDN's Used:
fearby.com : 
fearby-com.exactdn.com : KeyCDN
fonts.googleapis.com : Google
fonts.gstatic.com : Google
static.addtoany.com : Cloudflare
www.googletagmanager.com : Google
pagead2.googlesyndication.com : Google
www.youtube.com : Google
adservice.google.com.sg : Google
adservice.google.com : Google
googleads.g.doubleclick.net : Google
www.google-analytics.com : Google
s.ytimg.com : Google
stats.g.doubleclick.net : Google
fearby-com.disqus.com : Fastly
ssl.google-analytics.com : Google

FYI: https://www.webpagetest.org does want other (all) static content to be on a CDN to give a higher score than 42 out 100, sorry I did not get a subscore before enabling the CDN.

I was expecting a green A here for CDN but Webpage Test has given me more items to investigate to ensure WordPress plugins are also fast (minify or move to CDN). It appears WordPress includes are my next target for optimizing.

Web Page Test indicates the following WordPress assets should also be in CDN.

FAILED - https://fearby.com/wp-content/plugins/youtube-embed-plus/styles/ytprefs.min.css?ver=230d722825ddde2688088d563a906075
FAILED - https://fearby.com/wp-content/plugins/simple-social-icons/svgxuse.js?ver=1.1.21
FAILED - https://fearby.com/wp-content/plugins/youtube-embed-plus/scripts/ytprefs.min.js?ver=230d722825ddde2688088d563a906075
FAILED - https://fearby.com/wp-content/plugins/add-to-any/addtoany.min.js?ver=1.0
FAILED - https://fearby.com/wp-content/plugins/crayon-syntax-highlighter/js/min/crayon.min.js?ver=_2.7.2_beta
FAILED - https://fearby.com/wp-includes/js/jquery/jquery-migrate.min.js?ver=1.4.1
FAILED - https://fearby.com/wp-includes/js/jquery/jquery.js?ver=1.12.4
FAILED - https://fearby.com/wp-content/plugins/contact-form-7/includes/js/scripts.js?ver=230d722825ddde2688088d563a906075
FAILED - https://fearby.com/wp-content/themes/genesis/lib/js/menu/superfish.args.js?ver=2.5.3
FAILED - https://fearby.com/wp-content/plugins/crayon-syntax-highlighter/css/min/crayon.min.css?ver=_2.7.2_beta
FAILED - https://fearby.com/wp-content/themes/genesis/lib/js/skip-links.js?ver=2.5.3
FAILED - https://fearby.com/wp-includes/js/comment-reply.min.js?ver=230d722825ddde2688088d563a906075
FAILED - https://fearby.com/wp-includes/js/hoverIntent.min.js?ver=1.8.1
FAILED - https://fearby.com/wp-content/themes/genesis/lib/js/menu/superfish.js?ver=1.7.5
FAILED - https://fearby.com/wp-includes/js/jquery/ui/tabs.min.js?ver=1.11.4
FAILED - https://fearby.com/wp-content/themes/news-pro/js/global.js?ver=3.2.2
FAILED - https://fearby.com/wp-content/themes/news-pro/js/responsive-menus.min.js?ver=3.2.2
FAILED - https://fearby.com/wp-includes/js/jquery/ui/core.min.js?ver=1.11.4
FAILED - https://fearby.com/wp-content/themes/news-pro/style.css?ver=3.2.2
FAILED - https://fearby.com/wp-includes/js/jquery/ui/widget.min.js?ver=1.11.4
FAILED - https://fearby.com/wp-content/themes/news-pro/js/jquery.matchHeight.min.js?ver=3.2.2
FAILED - https://fearby.com/wp-content/plugins/contact-form-7/includes/css/styles.css?ver=230d722825ddde2688088d563a906075
FAILED - https://fearby.com/wp-content/plugins/youtube-embed-plus/scripts/fitvids.min.js?ver=230d722825ddde2688088d563a906075
FAILED - https://fearby.com/wp-includes/js/wp-embed.min.js?ver=230d722825ddde2688088d563a906075
FAILED - https://fearby.com/wp-content/plugins/bj-lazy-load/js/bj-lazy-load.min.js?ver=2
FAILED - https://fearby.com/wp-content/plugins/disqus-comment-system/media/js/count.js?ver=230d722825ddde2688088d563a906075
FAILED - https://fearby.com/wp-includes/js/wp-emoji-release.min.js?ver=230d722825ddde2688088d563a906075
FAILED - https://fearby.com/wp-content/plugins/wp-seo-html-sitemap/style.css?ver=230d722825ddde2688088d563a906075
FAILED - https://fearby.com/wp-content/plugins/simple-social-icons/symbol-defs.svg
FAILED - https://fearby.com/wp-includes/css/dashicons.min.css?ver=230d722825ddde2688088d563a906075
FAILED - https://fearby.com/wp-content/themes/news-pro/images/favicon.ico
FAILED - https://fearby.com/wp-content/plugins/genesis-tabs/style.css?ver=230d722825ddde2688088d563a906075
FAILED - https://fearby.com/wp-content/plugins/simple-social-icons/css/style.css?ver=2.0.1
FAILED - https://fearby.com/wp-content/plugins/add-to-any/addtoany.min.css?ver=1.14

It would be a huge effort trying to read and keep static plugin and theme related files in a CDN. I’ll ask the EWWW.io developer to see if this is possible in a future version (that would be nice). The developer did promptly point me here to opt into using the CDN to deliver CSS, JS etc.

Is that it? It can’t be that simple!

I can load my site at https://fearby.com and my images load from https://fearby-com.exactdn.com 🙂 I am impressed, blog posts are now loading images from the CDN network and I did not have to edit posts. I did have to subscribe to ExactDN and tick a checkbox in WordPress though.

Here is a sample image (my largest) from this blog post.

https://fearby-com.exactdn.com/wp-content/uploads/2017/10/Infographic-So-you-have-an-idea-for-an-app-v1-3.jpg?strip=all&quality=60&ssl=1

fearby.com WebPAGE TEST Results – Singapore

Web Page Performace Test Results – Singapore (After Setting up the EWWW.io ExactDN):

  • Load Time: 4.177s (down 4.646s from previous optimizations ((My Original load times were 23.672s))
  • First Byte: 2.240s (down 1.3133s from previous optimizations (My Original First Byte: 6.743s))
  • Start Render: 2.800s (down 2.68s from previous optimizations (My Original Start Render: 11.8300s))
  • Speed Index: 3009 (down 2585 from previous optimizations (Speed Index: 15024))
  • Requests: 67/68 (Document complete v Fully Loaded) 
  • Bytes: 535KB/538KB  (Document complete v Fully Loaded)

Quick Web Page Performace Score Card – Singapore (After Setting up the EWWW.io ExactDN):

  • First Byte: F (I should have captured the subscores). 
  • Keep Alive Enabled: A (I should have captured the subscores).
  • Compress Transfer: A (I should have captured the subscores).
  • Compress Images: A (I should have captured the subscores).
  • Cache Static Content: B (I should have captured the subscores).
  • Effective Use of CDN: X (I should have captured the subscores). plugins now need to be on a CDN.

The Web Page Test site does give detailed scores and recommendations if you scroll down or click the score car (A-F). Do read the recommendations and see what you may need to do next.  I am happy that I now have a CDN via EWWW.io. Clicking the first byte and CDN buttons at Web Page Test reveal sub-scores to allow you to see if you have made improvements, regrettably, I did not know of and capture sub-scores until after installing the CDN (I suggest you do).

Full Dump.

GT Metrix Page Speed Score

https://gtmetrix.com is giving a good score across the board (86%). It hints I should optimize Javascript files and “Remove query strings in static files” as some proxy servers do not cache URLs with “?” in them. That is (fortunately) not a problem with ExactDN, as the servers are configured to properly handle query strings.

GTMetrics

Gtmetrix.com does give some optimisation tips too (However, it does report low CDN optimizations if a single file is not delivered over a CDN).

I do like Gtmetrix.com email reports, you can see if your site performance is degrading.

GTMetrix Summary

Page Speed Insights

I am now looking at the Google PageSpeed Insights test for things to fix next. I think I can tweak my NGINX a little (adding caching). Read more about Google Page Speed Insights here.

fyi: My Google Page Speed Insights score (desktop) along with another local big corporate site I tested.

Page Insight

I am happy with 82 🙂

It appears if you want to get 100% you need to get..

  • Initial Response under 100ms
  • Animate, produce frame in under 10ms
  • Idle maximize idle time
  • Deliver all content in under 1000ms

Getting More from ExactDN by caching CSS and JS files.

I added the following to my /www/wp-config.php file as mentioned here: http://docs.ewww.io/article/47-getting-more-from-exactdn This will serve more resources from the CDN

define( 'EXACTDN_ALL_THE_THINGS', true );

I refreshed my site in a browser and now CSS, JS and fonts are loaded from the CDN too.

All

To reduce files served from my website I re-enabled the Fast Velocity Minify plugin in WordPress and pointed it at my CDN (https://fearby-com.exactdn.com/)

cdn-minify

Now I am getting a GTMatrix score of Page Speed Score = 93% (A) and YSlow Score  = 74% (C) and a 2.6-second load and a much lower 35 requests (post minification and pointing the minified files to the CDN).

WOW.

Minifiy

And WebPageTest.org is reporting Effective use of CDN 🙂 Awesome.

Effective use of CDN

I think I’m done (ExactDN CDN and the image optimiser and other tweaks have worked its magic).

How do I compare to other sites?

Given Apple, Microsoft and NBC speed scores are worse than mine I’m happy for now.  🙂

GTMetricx Graphs

GT Metrics graphs show the improvement, YSlow report does indicate I can reduce the website DOM elements to speed things up and tweak plugins but I’d rather keep my design for SEO and not play with plugins or they might break.

I could tweak the server side (NGINX/MySQL/Cache or DNS) but I don’t need too.

Conclusion

I still can’t believe setting up a CDN is a one-click solution and adding a wp-config.php lin (after you subscribe). Best of all the BJ Lazy Load plugin still works. I am very happy with the EWWW.io ExactDN. I now have a CDN and it has lowered my First Byte time, Start Render Times, Speed Index time (on more than just the front page) and all I did was subscribe and tick a checkbox.

It is nice that EWWW.io have not charged for Data delivered from the CDN network on top of a monthly subscription at this stage but they have left it open in the terms and conditions. It is important to ensure a service can sustain itself, so this is a good sign. FYI: my favorite Agile toolkit (Atlaz.io) that I reviewed here closed shop today so it is wise for aaS shops to plan ahead. I know how hard it can be to spin up servers and stuff and allocate times to keep them running. Good luck EWWW.io!

Also, I now have a clear set of steps (below) to resolve other non-optimized assets that are outside of the CDN.

Full Report

Thanks to EWWW.io and Web Page Test for helping make my site faster.

I doubt WordPress and my “down under”  server location can get me to under 1000ms but I will try.

Read on here to see how Cloudflare can increase your site’s performance.

Setting up a website to use Cloudflare on a VM hosted on Vultr and Namecheap

DNS Improved

Don’t forget your sites performance/SEO and security.

Website

Update November 2018

I have a much faster loading website after moving it to a new host, read my guide here.

GTMetrix After YSlow

Donate and make this blog better

Ask a question or recommend an article

[contact-form-7 404 "Not Found"]

Revision History

v2.0 November 2018 Update

v1.9 added Cloudflare, SEO, Google Page Speed Insights test and future optimization, page insight speed image, added GTMetrics comments, growth image, gtmetrix.com reports (typos are free), updated change is domain to .com, added EXACTDN_ALL_THE_THINGS, 93 page speed, minify, compare, graph, fixed typos, tidied up the conclusion.