Best WordPress Hosting Round 4

wordpress-performance-comparison-round-4

 

Welcome to Round 4 of our WordPress hosting performance comparison! For more information on how my testing works and what I am trying to accomplish please see the original comparison here: Best WordPress Hosting Round 1

The short of it is I use Siege to simulate a large amount of users who are hitting the test site over 15 random URL’s (yes I changed from the original 11) with a random delay of 1-3 seconds per visit per user. Each test runs for 10 minutes and is closely monitored, If I see the site is really struggling or becomes unavailable I stop the test. The goal is to show what each host offers in terms of performance and just how large of a difference caching makes.

 

Let’s Get Started with our Best WordPress Hosting Round 4!

 

funio-wordpress hosting-performance review

 

Funio Environment, Resources, And Price

My experience with Funio did not start off all that well. My order was held for manual verification (my fault) after 6 hours or so I submitted a ticket asking how long it might take. I had no response for 3 days after which my order went through. Never was any reply to my ticket at all, not a confirmation or a sorry for the delay … nothing.

Note: While Funio offers “WordPress Hosting” there is no difference between this and their normal shared hosting.

 

Plan Name: Wordpress Hosting Pro

Plan Cost: $7.99/mo USD

Panel: cPanel

Ram: 2048MB

CPU: 100% / 1Core

IO: 1024KB/s

Webserver: Apache

Server Side Caching: No

Opcode Cache: Yes, Zend.

Mysql: MariaDB

PHP Memory: 128MB Default. Changed to 512MB for testing.

PHP 7: Yes

 

Funio First Run (No WordPress Cache)

I started at the usual of 25 Concurrent users and unfortunately Funio struggled with it. I noticed in cPanel in the extended server information that the server seems to be constantly and heavily using SWAP which is really not a good thing for all the people on that server. I would say this is most likely why the test results are so poor, anytime your server is constantly in swap and especially as heavily as this one the entire server is going to suffer. The server must have some bad configuration for it to be swapping when only using 30-40% of it’s available Ram.

funio-server-swapping

Test Results @25C

siege -c25 -d3 -t10M -i -f /usr/local/etc/funio_urls.txt

Lifting the server siege...
Transactions: 3500 hits
Availability: 99.91 %
Elapsed time: 599.84 secs
Data transferred: 116.21 MB
Response time: 2.78 secs
Transaction rate: 5.83 trans/sec
Throughput: 0.19 MB/sec
Concurrency: 16.21
Successful transactions: 3500
Failed transactions: 3
Longest transaction: 29.71
Shortest transaction: 0.19

 

Well as you can see failed transactions most likely due to a timeout (look at longest transaction), under 100% availability and a response time of almost 3 seconds. Really struggling.

funio-gtmetric-25C

funio-25c-resource-use

 

Not so good, let’s see if we can help it out some with WP Super Cache.

 

Funio Stress Test (WordPress Caching)

I tested 5 times at 25 Concurrent users and could not manage to get 100% availability, but server response time was ok, longest transaction was 7 seconds, and it was only 1 failed transaction. So I decided to let it slide and try increasing concurrent users until there was a bit more instability or slowness, unfortunately this happened quite easily and at 35 Concurrent it was game over.

 

siege -c35 -d3 -t10M -i -f /usr/local/etc/funio_urls.txt

Lifting the server siege...
Transactions: 12493 hits
Availability: 99.92 %
Elapsed time: 599.85 secs
Data transferred: 414.25 MB
Response time: 0.17 secs
Transaction rate: 20.83 trans/sec
Throughput: 0.69 MB/sec
Concurrency: 3.56
Successful transactions: 12493
Failed transactions: 10
Longest transaction: 11.42
Shortest transaction: 0.06

Between the support issue, an apparent server configuration issue (swap), and overall performance I would not recommend Funio Web Hosting.

 

milesweb-performance-review

Miles Web Environment, Resources, And Price

Miles Web I fairly often see advertisements on Reddit for and noticing they used Litespeed which I am a fan of so I decided to give them a try.

Note: Miles Web offers “WordPress Hosting” but there is no difference between that and it’s regular shared hosting offers.

 

Plan Name: Unlimited

Plan Cost: $6/mo USD

Panel: cPanel

Ram: NA

CPU: NA

IO: NA

Webserver: Litespeed

Server Side Caching: LScache

Opcode Cache: Yes, xCache.

Mysql: Mysql

PHP Memory: 128MB

PHP 7: No (PHP 5.6)

 

Miles Web First Run (No WordPress Cache)

As usual I started at 25 Concurrent users, and was hoping for better than normal performance because of Litespeed. The server handled 25 Concurrent users ok but just ok. Notice the response time, I had a feeling that things were going to go downhill quick.

 

siege -c25 -d3 -t10M -i -f /usr/local/etc/milesweb_urls.txt

Lifting the server siege...
Transactions: 4672 hits
Availability: 100.00 %
Elapsed time: 599.77 secs
Data transferred: 153.72 MB
Response time: 1.71 secs
Transaction rate: 7.79 trans/sec
Throughput: 0.26 MB/sec
Concurrency: 13.29
Successful transactions: 4672
Failed transactions: 0
Longest transaction: 4.80
Shortest transaction: 0.35

I bumped it up to 35 concurrent users and it maintained 100% availability as it did at 40 but both times response times suffered badly so while the site was available load times were 3 to 9 seconds in the area from 25C to to 40C. At 50 concurrent users the site becomes unusable with load times from 5 to 13 seconds. But is still did stay at 100% availability, a testament to the toughness of Litespeed as Apache when suffering from load times like that inevitably starts dropping sessions.

According to Miles Web support their servers support Litespeeds Cache plugin. If so there should be a massive improvement in performance, let’s see …. woops seems like somebody **** the bed. Note this is the second time I asked support about this exact thing both times I was told yes.

miles-web-wordpress-litespeed-cache-oboy

miles-web-wordpress-litespeed-cache-fail

For me that ends the Miles Web experience. The performance  even with Litespeed was not good, the support either lied, was oblivious, or both, and that is enough for me. I would not recommend Miles Web hosting.

 

bluehost-wordpress-performance-testing

 

bluehost Environment, Resources, And Price

Note:  Bluehost is owned by EIG and in many eyes has went downhill as is typical after EIG buys a company. So just a warning for those that consider going with bluehosts WordPress hosting or any of their hosting for that matter, it may not end well. 

That being said their WordPress hosting is pretty well done and performs quite good.

 

Plan Name: blogger

Plan Cost: $24.99/mo USD

Panel: cPanel + Custom

Ram: 2048MB

CPU: 2 Cores

IO: NA

Webserver: Apache + Nginx (Reverse proxy)

Server Side Caching: Memcached + Varnish

Opcode Cache: No

Mysql: Mysql

— Mysql on bluehost WordPress hosting is hosted on it’s own VPS which should allow for a increase in the amount of traffic it can handle (in theory).

PHP Memory: 512MB

PHP 7: No (PHP 5.6)

 

Bluehost First Run (No WordPress Cache)

No point really to run a WordPress caching plugin as this is already handled server side. Given the setup I knew it was going to perform well so I started off at 500 concurrent users which it handled quite easily. I doubled it and hit it right at the point where it starts to have issues, which is 1000 concurrent users. Here is a test result with 1000 concurrent:

 

siege -c1000 -d3 -t10M -i -f /usr/local/etc/bluehost_urls.txt

Lifting the server siege…
Transactions: 176138 hits
Availability: 100.00 %
Elapsed time: 599.50 secs
Data transferred: 5819.49 MB
Response time: 1.87 secs
Transaction rate: 293.81 trans/sec
Throughput: 9.71 MB/sec
Concurrency: 550.69
Successful transactions: 176138
Failed transactions: 1
Longest transaction: 10.91
Shortest transaction: 0.21

bluehost-1000c-gtmetrix

1000 concurrent would be the max that I would feel comfortable having it handle on a production site, things go downhill fast after that. At 1350 concurrent users I started seeing database connection issues and the site went away completely. It did come back by itself once I stopped the test. Maybe I killed Memcached? 😛  All in all this is some wicked performance and enough to take over 3rd spot on our top 5 if it was not an EIG owned company. Sorry Bluehost but I refuse to feature a EIG owned host in our Top 5 WordPress Hosts.

2 thoughts on “Best WordPress Hosting Round 4”

  1. Siege is a nice tool, but it is not really a good tool for testing companies. Any competent hosting company will drop the siege user-agent, or at minimum restrict the amount of concurrent connections by a single IP.

    I was not able to tell from your report, but are you running Siege from multiple IP’s or a single location?

    1. Hi,

      So far I have yet to see a company drop the siege user-agent though I agree it is definitely a possibility. I have tested multiple companies that use CloudLinux and limit Concurrent Connections (EP), and Siege does not seem to trigger that limit even with Concurrent users far over the limit. I usually run Siege from multiple VM’s as I have found that running from just 1 was causing some inconsistencies.

Leave a Reply