SiteGround participated for the fifth time in our WordPress Hosting Performance Benchmarks (2018). This review is based off the results of that test. Last year SiteGround competed in four price tiers and earned 3 Top Tier awards. This year SiteGround competed in the <$25, $25-50, and Enterprise tier.
SiteGround is one of the oldest competitors in our benchmarks and have for years now topped our reviews based on what people say on social media leading the shared category by a whopping 13% over the next highest competitor at the time of writing (Sep 2018).
The Products
Plan Name | Plan Monthly Price | Plan Visitors Allowed | Plan Disk Space | Plan Bandwidth |
Plan Sites Allowed
|
GrowBig | $14.95 | Suitable for ~ 25,000 Visits Monthly | 20GB | Unmetered | Unlimited |
GoGeek | $29.95 | Suitable for ~ 100,000 Visits Monthly | 30GB | Unmetered | Unlimited |
Enterprise | $4,800 | Unlimited | $1 per 1GB for distrbuted storage + shared file system | 10TB | Custom |
Enterprise was a custom setup: 8 servers - 2 load balancers, 4 application nodes (PHP workers) and two MySQL database servers. Every server had 16GB of RAM. The total RAM for the whole solution was 128GB RAM.
Performance Review
Load Storm
Total Requests | Total Errors | Peak RPS | Average RPS | Peak Response Time(ms) | Average Response Time (ms) | Total Data Transferred (GB) | Peak Throughput (MB/s) |
Average Throughput (MB/s)
|
|
<25 | 316,826 | 14 | 242.02 | 176.01 | 5,361 | 445 | 21.18 | 16.98 | 11.76 |
25-50 | 317,462 | 16 | 245.1 | 176.37 | 4,569 | 404 | 21.23 | 17.25 | 11.79 |
Enterprise | 1,443,091 | 15 | 1,151.40 | 801.72 | 15,086 | 209 | 105.15 | 84.68 | 58.42 |
Sources: <25, 25-50, Enterprise
The Load Storm test is designed to simulate real users coming to the site, logging in and browsing the site bursting some of the caching mechanisms typically found on managed WordPress hosts.
SiteGround aced the LoadStorm tests having minimal errors across every price tier. The enterprise configuration really shined here and makes it clear the value of all the redundancy and scalabliity where it had half the average response time and handled 1.43 million requests without issue.
Load Impact
Requests | Errors | Data Transferred (GB) | Peak Average Load Time (Seconds) | Peak Average Bandwidth (Mbps) |
Peak Average Requests/Sec
|
|
<25 | 318163 | 4 | 16.97 | 0.556 | 264 | 605 |
25-50 | 316927 | 0 | 16.89 | 0.531 | 280 | 642 |
Enterprise | 1627518 | 7 | 87.96 | 0.618 | 1370 | 3100 |
Sources: <25, 25-50, Enterprise
The Load Impact test makes sure static caching is effective so that if a page gets a lot of traffic the site will keep responding without issue.
SiteGround again aced the Load Impact tests. Nearly no errors on <25 and Enterprise and zero on the 25-50 test.
Uptime
UptimeRobot | StatusCake | |
<25 | 99.99 | 99.99 |
25-50 | 99 | 98.94 |
Enterprise | 99.96 | 100 |
The <25 and Enterprise didn't have any uptime issues. Unfortunately the 25-50 plan ran into some issues. SiteGround issued a statement about the issue:
"The recorded downtime for the SiteGround GoGeek hosting plan during the test was not a result of actual server availability issues. The uptime is below 99.9% because during performance testing an automatic limitation system temporarily kicked in, in a way that prevented the uptime test to be properly executed. SiteGround confirmed that this particular limitation system should not have been active on the tested accounts and is currently not active on any of their production servers."
WebPageTest / WPPerformanceTester
PHP Bench | WP Bench | |
<25 | 10.49 | 1157.407407 |
25-50 | 10.44 | 890.4719501 |
Enterprise | 10.457 | 196.6181675 |
The WPPerformanceTester results were average. The distributed nature of the Enterprise shows quite clearly how having redundant and scalable systems can reduce thoroughput on the WP Bench (MySQL) when you run multiple and separate database servers.
<25 | 25-50 | Enterprise | |
Dulles | 0.432 | 0.508 | 0.812 |
Denver | 1.489 | 1.358 | 1.342 |
LA | 0.909 | 0.894 | 1.06 |
London | 0.937 | 1.082 | 0.998 |
Frankfurt | 1.087 | 1.235 | 0.999 |
Rose Hill, Mauritius | 1.945 | 2.108 | 2.636 |
Singapore | 2.088 | 2.564 | 1.908 |
Mumbai | 1.957 | 2.329 | 2.109 |
Japan | 1.617 | 1.769 | 1.394 |
Sydney | 1.781 | 2.126 | 1.588 |
Brazil | 1.392 | 1.565 | 1.187 |
Nothing special to say here, it performed as expected.
Conclusion
SiteGround had excellent load testing performances. Load Storm and Load Impact results were great across the board. The <25 plan also had the second fewest errors cumulatively in its price tier. The 25-50 plan looked solid except there was an uptime issue which disqualified it from earning top tier status. SiteGround's Enterprise offering was a new entry in the tier. They made quite a splash with an 8 server configuration which handled our heaviest load tests without batting an eye. It's nice to see SiteGround offering great performance at an entry level price point and being able to scale that performance up to an Enterprise grade solution.
Kevin Ohashi
Latest posts by Kevin Ohashi (see all)
- Analyzing Digital Ocean’s First Major Move with Cloudways - February 28, 2023
- Removing old companies - June 28, 2021
- WordPress & WooCommerce Hosting Performance Benchmarks 2021 - May 27, 2021
This is a great post and cuts through the mess of information on page speed out there for thrive theme users. Thank you. Even though this is exactly what I am doing. It”s going to cut the time you spend trying to work out how to get good page speed. I just wanted to throw my experience of using Siteground (Grow Big) and WPX hosting in the UK. I test on gtmetrix (london) and pingdom (stockholm). Siteground started out fast, but over the year I”ve been with them the speed has dropped steadily by about 3 seconds. Starting out at 2 and dropping to about 5 on average with GT metrix. Pingdom 1.5 to 4 seconds. When I switched over to WPX everything became simpler. I consistently get 1.7/1.5 seconds with GTmetrix and 800ms with pingdom. Even without cloudflare. The main difference being Siteground is shared hosting and WPX is dedicated. Siteground support staff also varied in their response. One time, I spent 7 days complaining that my website was down and they kept telling me it wasn”t, until a senior person realised that a server had broken. He was honest at least. But it was a disappointing experience.