DigitalOcean vs. Linode in ChinaThis article compares the latency of DigitalOcean and Linode in a test from Shenzhen, China
DigitalOcean and Linode are two very popular developer focused cloud hosting providers. Each of them provide numerous hosting locations around the world and make it easy for developers to launch and scale websites and smaller web apps.
This article focuses on the performance of the various data centers from the two providers in Mainland China. The tests the article is based on were done using CloudHarmony’s testing environment and focused on latency as the key parameter.
What is DigitalOcean?
DigitalOcean is a New York based company with data centers in New York, Toronto, San Francisco, London, Frankfurt, Amsterdam, Bangalore and Singapore. At the time of writing almost 24 million droplets were launched on DigitalOcean’s servers.
According to DigitalOcean’s website, among some of the more famous clients using DigitalOcean are Atlassian, jQuery, HP, SalesForce, Ericssona and Zendesk. To find out more, go to their website: https://www.digitalocean.com/
What is Linode?
Linode, similar to DigitalOcean, is a US based company with 4 server locations in the US, two in Europe with Frankfurt and London and two in Asia with Tokyo and Singapore with Tokyo, however, being sold out at the moment.
The company states on their website that they have more than 400 000 customers worldwide with CreativeCommons and WP Engine among the more notable ones. To find out more about Linode go to https://www.linode.com/
The test process and its limitations
As with the CDN test and the general server test before, the testing was done using CloudHarmony. CloudHarmony allows you to test latency and download speed for various servers and cloud computing providers with DigitalOcean and Linode among those.
To see what the test is like, feel free to have a go at it yourself by clicking on the following link: http://cloudharmony.com/speedtest-latency-for-linode-and-digitalocean:compute
The test itself was limited to latency mainly because of the reason that this can be tested with very small data packages and is less likely to be influenced by spikes up or down in bandwidth, which would be the case for testing download speed. A more detailed discussion on what latency actually is can be found here: http://www.speedtest.net/articles/what-is-ping-what-is-latency/
Before we get to the results, a quick word on the test process and its limitations. Unfortunately, I didn’t have any influence on what types of servers or service packages were actually used by CloudHarmony. So it might be that CloudHarmony has access to the highest performing computing engine for provider x but only to the cheapest version for provider y, which might influence the results. This, unfortunately, is out of my control.
The test itself was conducted in Shenzhen, China, which is very close to Hong Kong. The test data includes 8 sessions between August 7th 2016 and August 20th 2016 with a maximum of 1 session per day.
Finally, please let me point out, that the data shown below should be taken and consumed with the necessary amount of care. Provider y doing bad in the test doesn’t mean that provider y is a bad service provider or does not work in China, it just means that in the given time period in Shenzhen with the given internet connection the test provided by CloudHarmony didn’t work for provider y at the 8 times the test was run. So don’t jump to any conclusions based on this test alone, seek as many sources and reference points as possible before making any decisions.
After all those remarks, let’s get to the results!
Best performing server locations
This analysis took the results of each of the 8 tests, sorted the servers by latency with the lowest on top, highest on bottom and awarded points, 10 for the first down to 1 for the 10th.
The top 5 performers for this analysis were:
- Linode Tokyo
- Linode Freemont
- DigitalOcean sfo2 (San Francisco)
- DigitalOcean ams3 (Amsterdam)
- Linode Frankfurt (equally fourth)
Similar to the CDN test and server test, what strikes me is that Singapore servers were not performing better in the tests, especially given that there were no Hong Kong servers available to be tested here. There were a couple of problems with sea cables around Vietnam recently, so that might have had an impact, but we’ll have to wait and see for future results of those tests to really get a picture of what is going on.
The clear winner of the test was the Linode Tokyo data center. This isn’t too surprising having seen other Japanese data centers doing very well in the other tests. The big problem here for us developers is, that this data center is currently sold out. Linode says it’s working on a solution for that, but for the moment, this option isn’t available.
With this option gone, The US with Linode’s Freemont data center and DigitalOcean’s San Francisco data center seems to be the best choice closely followed by European data centers.
With this test only targeting latency it’s difficult to say, what the bandwidth and loading speed would look like from those data centers but I would assume that they are just fine. In fact, China Website Audit is running on DigitalOcean’s San Francisco servers and the loading speed within Mainland China is very respectable.
The raw data
Here you can download the screenshot of the 8 test results that were taken as basis for the analysis. Feel free to download the ZIP file and build your own conclusions based on the test result from CloudHarmony.
All in all it can be said that both providers are viable alternatives for more price-sensitive customers who still want to have a decent performance in China. Linode seems to have a slight edge over DigitalOcean based on the test results presented in this article, but the most important thing to see is that most of the times, almost all of the data centers are performing very respectively in the tests from Shenzhen.
For bigger websites, Hong Kong would still be the better option for hosting in my opinion when targeting Asia and China specifically. For other websites, the US West coast also seems to be a great alternative especially when combined with a well performing CDN.
A quick introduction to CDNsun, one of our top performers in the most recent CDN test
A quick introduction to Akamai CDN, one of our top performers in the most recent CDN test
A quick introduction to SkyparkCDN, one of our top performers in the most recent CDN test