A ton has changed since this post was originally published! Why not check out a more recent post?
New year, new set of benchmarks. Decided to start the year off with a comparison at the 5 dollar tier.
Best as I can tell, this price point tends to be a favorite so I’m probably going to stick with it this year. Either that or every month moving up to a new tier, but I’m not sure that each provider has 12 tiers so that may not work.
|Location||New York 1||Virginia, Zone A||Newark, NJ||New York (NJ)|
|RAM||1 GB||1 GB||1 GB||1 GB|
|CPU||1 Core||1 Core||1 Core||1 Core|
|SSD||25 GB||40 GB||25 GB||25 GB|
|Transfer||1 TB||2 TB||1 TB||1 TB|
The tale of the pricing pages is pretty evenly matched with the exception of Lightsail offering 15 GB of additional storage and double the bandwidth.
|Cache Size (KB)||27136.00||30720.00||16384.00||16384.00|
Bit of a swing here between providers. Worth noting is that these numbers are the average of 3 separate servers on each provider. Based on the peculiar looking numbers for DigitalOcean and Lightsail, it seems like each instance came in with a different clock speed.
When you’re picking a provider, any provider, you should always spin up a few boxes and pick the one with the most favorable specs. Often times they may all be exactly the same while other times you may end up with server sitting on legacy hardware and crappier specs.
|Events per Second||965.74||859.84||680.53||781.90|
Even with having the slowest average clock speed, DigitalOcean’s actual processing power was speedier.
Both Linode and Vultr brought up the rear, those maximum times most likely to blame. Even with decent average speeds, when there is a high maximum you can experience lag. Often times this is due to noisy neighbors that are choking out the CPU.
|Ops per Second||3872902.31||812003.21||652546.66||3388041.42|
DigitalOcean and Vultr both broke 3 million operations per second while Lightsail and Linode were below 1 million.
Even putting up some decent numbers, Vultr’s maximum time to process came in third.
|Ops per Second||3861697.14||824843.47||629582.55||3359018.29|
Generally speaking, the memory write metrics tend to marry up pretty close with the memory read one. This particular benchmark is no exception.
Lightsail’s maximum did come in well below what it did during memory reads while Linode’s maximum came in well above.
|Reads per Second||1246.86||1112.42||1094.47||2159.02|
|Writes per Second||831.22||741.61||729.61||1439.33|
|Fsyncs per Second||2652.49||2366.50||2326.54||4601.03|
Vultr really shined for file I/O with the rest of the bunch fairly evenly matched.
|Transactions per Second||670.00||2146.67||1556.00||3055.67|
|Queries per Second||13400.00||42933.33||31120.00||61113.33|
Vultr continued the trend into the MySQL benchmark coming in nearly 1,000 transactions ahead of the second place Lightsail.
What’s somewhat surprising is that DigitalOcean’s transactions per second came in so low with a maximum time at 6 and a half seconds.
Poor MySQL benchmarks aside, DigitalOcean crushed everybody on download speed and inched out ahead of Vultr for uploads.
Considering Vultr still touts themselves as having the fastest network, their numbers weren’t all that impressive.
Apache Benchmark (against nginx on the servers)
|Requests per Second||258.93||269.21||253.82||263.67|
|Time per Request (ms) (mean)||1931.14||1857.98||1980.04||1899.95|
|Transfer Rate (Kbytes/sec)||217.21||225.83||212.93||221.19|
This is always the “take it with a grain of salt” metric. It’s just me running
ab locally to an out of the box nginx server serving up a static page. Too
many factors in play to get a clean reading.
Nonetheless, everybody came in pretty evenly matched, which usually isn’t the case.
As always, please take these metrics and jump to your own conclusions based on your needs.
For me, I think DigitalOcean put out some of the better results this go around, with the exception of how MySQL performed.
Obviously, for most specs for the buck, Lightsail is going to provide you the most storage and bandwidth.
There’s still a lot to love about Linode and Vultr and other offerings like block storage and ease of use of the user experience don’t factor into these raw metrics but should be considered when making your decision.
If you found this comparison helpful and are planning to sign up for one of the providers mentioned, it would be greatly appreciated if you used one of my sign up links for DigitalOcean, Linode and/or Vultr.