After many months of enduring inquiries about why I was no longer including
Amazon’s Lightsail VPS offering, I’m giving
up in and including it in the mix
Originally it was dropped because they don’t currently offer Ubuntu 18.04 LTS which I was doing my benchmarks against.
For me personally, I prefer to run the latest and greatest version of my distro of choice when I can, so it made sense to do these reviews on the latest and greatest Ubuntu LTS release.
So yeah, got tired of being asked why it was dropped, so I spent some time today
time today retrofitting my [server benchmarks script][bench] to allow for
sysbench 0.4.x, as found on Ubuntu 16.04 LTS, and 1.0.x, as found
on Ubuntu 18.04 LTS.
To get that apples to apples comparison as I prefer, all hosting providers were running Ubuntu 16.04 LTS and as per usual, I spun up 3 instances of each and averaged the results.
|Location||New York 1||Virginia, Zone A||Newark, NJ||New York (NJ)|
|OS||Ubuntu 16.04 LTS||Ubuntu 16.04 LTS||Ubuntu 16.04 LTS||Ubuntu 16.04 LTS|
|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|
Right out of the gate, Lightsail is offering up double the included bandwidth and nearly double the storage as everybody else.
|Cache Size (KB)||30720.00||30720.00||16384.00||16384.00|
Vultr’s processor upgrades earlier in the year are still holding strong.
|Events per Second||686.04||861.32||776.33||920.52|
Even though Lightsail was edged out by Vultr, Lightsail’s maximum execute time was leaps and bounds better than everybody. If you’re doing a ton of CPU heavy work, Lightsail should give you a more consistent experience without any random slow downs.
|Ops per Second||1832488.45||610450.95||1204406.60||2308807.56|
On the other hand, if you’re doing anything that’s read or write intensive on the memory, Lightsail wouldn’t be a great choice. Vultr was able to put up some great numbers with DigitalOcean and then Linode taking up the middle.
|Ops per Second||1855140.50||610214.85||1194220.77||2436094.29|
Pretty consistent with the Memory (Read) data, Vultr up front with Lightsail way behind.
Vultr and Lightsail practically even but Lightsail’s maximums clocked in way lower.
|Transactions per Second||251.08||250.25||336.84||296.80|
|Queries per Second||4770.60||4754.80||6399.92||5639.20|
This is the test that gave me the most amount of grief. Partly because it’s the
test that worked differently depending on which version of
sysbench works but
also because Lightsail doesn’t let you log in as
root over SSH, which threw a
wrench into some of my scripts.
All that said, Linode really shined on this one coming in ahead of Vultr and really far head of DigitalOcean and Lightsail. Even though DigitalOcean and Lightsail were fairly comparable, Lightsail really struggled with the maximum time.
Since Amazon doesn’t have a New York / New Jersey based data center, they got a bit of an edge in regard to distance. Even with said advantage, they clocked in as the slowest of the bunch.
Vultr touts their network speed in comparison to the rest, but couldn’t make good on that, at least in regard to their download speed. Uploads they did fine, but not crazy better or anything.
Apache Benchmark (against
nginx on the servers)
|Requests per Second||239.01||235.96||231.79||286.31|
|Time per Request (ms) (mean)||2111.80||2152.55||2161.79||1764.76|
|Transfer Rate (Kbytes/sec)||200.50||197.94||194.44||240.18|
The most subjective review of the bunch, since I’m running
ab from my home
machine and with a decently connected house and family, a lot of things could
factor into this value.
Taken with a grain of salt or two, Vultr was better than the batch while the rest were all clumped up in essentially a tie for second place.
The usual disclaimer here, YMMV and you really should take the time to analyze these benchmarks and make a decision based on your specific needs for your use case.
That said, feel like Vultr put up some really good numbers this month coming in the top 2 slots for most of the categories.
Linode also had a good showing, especially for MySQL, but fell short on the speed test and memory interaction.
DigitalOcean, while putting up some decent metrics, didn’t seem to stand out except for the speed test, but even still, their latency was the worst.
Lightsail for the most part, showed a lot more consistency between their minimum and maximum times, but really fell short in the memory category. With that, if you’re looking for more storage or bandwidth for your buck, they are the obvious call.
As always, if you found this post helpful, please with sugar on top, use one of my referral links below.
Amazon doesn’t have any referral incentives, so if you go with them and feel especially generous, you can buy me a beer over on GitHub Sponsors.
Got something special in store for next month’s comparison, stay tuned!