Before you go commenting on the obvious, let me go ahead and clear the air a
bit.
First off, yes, DigitalOcean and Linode are purposefully omitted from this
comparison because they lack of a $3.50 tier.
Second, yes, I’m aware there are other providers that offer $3.50 pricing tiers.
Most notable (thanks for all of the requests I receive) are OVH and RamNode.
So why did I choose to omit them?
RamNode didn’t make the cut this go around because they don’t offer Ubuntu 18.04 LTS
at this time. Since last month’s post where I dropped back to an older Ubuntu,
Amazon did start to offer the latest Ubuntu long-term support release and my own
desire to only review the latest and greatest, I made the call to skip RamNode
this month but will most likely go back to using Ubuntu 16.04 LTS to allow for
it’s inclusion.
OVH was a different story all together. I really wanted to get them in the mix,
but while poking around their site, I couldn’t find a way to create an account
w/o opting into some promotion that required a one year commitment. I spin these
boxes up for a few hours a month and being locked into a year term was no bueno.
Admittedly, I didn’t spent a ton of time trying to figure out their site, so I
very well may have missed a way to sign up / created an account without being
roped into some promotion.
With all that said, I think I’m getting close to expanding the hosts offered in
these reviews and appreciate everybody’s patience as I work out the kinks on my
end to make it happen.
Okay, so with all that out of the way, here’s the deal for this month. I’ve been
aware for a while now that there are sub-five buck hosting out there. Since
these posts did start as Linode vs DigitalOcean many moons ago, I’ve always
wanted to keep them in the mix and have always done apples to apples comparisons
on either price point or amount of RAM.
Both DigitalOcean and Linode bottom out at $5 and at 1 GB of RAM so there wasn’t
really a way to work them into this month’s review.
It’s also worth pointing out, that I believe that hosting as a whole,
especially of late, seems to be a race to the bottom. At one point Linode’s
stance was that they’d never offer a plan less than $20 and yet market pressure
has resulted in them offering two plans less than that.
It won’t shock me if in a year or so I’m including more providers in the sub-$5
price range.
All that being said, here’s how Amazon Lightsail and Vultr stacked up against
each other at the $3.50 price point.
Overview
| Lightsail | Vultr | |
|---|---|---|
| Location | Virginia, Zone A | New York (NJ) |
| RAM | 512 MB | 512 MB |
| CPU | 1 Core | 1 Core |
| SSD | 20 GB | 20 GB |
| Transfer | 1 TB | 0.5 TB |
Pretty comparable product offering outside of Vultr only offering up half of the
bandwidth as Lightsail.
It’s worth noting that Vultr also offers a $2.50 plan with the same specs, but
only available with IPv6. I’ve also heard many times that instances at that
price tend to be unavailable. YMMV but I did just spin one up without issue!
CPU Info
| Lightsail | Vultr | |
|---|---|---|
| CPU MHz | 2400.10 | 2398.15 |
| Cache Size (KB) | 30720.00 | 16384.00 |
| BogoMips | 3200.33 | 3196.00 |
Neck and neck on this one.
CPU
| Lightsail | Vultr | |
|---|---|---|
| Events per Second | 888.89 | 742.34 |
| Minimum (ms) | 1.03 | 1.19 |
| Average (ms) | 1.13 | 1.35 |
| Maximum (ms) | 1.59 | 9.67 |
Even though the CPUs were stacked pretty evenly, Lightsail performed
significantly better. Not only did they process more events per second, the
maximum time to execute was quite low.
Memory (Read)
| Lightsail | Vultr | |
|---|---|---|
| Ops per Second | 845991.70 | 3108473.66 |
| Minimum (ms) | 0.00 | 0.00 |
| Average (ms) | 0.00 | 0.00 |
| Maximum (ms) | 2.93 | 10.52 |
Lightsail server’s CPU may perform better than Vultr’s but for reading from
member, there’s absolutely no comparison. Max execution times came in higher
with Vultr but the number of operations per second was over triple that of
Lightsail!
Memory (Write)
| Lightsail | Vultr | |
|---|---|---|
| Ops per Second | 852791.94 | 3095217.65 |
| Minimum (ms) | 0.00 | 0.00 |
| Average (ms) | 0.00 | 0.00 |
| Maximum (ms) | 2.83 | 5.47 |
In past comparisons, the memory read and write metrics tend to be in line with
each other. Pretty much the norm here, with Vultr trouncing Lightsail.
File I/O
| Lightsail | Vultr | |
|---|---|---|
| Reads per Second | 1178.61 | 1432.22 |
| Writes per Second | 785.73 | 954.78 |
| Fsyncs per Second | 2507.92 | 3049.65 |
| Minimum (ms) | 0.00 | 0.00 |
| Average (ms) | 0.22 | 0.20 |
| Maximum (ms) | 15.99 | 12.98 |
Vultr edged out Lightsail on file I/O but with not nearly the gap we saw with
the memory benchmarks.
MySQL
| Lightsail | Vultr | |
|---|---|---|
| Transactions per Second | 1994.33 | 1732.67 |
| Queries per Second | 39886.67 | 34653.33 |
| Minimum (ms) | 2.36 | 2.93 |
| Average (ms) | 5.01 | 6.54 |
| Maximum (ms) | 40.39 | 39.90 |
Even with slower file I/O and memory reads/writes, Lightsail’s MySQL performance
was a bit better than that of Vultr.
Speed Test
| Lightsail | Vultr | |
|---|---|---|
| Distance (km) | 2090.30 | 2124.90 |
| Latency (ms) | 45.065 | 786.966 |
| Download (Mbit/s) | 282.09 | 212.41 |
| Upload (Mbit/s) | 173.08 | 357.60 |
Lightsail tends to have the advantage on this one since Amazon’s Virginia
data center is a bit closer to the Texas server I benchmark against than
everybody else’s New York / Newark data centers.
Even with the slight advantage, Lightsail was only able to outperform on
downloads while Vultr took it for uploads.
Vultr’s latency this month leaves a lot to be desired.
Apache Benchmark (against nginx on the servers)
| Lightsail | Vultr | |
|---|---|---|
| Requests per Second | 294.45 | 289.59 |
| Time per Request (ms) (mean) | 1698.51 | 1726.95 |
| Transfer Rate (Kbytes/sec) | 247.00 | 242.93 |
I always catch flak for this one because the benchmark is against an out of the
box nginx serving up a static HTML file. Even still, both providers performed
about the same.
Conclusion
Even with half the bandwidth that Amazon is offering with Lightsail, Vultr seems
to be the better all around pick at the $3.50 price point.
As always, YMMV and you should pick a hosting provider based around your
personal needs.
If you found this review helpful in making your decision on a $3.50 VPS, pretty
please use one of my referral links below.
If you want to run these benchmarks yourself, feel free to pull down my
[benchmarks][bench] script.
Got something fun planned for next month, then hopefully in January expanding
these reviews to include RamNode and OVH!
