Linode vs. DigitalOcean – performance benchmarks

From DigitalOcean's TOS: "You agree that you will NOT use DigitalOcean's services to: violate any applicable state and federal law and regulation, including, but not limited to, any copyright, trademark, patent, anti-piracy, or other intellectual property law or regulation, or encourage or enable others to violate any such law or regulation. Transmit, distribute, post, store, link, or otherwise traffic in information, software, or materials that is offensive, abusive, inappropriate, malicious, or detrimental, including, but not limited to, those that: Are obscene, fraudulent, or discriminatory, including any containing profanity, or obscenities. DigitalOcean permits adult websites that abide by state and federal law and regulation. ..."<p>I think it's funny how they can go from barring any information "offensive, abusive, inappropriate, malicious, or detrimental, including, but not limited to, those that: Are obscene, fraudulent, or discriminatory, including any containing profanity, or obscenities" to specifically permitting "adult websites that abide by state and federal law and regulation."<p>In any case good, luck keeping your blog, blog, comments, or whatever else you host with them clear of all profanity all of you who decide to try their services, because you've just walked into a convenient TOS violation. Damn! (Oops. TOS violation. Goodbye HN. Ha ha.)

For me the performance of Linode isn't the greatest factor, the greatest factor (and the reason I spend ~$300/m with them when I could get the 10 servers for 1/5th the price elsewhere) is that support is second to none. I have never experienced better support than I have with Linode. If I have a problem and submit a ticket (which is rare) it's answered within a few minutes, if there's a network or hardware issue they open a ticket with me before I've even noticed. No other provider I've used has ever had that quality of service.

I just switched <a href="http://jsonip.com" rel="nofollow">http://jsonip.com</a> from Rackspace/Slicehost to Linode about a month ago. I was a multi-year Slicehost customer and couldn't have been happier. It was only a matter of time after Rackspace aquired them that I was expecting problems. That occurred about a month + one week ago. (Sorry for the downtime to any users that might be reading this.)<p>I've since migrated the service to Linode and holy shit, it's such an improvement. The support is great, but even more their admin interface is excellent. Gives me so much more info than I ever had before. It's even a little cheaper than the plan I had with Slicehost.<p>Just to throw some numbers out there: jsonip.com is a node-based app that supports upwards of 6-10 million requests per day. Lets just say that I couldn't be happier with Linode.<p>Note to Linode: Do not ever sell out to Rackspace or else I will be forced to go live in the woods somewhere. Thanks.

(shameless self-promo ;-)<p>Use the DigitalOcean API from Ruby and Chef/knife to create/shutdown/destroy virtual servers ("droplets"):<p>* <a href="https://github.com/rmoriz/digital_ocean" rel="nofollow">https://github.com/rmoriz/digital_ocean</a><p>* <a href="https://github.com/rmoriz/knife-digital_ocean" rel="nofollow">https://github.com/rmoriz/knife-digital_ocean</a>

A new broom sweeps clean. Results even with VPS depend on who else is using the equipment and what they are doing.<p>Easy to spin up a machine with no usage and get great performance if there is nobody on it. That doesn't mean this data <i>isn't</i> correct of course but keep in mind the OP is making a comparison between 1 machine at Linode and 1 machine at Digital Ocean at a given point in time. Same equipment or different equipment at a later date could yield different results as the equipment gets filled up or depending on who else is using it. We had VPS's at Media Temple that would literally stall periodically for a few seconds. We then migrated to a different machine at a different data center and it hasn't happened since.

I've been using digital ocean for a minecraft server and vpn. Network reliability has been a problem. Some days I get frequent disconnects, and other days it is fine.<p>I will continue using linode for my servers that matter, and digital ocean for toys and experiments, just because it's so cheap.

One of the many reasons I pay for Linode is the "good neighbor" aspect the price point brings.<p>I'm not saying that DigitalOcean suffers this problem but the cheaper a VPS provider is the greater chance of it being used by folks running bit torrent, warez downloads, etc and/or folks who think they need to smash the (shared) resources out of server to get their moneys worth.<p>Paying a few dollars extra for a Linode instance (and I have 100's of them, so it adds up) is worth the peace of mind of being unattractive to these kinds. IE 'bad neighbors" who will pee in the pool.<p>Support is also top notch (although I rarely need it).

I fail to see how anyone can create a repeatable objective benchmark between different VPS providers.<p>I respect the fact that the author tried hard and didn't bias the results, but I fail to see where he even repeated the benchmark over a number of days to attempt to eliminate variations in what the other guests were doing while the benchmark was running.<p>As I see it, the only way I would trust a benchmark of this nature was if it was somehow ran over multiple hosts and was able to sample a large fraction of the physical boxes over a large period of time. As you can image, that's impossible to do on S3 and any of the other larger VPS providers<p>Even if somehow someone did manage to pull off a benchmark like the above, I still wouldn't trust it because it would only be valid until the provider added "Joe's Cheap Video Encoding Service" as a client that's sharing my box. Or someone else who's running a benchmark just like you are?<p>With all of the above said: I'm sure that there is differences between the VPS providers in how they allocate CPU and IO resources and Digital Ocean may indeed be much better on average. I'm just not sure how you can measure it objectively to make sure they are or that they will continue to be so even 5 minutes in the future.<p>Which is a shame, because an objective metric that was trustworthy that compared all the different providers would be absolutely awesome.

There are plenty of VPS providers out there who are cheaper than Linode, including some with compelling SSD packages.<p>With Linode, I know that I am getting excellent support. There's the peace of mind of knowing that when I submit a ticket, someone will read and respond in short order.<p>A $5 SSD VPS sounds great, but I automatically wonder what kind of hardware I will be on (does it compete with Linode's RAID 10 configuration?), and how long my support requests will take to be addressed, and whether or not the low prices are a sustainable business model.

My favorite site for finding all kinds of VPS deals (some for as little as $15 a year) is <a href="http://www.lowendbox.com" rel="nofollow">http://www.lowendbox.com</a>. It's great for finding a VPS nearly anywhere in the world for your SSH tunnel, offsite storage, or just to tinker with.

DigitalOcean is promising, but they should probably hire a lawyer to fix up their Terms of Service. I notice they've attempted to appease adult content websites by removing the restriction against nudity and adding a statement that adult sites are expressly allowed if legal. But, they've left in a prohibition on profanity? That seems a bit backwards.<p>Inappropriate content is also disallowed, what does that mean? It's not a legal term of art that I'm familiar with. Why does a sentence begin with "Transmit"? And please, consider using bulleted lists.<p><a href="https://www.digitalocean.com/tos" rel="nofollow">https://www.digitalocean.com/tos</a>

Since their US servers are based in New York, what are the state's policies as far as sales taxes? I believe there was a issue with Linode's Texas hub that any sales to Texas residents required you collect sales tax. Does NY have a similar issue where a server constitutes a physical presence?

The article says Linode has "predictably greater CPU performance". Am I missing something? It looks like Digital Ocean is the winner in every single benchmark other than Apache Static Page Serving.<p>(I'm assuming lower is better for encoding time.)

For a while, I was pretty happy with another low-price host (ChicagoVPS)--but I had to switch away (to Linode, as it happens) due to something not mentioned or measured in this article at all: network throughput volatility. Someone or something was regularly burst-saturating the physical pipe going into the machine, such that it would seem to all the rest of the world that my website would be down for random 90-second periods every few hours. This is really bad when you're trying to run a realtime game with persistent connections :)<p>I'm not sure quite what it was--whether malicious user behavior, a lack of separation in bridge-interface setup making network traffic on the box un-QoS-able, or maybe someone on the box I was on bearing the brunt of temporary, persistent DoS attacks--but I didn't care to roll the dice on another box from the same host, when I knew that this kind of thing "just doesn't happen" with the major providers.<p>Still, if there was a low-end host that specifically mentioned stable network throughput in their SLA, I'd love to try another one. $7/mo sure does buy a lot of horizontal scaling and High Availability, once you can ensure each node can hold onto the little bit of traffic you send it. :)

I recently switched from Linode to a dedicated server from Hetzner as I found I could get better specs for only a few quid more / month (than Linodes 1GB package).<p>Linode was a great provider, but now I can manage my own VMs using Xen myself.

I rarely see Rackspace Cloud mentioned on Hacker News in these sorts of VPS discussions. Anyone have insight as to why that is? Too expensive? Crappy service?<p>Background - my employer hosts on dedicated servers at Rackspace and we're considering using more of their cloud offering.

ServerBear has all the $5/month hosts in one place: <a href="http://serverbear.com/compare?Sort=BearScore&#38;Order=desc&#38;Monthly+Cost=500000-500000&#38;HDD=-&#38;RAM=-&#38;BearScore=-&#38;I%2FO+Benchmark=-&#38;Virtualization=" rel="nofollow">http://serverbear.com/compare?Sort=BearScore&#38;Order=desc&...</a>

I made the switch from Linode/AWS micro instances to Digital Ocean recently.<p>Pros: Great pricem, performance is excellent,UI is great for creating servers<p>Cons: I have had two periods of about 20 minutes of down time in the early mornings recently.<p>Digital Ocean is great hosting for developers. They are really doing it right.

I was choosing VPS providers recently. I used <a href="http://www.serverbear.com" rel="nofollow">http://www.serverbear.com</a> for the benchmarks. Then I looked on the web for anecdotal information on how good and reliable the VPSs were. I chose Linode. Then I needed another VPS. It was favoring Ramnode, but they had temporarily run out of SSDs. So I went with Rackspace, which was used by many people, even though they seemed to generate more complaints then Linode.

Shiny new services will always be popping up with great deals and "faster" servers but I prefer to not move my infrastructure every time the next host du jour comes out. Plus, they're just going to raise their prices once they get a critical mass of customers.<p>Linode has been there through thick and thin for me. Sounds cheesy, but they report on and fix issues before I even <i>notice</i> them, I've only had to submit a ticket once (for something that was my fault), and their infrastructure is tight as a drum.<p>Linode is very, very reliable. Another thing that many people who work with larger applications than mine will tell you is that speed is not important, what's important is consistency. Linode's virtualization is set up to provide, above all, consistency (even on a loaded host). So sure your site might run 1.6778 times faster on DO, but when the host server is loaded your site's going to be crawling because the other neighbors are bursting their CPU into your space.<p>We'll see if DO's prices and performance stand the test of time. My guess is their support is extremely lacking in comparison. When everything's running perfectly, this is great. But when the chips are down can you really rely on them? I guess we'll see.

I've been very happy with DigitalOcean since switching to them. This blog echo's my anecdotal experience. If DigitalOcean would open up a Chicago and West coast DC I could switch completely over...<p>I wish this had been a $20 vs $20 plan comparison.

Doh. "Due to a high load that we are experiencing with Trial Accounts - we have temporarily disabled them." However email after signup tells: Your trial period will last 12 hours.

I personally use digitalocean for hosting <a href="http://en.wikipins.org/" rel="nofollow">http://en.wikipins.org/</a> and would like to say that I am <i>very</i> happy with their service and SSD IOPS - something that took me 6hrs. on EC2 takes me ~45mins on digitalocean hosted machines for a half (or less) the cost.

I think DO compares to Linode (or even better) in CPU performance because the software used (for Encoding or whatever) doesn't make use of extra cores. You'll need multi-threads aware software to take advantage of the available cores.<p>I have a Linode 1Gb (and another 512Mb) and I rarely go over 110% CPU utilization even when my server is busy (the max is 400%, 4 cores being used).<p>Also, price comparison, Linode charges an extra $5 (for 512Mb) and $10 (for 1 Gb) for Backups. DO seems to be giving them for free.<p>My yearly bill on Linode : $900<p>My yearly bill on DO : $180<p>I think I'm trying DO soon, especially I'll be in need for more servers.<p>Edit: I'm a little afraid from DO. They are simply too^8 cheap and it's scary. Will they be around for too long? What's their down time like? Will they lose my data (and backups)?...<p>Price wise, DO beats the hell out of anything. I can renew my smartphone yearly for free just by moving.

Hmm. They email your initial root password in cleartext, rather than either letting me choose it or displaying it on the site, both of which would be over SSL.

DigitalOcean is good value but I found their I/O speed not anywhere close to SSD levels. I ran this bench mark from github. RamNode is the best.<p><a href="https://github.com/mgutz/vpsbench" rel="nofollow">https://github.com/mgutz/vpsbench</a>

Just hope those SSD are made by intel and are in raid10.<p>Otherwise 3-4 year failure is quite common.

I've never heard of Digital Ocean. Are they reliable? Do they have a DNS manager comparable to Linode? (I have no aspirations of running multiple Bind instances on my own, thanks).<p>I might consider moving off Linode for pricing alone.

DigitalOcean does not currently offer IPv6.<p><a href="http://digitalocean.uservoice.com/forums/136585-digital-ocean/suggestions/2639897-ipv6-addresses" rel="nofollow">http://digitalocean.uservoice.com/forums/136585-digital-ocea...</a>

Off-topic:<p>Who is this "Terrence Andrew Davis" fellow who comments on pretty much every single post that reaches HN?<p>He sounds like a lunatic and religious fanatic constantly quoting bible verses and sermons. Pretty darn creepy to read at 4am.

I've been a Linode customer for many years. I couldn't be happier. They have a buttload of locations to choose from, fast networks, and as everyone else has already mentioned: out of this world support. I've even managed to earn a few hundred bucks in referral cash from them.<p>I had a handful of dyno's, a worker, and a few add-ons running my iPhone app's API within Heroku up until about a week ago where I transitioned everything onto a small collection of Linodes in the UK datacenter. We're launching in Europe so it made sense to bring everything closer. If you're comfortable and skilled with a terminal, you can get a lot more bang for your buck with Linode.<p>That being said, I am always interested in trying something new. For a $40 Linode you get 1GB of RAM, with DigitalOcean the same $40 gets you 2GB of RAM and they run on SSD's. I'm going to deploy a copy of my primary API server here to take it for a spin.<p>First impression thus far: fast deployment. From pressing the submit button to shelling in took literally less than a minute.<p>Unfortunately though the latency on the box seems a bit high. There is a bit of lag in the terminal, something I do not experience with Linode's UK datacenter.<p>I'm in Sweden on a 4G modem, here's a traceroute between me and both boxes. DigitalOcean is first, followed by Linode.<p><pre><code> ### DigitalOcean – Amsterdam Datacenter traceroute to 192.81.221.48 (192.81.221.48), 64 hops max, 52 byte packets 1 tiny (192.168.0.1) 10.573 ms 1.658 ms 1.393 ms 2 10.9.14.115 (10.9.14.115) 33.218 ms 24.250 ms 59.223 ms 3 10.9.14.113 (10.9.14.113) 51.009 ms 35.878 ms 30.836 ms 4 90-229-25-24.link.se.telia.net (90.229.25.24) 33.153 ms 30.662 ms 34.999 ms 5 s-b2-link.telia.net (80.91.247.96) 33.962 ms s-b2-link.telia.net (80.91.246.228) 74.544 ms s-b2-link.telia.net (213.155.133.147) 77.395 ms 6 s-bb1-link.telia.net (80.91.246.148) 36.792 ms s-bb2-link.telia.net (80.91.246.234) 29.966 ms s-bb2-link.telia.net (213.155.133.142) 31.228 ms 7 kbn-bb2-link.telia.net (213.155.134.141) 151.952 ms kbn-bb2-link.telia.net (213.248.65.29) 42.281 ms kbn-bb2-link.telia.net (80.91.246.178) 123.130 ms 8 kbn-b3-link.telia.net (80.91.249.51) 39.032 ms 45.868 ms kbn-b3-link.telia.net (80.91.245.159) 44.708 ms 9 te4-7.ccr01.cph01.atlas.cogentco.com (130.117.14.33) 67.971 ms 60.729 ms 96.055 ms 10 te0-7-0-0.ccr22.ham01.atlas.cogentco.com (154.54.62.117) 132.391 ms te0-7-0-6.ccr21.ham01.atlas.cogentco.com (130.117.2.149) 97.568 ms te0-7-0-0.ccr22.ham01.atlas.cogentco.com (154.54.62.117) 137.713 ms 11 te0-0-0-3.mpd21.ams03.atlas.cogentco.com (130.117.50.57) 101.141 ms 89.885 ms te0-0-0-3.mpd22.ams03.atlas.cogentco.com (130.117.1.101) 147.596 ms 12 te2-1.mag01.ams03.atlas.cogentco.com (130.117.48.62) 104.005 ms te1-1.mag01.ams03.atlas.cogentco.com (130.117.0.242) 52.704 ms 127.224 ms 13 nxs-internet.demarc.cogentco.com (149.6.128.6) 98.294 ms 130.686 ms 142.924 ms 14 192.81.221.48 (192.81.221.48) 148.148 ms 107.753 ms 134.880 ms </code></pre> and now Linode...<p><pre><code> ### Linode – London, UK Datacenter traceroute to blitzen.dbld8.com (176.58.120.7), 64 hops max, 52 byte packets 1 tiny (192.168.0.1) 9.671 ms 2.313 ms 1.411 ms 2 10.9.14.115 (10.9.14.115) 32.881 ms 21.701 ms 34.152 ms 3 10.9.14.113 (10.9.14.113) 36.843 ms 21.986 ms 26.750 ms 4 90-229-25-24.link.se.telia.net (90.229.25.24) 26.421 ms 24.166 ms 26.982 ms 5 s-b2-link.telia.net (213.155.134.55) 27.415 ms s-b2-link.telia.net (80.91.248.199) 33.347 ms 37.758 ms 6 s-bb2-link.telia.net (80.91.254.114) 34.993 ms s-bb1-link.telia.net (80.91.254.58) 92.886 ms s-bb1-link.telia.net (80.91.247.216) 38.297 ms 7 hbg-bb1-link.telia.net (213.155.135.236) 83.932 ms hbg-bb1-link.telia.net (80.91.251.41) 36.076 ms hbg-bb2-link.telia.net (80.91.247.147) 47.708 ms 8 ldn-bb2-link.telia.net (80.91.247.201) 67.937 ms ldn-bb2-link.telia.net (213.155.132.247) 106.801 ms ldn-bb2-link.telia.net (80.91.247.169) 63.880 ms 9 ldn-b3-link.telia.net (80.91.249.170) 54.534 ms ldn-b3-link.telia.net (80.91.251.237) 101.259 ms ldn-b3-link.telia.net (80.91.247.25) 64.316 ms 10 telecity-ic-150799-ldn-b3.c.telia.net (80.239.167.94) 59.564 ms 70.329 ms 64.835 ms 11 85.90.238.70 (85.90.238.70) 70.606 ms 68.463 ms 60.808 ms 12 212.111.33.234 (212.111.33.234) 69.237 ms 60.927 ms 70.726 ms 13 li522-7.members.linode.com (176.58.120.7) 65.147 ms 60.434 ms 62.646 ms </code></pre> Quite a difference here. Geographically Sweden is closer to Amsterdam, but i'd imagine there is a fat pipe crossing the North Sea that is faster than being routed through Northern Europe to AMS.

One of the key featuers of Linode (at least it was when I used them) is the control panel which allows you to manage various vdisks (and even various whole OS installs on those vdisks if you so desire, though not with several running on a single account) within the space limits of your account. That and the general support. At the time I used Linode very few other hosts at that price-for-stuff value point offered such control. Neither of these things are mentioned in the comparison so I don't know if DO offers similar - obviously it isn't important to this reviewer but that sort of thing is for some.<p>I currently use one of the cheap servers from <a href="http://www.kimsufi.co.uk/" rel="nofollow">http://www.kimsufi.co.uk/</a> (9UKP/mo for 2Gb RAM, 500Gb space and 5Tb bandwidth (after 5Tb you are not cut off but bandwdth is throttled, though I've never come close to that to see how much effect it has)). Excellent if you don't need any hand holding (it is a completely self managed service) or much support (I've not needed to contact support so can't comment, but again "completely self manged"), or a very specific OS/kernel (they have their own builds, getting stock versions intalled is possible but a bit of a faf), you want a pile of space, and your tasks are not CPU intensive (the cheaper machines only have Atom or Celeron processors). Having your own dedicated drive makes I/O contention a moot point unlike with a VPS (though depending on how loaded DigitalOcean's host servers are they may still outperform this for some I/O patterns due to the use of SSDs).

Where are Digital Ocean's data centers? I can't seem to find that information anywhere on their site.

In this article, DigitalOcean's virtualization technology is misstated.<p>Its KVM not VMware

For over VPS benchmarks, I love serverbear.com<p>I host with Ramnode.com, and I love their prices and SSDs. Not one issue yet except the odd DDoS attack. Used the server bear coupon code to get 31% off my services for life.<p>I wonder how it stacks up to DigitalOcean. I also have a linode server, but I'm thinking of cancelling because it is too expensive, but it is solid as a rock.

Not a bad deal for Digital Ocean especially when there are $20 off promo codes available. Assuming you can use it for 4 months, it's essentially a 4 month trial. I found a few codes with a Google search. ;)

It's still just 512 megs of RAM. I'd love to find a VPS provider that gives me like 2 gigs of RAM, and charged me by how much of that was hot during the month. It's frustrating to have to pay in advance for the maximum amount of RAM I might need.<p>I'd like to be able to leave processes running that might have a lot of data mapped into their address spaces, but for much of the time that data can safely be paged out to disk and not used. When I come and use the machine, I'd like to have a large amount of RAM, but when I'm not using it, I don't want to pay for it.

I've been with Linode for a few years, nothing to complain about. Like others have mentioned, super fast support, good wiki pages for setting things up, and good IRC channel.<p>With RAM prices being what they are, Linode should price their RAM more competitively and offer SSD for those who want it, at a reasonable price.<p>On the other hand they also have a data centre in Tokyo, which means noticeably better pings to Asia than DigitalOcean.<p>In any case, will give DigitalOcean a try for a couple of projects.<p>Might also want to try OVH and Hetzner for dedicated servers for about £50/month.

Very confused, but I understand brand loyalty. Linode scorched Slicehost at one point. I like the idea of DigitalOcean and I'm moving all my non critical/beta/alpaha sites to DigitalOcean.

There is routine 10x variation in disk performance between different VPS's on Linode. It's luck of the draw and performance will change (i.e. get worse) over time.<p>You'd need to provision and benchmark a bunch of Linodes to get meaningful numbers.<p>That's one of a dozen ways in which a VPS is not like a real server.<p>5 second benchmark which corresponds well to real-life disk performance:<p><pre><code> sysbench --test=fileio --file-fsync-freq=1 --file-num=1 \ --file-total-size=16384 --file-test-mode=rndwr run --max-time=10 \ | grep "Requests/sec"</code></pre>

We just switched over to Digital Ocean and I'm very happy with it!

There might be better performing hosting providers, and there are definitely lower cost providers than Linode, but I don't love them like I love Linode.<p>It's not entirely rational, but if you ask around, I'd wager most people who use Linode love it.<p>Maybe it's because it used to be just caker doing everything, or the free disk / memory upgrades he kept handing out. I remember him asking about LASIK surgery in the (very helpful) forums; there's a family feeling to it. It's the anti-Rackspace. It's home.

I just launched this project on a Digital Ocean $5 VPS ... it seems pretty fast. (nginx + passenger + Rails)<p><a href="http://romancepass.com" rel="nofollow">http://romancepass.com</a><p>I've been a happy Linode customer as well. But being able to deploy silly side projects like this for just $5/mo is really cool. This is my 3rd DO VPS and they've all been great. I had a PostgreSQL problem on another one and my support requests were replied to really fast.

Anyone know of any similar hosts (VPS, SSD) for Windows Server? Seems like installing a virtual Win instance may not work [ <a href="http://digitalocean.uservoice.com/forums/136585-digital-ocean/suggestions/2536799-can-i-deploy-windows-" rel="nofollow">http://digitalocean.uservoice.com/forums/136585-digital-ocea...</a> ]<p>Alternatively, if anyone has successfully installed Windows on a VPS instance, please let me know :)

DigitalOcean seems like they are worth a look

I have both Linode 512 and DigitalOcean 512 VPS. Both VPS are running the same version of Debian, same version of Apache (Apache/2.2.16 (Debian), same apache2.conf files, list of loaded modules are same. For benchmark is used same web site on localhost. (simple joomla site). Results are quite different and I can't figure why. Linode has much more "Requests per second".<p><pre><code> #linode London Server Software: Apache/2.2.16 Server Hostname: localhost Server Port: 80 Document Path: / Document Length: 7483 bytes Concurrency Level: 20 Time taken for tests: 33.140 seconds Complete requests: 1000 Failed requests: 0 Write errors: 0 Total transferred: 8037000 bytes HTML transferred: 7483000 bytes Requests per second: 30.18 [#/sec] (mean) Time per request: 662.791 [ms] (mean) Time per request: 33.140 [ms] (mean, across all concurrent requests) Transfer rate: 236.84 [Kbytes/sec] received Connection Times (ms) min mean[+/-sd] median max Connect: 0 0 2.1 0 16 Processing: 118 658 1143.2 487 19006 Waiting: 113 655 1138.1 485 19006 Total: 118 659 1143.2 488 19006 Percentage of the requests served within a certain time (ms) 50% 488 66% 563 75% 616 80% 658 90% 779 95% 932 98% 3635 99% 6406 100% 19006 (longest request) #DigitalOcean Amsterdam Server Software: Apache/2.2.16 Server Hostname: localhost Server Port: 80 Document Path: / Document Length: 7483 bytes Concurrency Level: 20 Time taken for tests: 118.087 seconds Complete requests: 1000 Failed requests: 0 Write errors: 0 Total transferred: 8032000 bytes HTML transferred: 7483000 bytes Requests per second: 8.47 [#/sec] (mean) Time per request: 2361.734 [ms] (mean) Time per request: 118.087 [ms] (mean, across all concurrent requests) Transfer rate: 66.42 [Kbytes/sec] received Connection Times (ms) min mean[+/-sd] median max Connect: 0 1 4.0 0 33 Processing: 1256 2355 234.1 2284 3604 Waiting: 1256 2328 232.5 2264 3604 Total: 1256 2356 234.1 2284 3604 Percentage of the requests served within a certain time (ms) 50% 2284 66% 2328 75% 2412 80% 2500 90% 2666 95% 2768 98% 2877 99% 3204 100% 3604 (longest request)</code></pre>

Maybe this is the reason, (from DigitalOcean): Backup User is solely responsible for the preservation of User's Data. Even with respect to Data as to which User contracts for backup services, DigitalOcean shall have no responsibility to preserve Data, the service is provided as is, without warranty.

Doesn't something like Rails appservers (Unicorns for an example) run better on Linode because of more access to cores?<p>Or does that depend on implementation? I remember reading something about MRI only being able to utilize one core.

How is it possible that amazon charge $0.050 per GB (for more than 350 TB / month ) while DigitalOcean charge $0.02 per GB for each extra GB and 0.005$ / GB for the 5$ plan? How can they make any profit?

Has nobody else noticed that the benchmarks use ext3 for Linode and ext4 for DigitalOcean? Not that the SSD's won't outperform hard drives anyway, but that's one extra variable that will skew results.

Linode customer for years. I'll buy 4 servers and give digital ocean a shot just because it seems the community approves the service and they accept paypal (Linode doesn't and it bothers me)

How does KVM compare with Xen? Can you over-subscribe ram with KVM?

How are the "requests per second" measured? (I assume that something like `ab` is used, but I'm wondering if it run on the same machine or from an external machine ...)

When Digital Ocean has a west coast datacenter, I will give them another look. Latency has a larger perceived impact to users than any of those benchmarks given.

The tests on disk performance are kind of redundant as ext3 is being compared to ext4. This will skew the results in favour of DigitalOcean somewhat.

As someone living in a land far beyond the mountains and the sea (aka Switzerland) I'm thinking this: 5$ for a nice Hulu / Netflix proxy? Sweet!

But what about features set comparison? ipv6 support, backups, node balancer, dashboard, API, community, documentation...?

I signed up for DigitalOcean, thanks.

Performance Benchmarks for what.. Are these Benchmarks for a Remote Server for Work.. Or is it as a Webserver..<p>Benchmark does not prove thing for Webserver other than Disk I/O is greater with Digital ocean. If Disk I/O was so critical then no one would go with AWS since they have got the worst of all.<p>Benchmarks need to prove a thing in its application, unless it does its waste of time.

Well looks like I'm just spunking away ~£12 every month!

These graphs are meaningless without methodology.

It amazes me that people on here are so quick to forget about the Linode Bitcoin incident.<p>Because it is during security incidents that you really find out what type of company you are dealing with. With Cloudflare they informed users quickly, were transparent about what happened and took their licks. With Linode they didn't inform users and we still don't know what happened.<p>As a former customer I would never, ever use Linode again.

i think its bullshit, just buy 10 vps and write garbage to disk, in a month you can wipe their entire stock of ssd disks.