Digital Ocean vs. Linode
blog.schneidmaster.com
blog.schneidmaster.com
• Complaints about DO shutting down instances and locking accounts until people FAX'd in a copy of their driver's license, passport, etc. (If a user isn't breaking the law or violating the ToS, why does DO need identification? If a user is breaking the law or violating the ToS, then terminate their account!)
• Having to stop accepting signups in their Amsterdam location for a while because they ran out of IPv4 space but didn't support IPv6. (A message from August 2012 says, "we hope to have something in the works by Q4 2012 or Q1 2013." As far as I can tell, they still don't support IPv6 in any of their locations.)
• Not supporting the booting of custom kernels. (They've been stalling on this at least as long as IPv6 support.)
• Botching their Ubuntu image and having everyone's Ubuntu droplets sharing the same ssh host keys unless the user explicitly generated new ones. (Probably an honest mistake, but that's sloppy, especially considering they're effectively forcing users to use their distro images.)
The recent kerfluffle about not always wiping data after a user's disk image is destroyed would've also been a gigantic red flag (that had been doused in kerosene and lit on fire before being run up the flagpole) if it had happened a few months earlier.
To be fair, Linode's security/privacy record is far from spotless — they've been super evasive about breaches, and their handling of a data leak bug/vuln I reported didn't leave me with a super-awesome feeling — but my gut feeling is that Linode sucks just enough less to justify the US$15/month extra. That could easily change, though, but I imagine it would turn into "both suck, screw it, I'll suffer with an EC2 t1.micro instance."
* No verification required for me. Perhaps this is a step when paying by paypal? I've been asked many times for more info when using that. If they pulled this on me, my backup MX will handle email until I get to sort it.
* AMS1 was unfortunate. I signed up when it wasn't available and I'm in Europe. I moved my machine to AMS2 when that came up (after their first day of scary high load). I had my host in NYC2 to start with and it was perfectly usable from London, UK over SSH. I couldn't tell the difference between it and my server in the house.
* Custom kernels. This is one of two gripes for me. I really want FreeBSD but failing that I want the latest Debian kernel. Keeping an eye on this one. I've got ufw/iptables up front which I have my fingers crossed will protect from any network level issues, logwatch, fail2ban etc to monitor casual attempts and patches are tracked.
* SSH key generation: I always regen my SSH keys anyway if I didn't see it happen so this would have been a non issue for me. This is crypto paranoia on my part (and well justified).
* Wiping data: when I saw the "securely destroy my data" checkbox I ticked it. Why would you not tick it?
To be fair, for $5/month it's not bad. Having played with shared hosting for 17 years, managed massive colo custom deployments and paid through the nose for other hosting, it's the best compromise so far.
I looked at Linode but the bottom end was slightly too expensive and I'd rather have SSDs behind it as IO contention on VPSs is usually rather high so that got ruled out.
I tried an Amazon EC2 micro instance and it was horribly slow so they didn't get my cash.
And it's actually ticked by default in the GUI, just not in the API (which is dumb, yes).
Others have noted that it is ticked (I don't know).
But I will say that as a marketing move it's a good idea. Because you are actively informing users of something that they might not have thought about or that competitors don't mention or mention in a way that is ignored.
Amazon does a version of this with shipping. They allow you to select slower shipping for the same price as the default fast prime shipping.
IMHO, it should not even be an option, it should be default, always.
They offer a cloudy-type self-managed virtual machine platform which is close to your £12 budget.
probably some bad apples were using it to host botnets or using credit card fraud
also performance on any given VM may be 5x difference than another vm in the dc. (at least at the $5 tier).
and yes, though there are technically 5 DC's to pick from, quite often there are less than 5 available when you want to provision a server. a few times I've seen only 1 DC available.
That said, if you understand and can tolerate this "sloppyness" then digital ocean is a great system. I am perfectly happy with their service right now.
EDIT: clarifying details, my SaaS has been going for about 6 months, and currently runs 25 digital ocean instances across all 5 of their DC's, so I think i have a pretty good sample size.
I haven't considered switching to a cheaper service like Digital Ocean, because Linode offers peace of mind and excellent customer retention. There is something special about dialing a number and getting a real person who knows how to configure DNS and troubleshoot your international propagation and provide real assistance and empathize without sounding like they are feeding you canned BS.
I've never rebooted my Linode, except for their free upgrades. Recently I've purchased their backup service, which saved my ass when I decided to upgrade my Arch Linux install and butchered a package conflict which resulted in effectively destroying my box. I was able to quickly spin up a new install of Arch, mount my previous server, and copy over configurations. The LISH console has also saved my ass as well... too many times; I'm not a server admin!
I feel very much in control of the remote box and that is primarily why I will always recommend. However, they had one hiccup during my time with them, where they leaked sensitive information and withheld accountability for a while. It was frustrating, because I gave them benefit of the doubt, but secretly agreed that it was sketchy. If anyone reads this who works there, staying quiet matters negatively towards your credibility.
I don't know if DO would fare any better, and they have issues[2] of their own though.
Evaluating the two, DO makes me less uncomfortable and costs less.
1. http://news.ycombinator.com/item?id=5552756 2. https://news.ycombinator.com/item?id=6983097
I hear this repeated over and over but did you just read the headline and draw your own conclusion? From all the articles I read, the hacker maybe got some worthless encrypted cc data. I don't think anyone lost sleep over this and kudos to Linode for encrypting user data. Also the breach had nothing to do with the individual VPS accounts.
Around that same time Amazon had the most epic multi-day outage in history and their storage servers were leaking user data. By comparison, the Linode hack wasn't a big deal, IMO.
PS: As far as Linode holding back information, were you a Linode cusotomer last year or are you just repeating rumors? Your Slashdot article was posted April 15th. Linode announced the "security incident" on the 12th https://blog.linode.com/2013/04/12/security-notice-linode-ma... also https://blog.linode.com/2013/04/16/security-incident-update/ I did not get the impression they were holding back information.
I don't care what the hackers got per se, just that Linode was so opaque. However, the opaqueness in light of credit cards being stolen (encrypted or not (can most likely be decrypted anyway)) is even worse imo.
<snide> Though I'm skeptical enough of anyone still using Cold Fusion. </snide>
Not saying you should trust Linode, but that your confidence in thar rumour probably shouldn't have improved after the parent comment.
If its some kind of business perspective then I don't know why anyone cares about monthly fees equaling 10 minutes of developer salary or 20 minutes of developer salary, or why any businessman would care which CSS framework his sysadmin occasionally uses. On the other hand an extremely long term proven track record of reliability and responsiveness to problems is kinda important.
Or is it written for bitcoin miners trying to get the most CPU cycles per dollar, or private dropbox clone who needs as many GB per $ as possible, or a developer who spins up test environments on a regular basis but doesn't know how to use puppet, or a vanity VPS where you spin up an instance in 2008 or whatever and just keep "apt-get upgrade"ing Debian since then or ...
As a disclaimer I am a happy very long term linode subscriber. I admit I laughed out loud at the suggestion that I switch because DO uses a different CSS framework.
Stop being obtuse. He listed a number of points he finds in their favor: hourly billing, less expensive, SSD performance, and a clean, easy to use interface. He even argued your point that Linode has a long track record of reliability.
This "sentence," however, is one of the most incomprehensible things I've ever read:
> Or is it written for bitcoin miners trying to get the most CPU cycles per dollar, or private dropbox clone who needs as many GB per $ as possible, or a developer who spins up test environments on a regular basis but doesn't know how to use puppet, or a vanity VPS where you spin up an instance in 2008 or whatever and just keep "apt-get upgrade"ing Debian since then or ...
As a disclaimer, I'm an AWS user, and I've had an eye on DO for personal projects for a while now. This article articulates every reason why.
He's compare and contrasting two services for ... who? While claiming it somehow explains why he switched services.
I can't figure out who he's tailored the writing toward. Its not focused, just a list of random things which most readers would consider irrelevant, although each reader would probably disagree on which individual points are irrelevant.
As far as I know it is all factually correct, just audience and analysis free. It could be a good example of data vs information vs knowledge in that its a good example of dense comparative factually correct data yet provides no information and no knowledge. Not necessarily bad, just a peculiar way to explain why he switched services. I would expect an article explaining a business decision to have at least some analysis and some applied knowledge/wisdom.
If I had to rewrite his article it would look like this. First, I think he's a starving student who spins up and down a lot of test systems for learning purposes, so thats going to be listed as his goal. Could be wrong of course, but I'll take this theory and run with it. So I'd start explaining what he's trying to do and whats important to meet that goal (see above). Then gather data, and toss the irrelevant stuff (who cares how many GB per $ in the assumed situation, who cares what CSS framework they selected in this situation, it can't possibly impact his particular goal). Then analyze how the data applies to his goal and how the data pieces interact with each other. Finally some knowledge/wisdom to rank and prioritize his analysis for his situation, like a starving student needs $15/month savings a lot more than long term reliability track record etc. Which leads to the conclusion that in his individual circumstance, it made sense for him to move.
I host a handful of things. My personal site, my blog, my Gitlab; Git Reports (http://gitreports.com/); a couple of medium-sized websites for organizations related to my university; and (currently) a couple of staging apps for some freelancing. I feel like my use cases are fairly common in that regard, although I suppose I could be wrong.
So I attempted to explain why I found DO to be much more useful for my use cases. I'm not exactly starving but I am a college student so price is important to me. The things I run aren't entirely inconsequential so specs and uptime are also important to me. Like I said in another comment, being able to easily spin up fresh environments is convenient especially for staging. I found the management features convenient and easy-to-use. I'm not sure why thinking that a better interface is a good thing is an invalid opinion.
If you're saying that I should have included more context about my use cases, I suppose that's fair. Other than that, though, I'm not exactly sure what your complaint is.
Some persuasive essay goals...
write for a specific preferably defined audience. Original post needed what you have in the reply above, your context of use cases.
write to explain why one particular idea is valid. no problemo laser on target.
rational arguments vs emotional. Well there's a time and place for both but you made the right choice.
Straight line of development? Stayed on the track, check.
Anticipate possible objections and deal with them directly, well ...
It's also very hard to understand your writing style.
RAM: DO gives 2x as much RAM at each price point
CPU: Unclear. Linode gives you 8 cores across the board, but fewer "priority" ones. Probably needs benchmarking to be a sensible comparison.
Disk: Linode gives between 1.2x and 2.4x as much storage at each price point
Transfer: DO gives more transfer at the low end, but Linode ramps it up faster. DO gives more transfer at $20, they tie at $40, and Linode gives more at $80+.
I agree with the author that Digital Ocean is generally cheaper for my use-cases as well, but there are some where it isn't. For example, if you want 8GB RAM + 16TB transfer (to pick a configuration favorable to Linode), that's $160/mo at Linode but $300/mo at DO. On the other hand, if you need more transfer only occasionally, DO can be cheaper because it somewhat makes up for less included bandwidth with much cheaper overage rates ($20/TB vs. $100/TB at Linode).
"dd if=/dev/zero of=test.tmp bs=4k count=1000000"
Local Hardware (Bare metal, SSD): 633 MB/s
Linode SSD: 338 MB/s
Ramnode SSD: 212 MB/S
DO SSD: 199 MB/s
Linode HDD: 30.6 MB/s
Local Hardware (Bare metal, SSD): 240 MB/s
Local Hardware 2 (Newer, Metal, SSD): 387 MB/s
Linode SSD: 355 MB/s
Ramnode SSD: 293 MB/s
DO SSD: 236 MB/s
Linode HDD: n/a, out of space
With that aside - when I'm looking for remote dev nodes to mess around on, DO and Ramnode are both good bets. They are cheap and discardable.
But when I want top-notch support and uptime, I go with linode. (Currently I have nodes at all three.)
"dd if=/dev/zero of=test.tmp bs=4k count=1000000 conv=fdatasync"
Even then, the host could be lying and telling your VPS that a write was successful when it was not actually written to the host disk. Still in the hosts disk cache, which means great performance but if the host ever crashes, all data on your VPS disk will likely be lost.
that tested what you got for $20 and found Linode faster. Personally I use DO and have not tried Linode as I only need the $5 thing.
This is from Digital Ocean:
FastResize is limited to the resources available on the physical hypervisor that your server is on. If there are no resources available, no sizes will be listed on the FastResize Page.
(Linode automates all these steps...)
If this is the case, you will need to take the slower method of snapshot & redeployment which is not limited to the same hypervisor:
1) create a snapshot of your server 2) delete your server to release the IP 3) immediately after the delete is finished, create a new droplet (choose the same datacentre) "select images" section switch to "my images" select your snapshot
About to do this tonight, fingers crossed!
Droplets look cheap but maybe they're competitive when you look at the number of cores. Seems to me a 4G Linode is faster than a 4G Droplet. But not 2x faster, IMO. A 4G Linode costs 2x the 4G Droplet. Digital Ocean feels like a better deal so far, but I need to upgrade to the 8G Droplet before I can really say for sure--the 4G Droplet isn't quite fast enough.
As far as SSD, can your user tell the difference between a cache file that takes .001s vs .0001s? Probably not. For many of us, I think the more critical question is: How much does SSD improve MySQL performance?
Doesn't MySQL store cache in ram, so wouldnt more ram/$ be better?
I started up a droplet last month because I needed the better deal on RAM. I've been charged only ~8$ for December.
DO is young but they come from the same team that built other hosting services -- which is why I find it odd that they've made a few of the same mistakes a lot of hosting providers make (re: security). I think we can give them a pass on that too since they're so relatively new.
Linode is great but I definitely see DO winning out over the next few years.
As for the people who work there? I'm sure they're reading this, and I don't mean to offend, but I think a lot of them need to take a look at the way they've handled a few of their recent security issues and the tone their responses have had. It didn't come across as 100% professional to me.
Happy to cite claims on request.
It looks like they've added things since I last looked because I don't remember the ssh_keys and domains stuff being there. I'll revise my comment accordingly.
Not for my money, Linode rocks.
Companies that deliberately withhold key information from their users during security incidents don't deserve to be hosting other people's services. In fact they really shouldn't be in business. Transparency is that paramount. And Linode has done it not once but twice. I will never forget learning that my entire Production Linode stack could have been hacked through a Reddit post rather than from the company itself.
You really only have one obligation to your customer. Don't hide things from them.
You get what you pay for.
Disclosure: I use linode as well as other providers. Completely happy with linode, not a second of downtime in close to a year, great performance (including disk) and even a far nicer management interface than the screenshot in the link (why do people like not being given choice? Is it some ADHD thing?...).
Another thing to compare them on is DDOS protection and mitigation. I don't have enough data to compare both, in fact nothing on DO's side but I know that getting null routed on Linode is not a very good experience. You don't get much info from them and it's one of those helpless situations I guess. Does anyone else have experiences with either?
That is why i am always looking for a Hybrid, a company that offers dedicated model while providing Cloud / VPS hosting. However i have yet to see a single company that does it with a price competitive option. SingleHop offers that but at a relatively expensive price. Hivelocity does very good dedicated but their Cloud / VPS sparknode gets no love. LimeStone Network and Incero are bothing testing their Cloud. While OVH still have absolutely no idea what it is doing although they think they do.
Back to DO vs Linode I am still skeptic of DO. After all their problems with VM Data Left over, Backup, Data Loss etc. I still think they are more of a Beta services compare to Linode, which is rock solid. Apart from the hacking issue Linode had.
Linode offer much better CPU performance then DO at all level. And with SSD (cache?) they should be competitive with IO in 2014 Q1. The thing that i dont understand is Linode decide to have SSD + HDD solution, where 500GB HDD + 500GB SSD works together. I dont get how this solution is better then just 500GB SSD alone.
Linode are also much better with Bandwidth and Network. This could be just that DO have growth problem where their user base exceed their DC bandwidth. But in general Linode's Network are much better tuned.
Support is also better at Linode.
So basically Linode is better, but more expensive. I remember Linode does offer daily price based where they charge their client a full Mouth price and then refund it to their account on a per day basis.
I sorta agree although I think the barrier is somewhat higher than $100/month.
One of the biggest wins with a service like Linode is that if hardware fails, you often don't even need to know about it. And if your server were to fail, they'd either get you running again from your image pretty quickly or you could at least spin up a new VPS from your backup.
With a dedicated server, no matter how cheap, getting back up and running again can take a while even in the best of situations unless you're running hot spares or are already load balancing. It takes time to commission new boxes even at the best of the dedi providers and usually you'd just wait for hardware fix/replacement which can take hours.
I've been using DO recently because I was given some free credit - mainly for testing Ansible. Their pricing is very aggressive and they do have that new fresh out of the box feeling.
I haven't used anything from them in production yet though so can't do a full comparison. Negatives for DO are they don't support private networking in all of their datacenters, they also don't pool bandwidth (as far as I'm aware) between all of your instances, I haven't seen any references to permanent/fixed IP addresses in the DO control panel - they seem to be dynamically assigned on each droplet spin up.
I haven't used DO support but Linode support has always been fast and accurate for me. For me, at the moment, DO isn't offering anything more than Linode so moving sites doesn't make sense in my case (disk IO isn't an issue for me at present).
I look at the offerings in general and despite the fact that I've used neither, but am going to, it does look like Linode offers slightly 'more', albeit for a substantially higher cost?
It really depends on your use case. I run my personal website, my blog, and a couple of other websites that get decent but not spectacular traffic (think a few K visits a month tops). I think that made Digital Ocean a pretty good choice for me because I take advantage of better (IMO) management tools, better third-party apps, etc. However, if you're running some seriously high-load stuff, Linode has some enterprise-level features that DO lacks, like load balancing. Linode also seems to have somewhat more technically competent support from what I've read; I've never had to use either company's support though.
Deleted comment
I created a ridiculous support ticket (with lowest priority) about how annoying their youtube ads are. Got response within two hours that they were disabled for my IP. It's a small and silly issue, but I was surprised that they even responded, not to mention they fixed my "problem".
Linode is for more mission critical stuff where I need better uptime and tech support.
DO is for less important sites (blogs, family websites, etc)
Also, Linode don't just charge a flat fee for the month, they pro-rate costs based on when VPSs are created and destroyed in the billing cycle.
I've been meaning to switch to DigitalOcean because it's twice as cheap but it's too much hassle to switch all domains, etc. so I've stuck with Linode.
I've never had to deal with Linode support which I guess is a good thing. DigitalOcean support is also great.
My credit card was stolen from Linode (I only ever used it to pay for Linode and Amazon). Luckily it was blocked by my bank automatically. Because of this, I will never call Linode rock-solid, secure or trusted.
Also, I miss is backups and on-the-fly snapshots. I looked all over DO's site and eventually had to write a support ticket to find out how to activate the backup section in the dashboard. It turns out they had removed this option. Bring it back! I could consider moving to Linode just to get backups.
On AWS I could take a snapshot without rebooting. On Do it says it can take up to an hour, which seems a long time.
As a safety precaution, I always replicate snapshots that I care about to all their regions before I destroy a droplet. And I always tell others to do the same.
I ended up at DO on a recommendation and have been happy since. The Linode stuff was recommended as well, but if I'm being honest it intimidated me (not as much as Rackspace/AWS). I learned a lot from those experiences and would feel better on a lot of other servers now...but the biggest thing I learned is at the very least find a company that a lot of people are talking about, as that means you won't be a lone voice trying to get support 3 days later.
Their 2 core, 4GB, 150GB, 5TB plan is actually cheaper than DO's (~$27 vs $40).
Not affiliated, I just have a (personal) VPS with them for a couple of years, and it's been fine.
I realize that you can do this on the user end, but I neglected to when I set up my droplet, and I feel like if they're willing to charge you for overages, they should provide some method of bandwidth monitoring.
https://www.digitalocean.com/community/questions/bandwidth-t...
To add, I did my ping time testing with digitalocean using their test servers[0]. I also created my own droplets and saw the same latency/speeds.
Both have been great, but Linode's support and response time to issues has always been excellent.
Code comes from ServerBear, also a great site to find yourself a VPS that fits your needs.
I am happy with a European VPS provider (TransIP), costs about 12 EUR/mon (with VAT and all) for 1G Ram, 1 Intel Xeon, 50 GB, 1.000GB traffic.
[0]: https://www.digitalocean.com/community/articles/digitalocean... [1]: https://www.linode.com/backups/
It's a real shame, because if tools like Vagrant and Packer worked with Linode like they do DO...
Linode for production
Digital Ocean for staging
I started using digital ocean and it's just too wonderful. I should've made the switch a long time ago. linode is good too but digital ocean's droplets are a joy and at half the cost and SSD, it just adds more. I also noticed that the dns management was super simple with digital ocean, it recognizes gmail mx records and adds the rest for you. these little subtle attention to details provides a very pleasant experience. in the end, it was financial but user interface that sold me and I'm here to stay with them.
the fact that you can take snapshots of droplets and clone them, and have an api to spin up new ones and also shutdown ones that are idle is just too awesome to ignore. not to mention quick ram upgrades and disks.
one of my droplets cost $5 but it shows the charge by hourly and if I shut it down, I can save even more money. I know this is little pennies but I love the fact that they give you control over this. It's almost like Amazon AWS but way more friendlier and easier to understand.
I went to DO because their pricing is clear and only then did I discover that everything just works.
I have a few servers and could not even think of switching.