Why we moved to Linode | Hacker News Reader

Why we moved to Linode

blog.superfeedr.com

111 pointsjulien42 comments

I just want to add one alternative voice; I've been a slicehost user for about 2 years and I've couldn't be a happier customer. It's the first time I've used a VPS like this so I can't comment on how their service compares price/speed/functionality-wise with Linode et al., but I can say that:

- My sites haven't once gone down.

- I have found the whole interface and management backend very intuitive.

- Their articles are fantastic. I went from pretty much knowing nothing about servers or linux to running 6 sites on my slice, with email server, FTP, securing it with iptables, django installed, installing postgres with remote access, etc etc, all through their articles.

- The support is exemplary. Last week I accidently made a double payment (my stupid fault). I emailed them, a human responded within 40 seconds saying it had been forwarded to the billing department; 90 seconds later the billing department emailed me saying they had reversed the transaction.

(edit: formatting)

I'm with Linode, but I have to give kudos to Slicehost for their documentation. This is my first time running a VPS and I have gone to Slicehost's tutorials just as often as I have to Linode's. I've also found the videos through vpsbible.com to be a big help as well.

I assume you've checked out Linode's Library? It's pretty comprehensive, with new articles being added almost daily (they have several full-time staff members who spend a great deal of time writing/testing library articles).

You wouldn't be happier if you got virtually the same experience for 40% to 50% less money each month?

Having used both Slicehost and Linode, the (user) experience is similar, node performance is better, and the price is cheaper.

I never said I wouldn't be.

I have nothing against slicehost, but a key point in the OP was price. Ever since their June RAM bump, they (on paper) offer a much better value.

Its also worth pointing out that of the various benchmarks I've seen (a handful on blogs and there are some pretty key threads on webhostingtalk.com), Linode typically comes out on or near the top.

On top of that, you add their participation to things like Rails Rumble and their data center choice (just missing Asia), and it all makes a compelling argument.

Even before the Linode memory upgrade, Linode was more competitive for many uses because they offered 32 bit systems, whereas Slicehost only has memory-hungry 64 bit systems:

http://journal.dedasys.com/2008/11/24/slicehost-vs-linode

After the ram upgrade, Linode is a clear winner. I just hope they don't screw up the good service as they gain customers.

> whereas Slicehost only has memory-hungry 64 bit systems

http://www.slicehost.com/articles/2010/3/1/32-bit-images-now...

Good catch, I didn't realize that. Before writing my article and switching to Linode, I politely asked if that was going to happen any time soon, but they said no, and the rest is history. It took them, it looks like, over a year to add that functionality.

I'm with Slicehost as well and agree. I did have on major downtime though - a incorrectly configured RAID wiped my entire slice. Luckily, I only had a couple smaller sites on it at the time... but what impressed me was they gave me a year's worth of credit and did so without me even saying/asking anything, it was right in the email they sent explaining what happened. Mistakes happen, and they responded in a great way with it and because of that I have recommended them to many.

That said... I am looking at Linode now just b/c of pricing. I know you get what you pay for, but I've only heard good things from Linode as well.

I was rocking with some dedicated servers from SoftLayer (who are awesome) but "downsized" to Linode about 18 months ago. I discovered VPSes weren't as slow or crippled as I imagined and I now serve much the same stuff for a fraction of the cost. Win win.

Linode is awesome and great for small- to mid-range server needs. Even an EC2 small instance is overkill (both in computing resources and in price) for many websites.

However if you do need lots of resources then the cost effectiveness of VPSes - or any virtualized system for that matter - goes down the drain quickly. For high-traffic websites nothing compares to good old dedicated servers. The pricing difference between a dedicated server and EC2/Linode/etc can be as much as 5x for the same resources.

Linode is also a bad choice if you need a lot of disk space, and with "a lot" I mean >100 GB. Expanding disk space on Linode is very expensive, an order of magnitude more than EBS.

Yeah, disk space is pretty much my only complaint with Linode. I'm not a big fan of their new manager interface, but that one's not a big deal.

Otherwise, Linode is really amazing, and the author is right -- they do seem to always be improving their service.

How is the pricing difference for a dedicated server and EC2 5 times? Maybe if you only look at the physical hardware cost, but often that is the smaller component. You are comparing a capital asset with both fixed and variable cost components to a fixed-cost per-hour lease of resource.

If you really need to lease resources by the hour then by all means, go with virtualized servers. But for one of the web apps I'm building I know that there's a high chance that I'll need to keep the servers around for at least a year. The prices on e.g. hetzner.de (I'm not a customer, I just read about it on HN) totally blows EC2 Reserved Instances out of the water.

I don't know why you consider dedicated servers capital assets. Maybe if you're colocating your own servers, but I plan on renting dedicated servers that a hosting company provides. They'll be responsible for replacing the hardware if it fails.

Turns out that renting a dedicated server and colocating your own costs approximately the same, assuming that the hardware never fails. If the hardware does fail then renting a dedicated is cheaper.

"Renting" also has tax benefits in some jurisdictions since you can write off the full rent instead of depreciation over time.

I've had a very similar experience with Rackspacecloud (makes sense since they're the same company as Slicehost now): Incredibly slow control panel, features sometimes work and sometimes don't, useless tech support (took 5 days to point our server to our domain because someone else had it pointed first for some reason, despite whois being obviously ours).

Meanwhile, I've been using Linode for my personal projects and it's night and day. The control panel is incredibly faster and easier to use, the prices and specs are better, and the community/documentation is noticeably more intelligent and better-equipped.

My ping times from my server to Facebook's graph servers went from 40ms at my expensive ThePlanet servers to 2ms at Linode Fresno. I'm really, really happy with Linode, though the prices for bandwidth, disk and ram are ridiculously high. Best option is probably something like 10tb from ThePlanet for bulk assets and Linode for the front end.

I moved to Linode recently from vps.net, nothing major just a few personal sites, but so far Linode have been great. If you're ever tempted to, don't go with vps.net.

I've had a great experience with VPS.net so far. Why do you say that?

I signed up in September 2009 and I'm still technically a customer but I won't be once renewal rolls around.

The problems are numerous, the management are great, they really care (Nick, Terry etc) but the product and support team are pitiful. I assume as you're a customer you've got access to the customer forum? Take a look at the problems people face. vps.net is great when it works, but the second something goes wrong I've found myself nothing but screwed and personally how the system and team handle when things go wrong is what matters most.

"Self healing with automatic failover" is a joke. I've had 2 periods of 10 hour+ downtime on tiny servers, I've also had numerous "small" downtimes that lasted between 30 minutes and 2 hours. The reasons for these range from technicians making mistakes ("replaced ram incorrectly") to the technology just not being able to cope.

I wish I could stick with vps.net, their flexibility isn't beaten by anyone but the stability and support is incredibly poor. If I were to totally ignore any times my server has been rebooted because of me (so assume 1 year = 1 year of uptime) the longest my server(s -- I used to run multiple with them) have been online is 90 days. They can't go 90 days without a problem, where problems range from 30 minutes to 10 hours.

The last straw for me was my server being inaccessible for almost 90 minutes, the reason? I had been nullrouted and that isn't the bad part... I submitted a support ticket explaining that my server was down (I originally logged in to the control panel after I couldn't get in via SSH and saw many disk errors) I got a canned "reboot plz" response, I sent a reply saying that didn't work, he replied "I can connect from here!" so I replied "Well I can't!" and included a link that showed it wasn't just me (http://i.imgur.com/NtXi9.jpg) then his next reply "oh, it appears you've been nullrouted!" (this was an hour after I first submitted my ticket) so I reply asking when it'll be back, they ignore me and then 15 minutes later it's accessible again and the ticket was closed...

...it doesn't end there. I head over to the customer forum and ask if anyone else is experiencing the same, surprise surprise another customer says "hey I'm experiencing the same!" so we compare IPs, he's on the same (subnet?) IP part as me, xx.xx.xx. (where is our specific parts and xx is the same) so he then submits a support ticket about it. So specific to this incident: Support are not told when blocks are put in place, support don't proceed to contact all the other customers with the same problem when it is resolved and they don't offer any sort of "sorry" explanation. So either support doesn't have the facility to check for incidents specific to a client (why not?) or they just don't bother.

Anyway yes, that's just one incident but it was the final straw. vps.net have always been good on the customer relations side with Nick always talking to customers, but the product is poor (have a SAN die and then enjoy the wait time while it "self heals") and the support is just as bad. I'd love to pass issues off as one time, or things that will get better, but after an entire year of seeing it remain poor I'm finally moving on and it isn't just me, the customer forum is full of issues.

oh and don't even get me started on their inability to communicate, you'd think a product aimed at people with important things would have people monitoring the network and hardware, but apparently not, because it often takes at least 30 minutes for them to notify people of issues in progress via status.vps.net (which FYI a bunch of customers had to hammer them to get online, which they still don't publicise).

http://www.vps.net/forum/customers-forums/clients-only-forum... < A group of clients down, Managing Director says he's going to "look into it" and confirms there is an issue. He has to look into it? Don't they, you know, monitor? The problem? Licensing issue, a licensing issue took down hundreds of servers for over 4 hours? It's a joke. http://www.vps.net/forum/customers-forums/clients-only-forum... http://www.vps.net/forum/customers-forums/clients-only-forum...

Summary: Unstable/unreliable product, poor support, good customer relations, terrible issue awareness. Broken promises.

Edit: I didn't realise how long this had become, apologies. I'm just very frustrated that I've stuck with them for so long (I was considering leaving after 3 months of nothing but pain) and they still suck.

Edit edit: I figured an example of the horrible support where it caused substantial downtime would be good. Back story: Server went down, checked the client forums and many others had the same problems, after an hour Nick (MD) said via Twitter that anyone still down should contact support, so I'd already been down for an hour:

09:34 EDT Submit ticket, say that my server is still down and this ticket is as instructed via Twitter.

09:38 EDT Reply from support, "Will check this and update as soon as possible".

10:07 EDT I replied asking where my update is, explain that I've now been down for 2 hours.

10:39 EDT I finally get a reply saying that my server is queued for being brought online

11:46 EDT Server still down I ask what's going on...

13:36 EDT Still no reply, server still down.

19:40 EDT Finally a reply from the MD (Nick) saying that my server is now back online.

So as you can see, between "Your server is now queued to come online" and the MD replying, I got nothing, all the while my server was down. It's laughably bad, especially because this is one of the many terrible instances with support I've had.

yea, inquiring minds would like to know. I gave gigenet a try but they were having major issues with their SAN when I signed up. Looking to use vps.net once my product launches.

I recently built a project (http://www.coatapp.com - A simple weather forecast application with clothing suggestions) on Linode and couldn't be happier. The disk speeds are very impressive, and super useful for apps that deal with large amounts of data that gets updated frequently (i.e. weather data).

This previously posted benchmark shows some of the difference between Linode and a few other providers (from 2009): http://journal.uggedal.com/vps-performance-comparison

Cool project, would be really nice to have celsius support though.

Ok, I'll look into that too. The app only offers weather forecasts for the continental US so I originally assumed most folks were interested in fahrenheit, but I can see the appeal of celsius :).

My zip codes don't work in your app, but I really like the idea.

07030 - Hoboken, NJ 07647 - Rockleigh, NJ

It looks like those zip codes weren't in the original zipcode dataset. I'm adding them now, thanks for pointing that out! update: It was actually a bug in the code, fixed now :)

I'm very happy with my ARP Networks VPS. Their prices are very competitive, but it's definitely not a "managed" solution. They support just about any Linux distro as well as FreeBSD and OpenBSD. Once your VM is spun up, you just log in and go. Support seems to be email only, but I've never had a second of downtime in the 9 months or so I've been there.

My only gripe is that web-based management of the VPS is almost non-existent. Basically power on/off and reboot are all you get. I'd like to see bandwidth, disk I/O, CPU utilization, and so on. They have a text-based management interface with a few more options, including VNC console over SSH. I suppose the end result is that you basically get a low-end colo server experience at a VPS price. (Which is fine if you're like me and that's what you want.)

I also want to tip my hat to Liquid Web's cloud offering, Storm. (I'm biased because I helped build it.) Their prices are certainly not on the low end but the support team beats the snot out of any other hosting provider, period.

I wish I could move to Linode from Slicehost, I have a slice with my personal/hobby sites on it but it's too expensive (I could have a Linode twice the size for that price). All my new servers are Linode, of course (including historious, which is breezing along).

Their London datacenter is the fastest I've seen, even faster than some dedicated servers I've used.

I've got fond memories of Linode, as it was the first VPS I tried out. Having an always-on Linux box turned out to be a lot handier than I had anticipated, and these days I can't imagine being without a shell _somewhere_.

That said, I ran into a curious problem when attempting to run servers for a doom source port, where performance would be quite choppy at times. I never really figured out what was going on either, as their support just told me that the box my VPS was on was fine and they couldn't see anything wrong with it. I finally moved to a cheap Atom dedicated server which seemed to fix things, and now I'm evaluating a VPS hosted by interserver.net (prognosis is troubling, it uses OpenVZ and I seem to keep getting dgramrcvbuf failures under heavy server usage), but I'll always suggest Linode to people looking for a good VPS host.

I moved from Slicehost to Linode, mainly for performance reasons. The pricing difference encouraged me to wait no longer.

One of my sites was having serious performance issues on Slicehost, which seemed to be due to low disk i/o. I was cleaning the database, optimizing, etc. and stressing out about it, wondering what to change. After moving to Linode I haven't had the slightest hint of a performance issue.

I'm in the middle of making the same move. My motivations were price (a 512 slice costs half as much on Linode) and locality (Linode have a London data centre; Slicehost don't). So far I've been very impressed.

"Linode offers some kind of customization so you can make machines with more RAM, or more bandwidth if you need to." Any details on how this works?

Once you have a node you can upgrade the specs a little. Short of transferring to an entirely new node, you can boost your bandwidth, get some extra RAM, and add some disk space, within limits without changing what host machine you're on due to the slack they have.

Separate to that, Linode also appear to support significantly larger packages than their pricing table indicates. Once you're in the system, there are references to things like "Linode 16384" accounts, etc.

Ah, I see in the "Extras" page. The pricing seems dumb though-they want $20/mo extra for 360mb of ram, but for an additional $20/mo I might as well update my 512 to a linode 1024 and get an additional 512mb of ram, as well as more cpu power/disk/bandwidth.

I think that is exactly the point, if you look at all of the extras (bandwidth,disk,ram) the pricing all comes out to when you reach the amount for each (bandwidth, disk, or ram) upgraded server it is the same for both extras or upgraded server.

There's only so much resources on each physical host - when you hit the resource usage of an additional Linode, you've effectively used up the space that could have gone to another customer. So you have to pay their rent as well.

Although this really doesn't make that much sense for bandwidth... but when you look at the raw price for bandwidth it's not all that bad anyway.

They aren't listed on the home page, but as soon as you click through to "sign up," you can select plans up to 20GB.

I've bought additional IPs for our main server that we are hosting multiple SSL sites on. Extra IP is $1/month, I think you can get up to 5 extra IPs on one machine.

Linode indeed has great support,from my experience

linode for dev, test and launch and then dedicated hardware at 100tb.com or the like when needed.

I would have went with RapidXen instead of Linode. Better service, and the guy who owns it is a pretty well known FOSS coder.

Reply on news.ycombinator.com