Comments on: Hands Down, Best VPS Hosting at Slicehost

By: Angel Tipa

Angel Tipa — Wed, 16 Dec 2009 02:44:59 +0000

Never been used as one of my VPS host. I should try this after I finished my contract on Dreamhost. :) If its the best I should try it and I will add it to the top 10 vps list.

By: Mike

Mike — Mon, 07 Dec 2009 14:19:39 +0000

@rob t VPSs which exist as Xen domU's aren't floating in some ethereal layer above the hardware -- each virtual server is very much tied to the hardware it's running on. Migrating a Xen domU to a different host server isn't a big deal, but the way this currently works with Slicehost and many other Xen-based services typically involves an rsync of the VPS's data (in several steps) and a little bit (or a lot, if it's a huge filesystem) of downtime. @tech-no-junk-ie You're exactly right about two disks failing at the same time. 99 times out of 100, if a "disk issue" has resulted in VPSs needing to be migrated off a server with disks in a RAID 10 configuration, it means that two+ of the disks are failing at the same time -- support staff have got to get the customers off that box before data gets lost!! In datacenters with thousands of servers, you do see that from week to week -- it's just a matter of probability in terms of the failure of electro-mechanical devices. @Derek Broderick I've got to agree with Chris here. A hosting company should be able to tell other customers on a box that "some other customer's misbehaving VPS" is the cause of performance degradation for all customers on that box. Doing so need not reveal any personal information, IPs, etc., about the other account holder nor his VPS. It would be nice if the various virtualization platforms so completely and perfectly isolated VMs that they can't affect one another; however, that's note the case in the real world. For example, heavy swap usage by one VPS can lead to crappy I/O wait times for other VPSs on the same box that are hitting the disks pretty hard themselves. Newer versions of Xen are helping with that, but it's the kind of thing that support staff for these hosting companies are dealing with.

By: Chris

Chris — Mon, 07 Dec 2009 05:37:30 +0000

@Derek Broderick I'm not sure how this represents a privacy breach. As a customer with less-than-expected performance, I had the right to know why my slice was under-performing (a thrashing neighbor) and what was being done to mitigate the problem (kicking the slice). The fact that the account was inactive is irrelevant, and no personally identifiable information was revealed. I'd have no way of knowing exactly which slice was misbehaving, which account it was linked to, or who owned that account.

By: Derek Broderick

Derek Broderick — Mon, 07 Dec 2009 04:48:29 +0000

Sounds like Slicehost isn’t a big fan of respecting their customers’ privacy policies, if they told you about the neighbor of yours who was thrashing.

By: tech-no-junk-ie

tech-no-junk-ie — Sun, 06 Dec 2009 22:49:59 +0000

@rob t really, no migration needs to happen with raid 10 just a straight drive replacement unless both drives where failing. So he shouldn't have been informed unless there was a possible corruption of data. Also nice to hear that anyone that could be trashing the server is going to get kicked. Makes me feel very comfortable :p

By: Chris

Chris — Sun, 06 Dec 2009 21:20:55 +0000

@Fjord Actually, Linode was the runner-up in my VPS searching... to be honest though, I forget just what the deciding factor was. @rob t Good perspective. Although necessarily there must be /some/ (however brief or trivial) service interruption when migrating hosts, I agree that the issue could have been dealt with a lot more transparently, and I hope the folks at Slicehost are working hard on doing just that. More to the point is the fact that the staff there are responsive, helpful and friendly in dealing with circumstances from the most mundane to the most exceptional. How often, when dealing with a business (any business... a VPS hosting company, a department store, a pizza shop) do you walk away thinking to yourself "Wow - that was really exceptional service"? It says a lot about Slicehost as a business, and as a team - the technical stuff will continue to refine itself.

By: rob t

rob t — Sun, 06 Dec 2009 20:17:34 +0000

A HD failure should be handled transparently by the hosting company, and you should be migrated to another host without any service interruption, and the new host that you’re migrated to should be able to host your VPS with the same level of service.

Although they have good chat support, and I love (and use) their product, this story is not as positive as you make it out to be.

By: Fjord

Fjord — Sun, 06 Dec 2009 20:08:06 +0000

I feel compelled to give a recommendation for Linode (I’ve never tried Slicehost or any others). I’ve been with them for nearly 5 years now for my personal server and nearly 2 years for Birdstack.com. Support is great, and often just catching one of the admins on their IRC channel has been enough to solve my problems (which have been very very few). Over the last 5 years, I’ve had maybe 10 hours total downtime that was their fault–I don’t really remember for sure. And, a recent performance comparison puts them at the top: http://journal.uggedal.com/vps-performance-comparison

Anywho, maybe Slicehost is awesome, just wanted to give a shoutout for Linode. :-)

By: Mike

Mike — Sun, 06 Dec 2009 19:32:35 +0000

@Jason Rackspace Cloud Servers are billed like a utility: you pay by the hour for the VPS, and you pay for bandwidth usage on top of that. Slicehost offers monthly plans (prorated per diem), which include a monthly bandwidth allowance. So whether Cloud Servers are actually cheaper than Slices will depend on your usage.

By: Jason

Jason — Sun, 06 Dec 2009 18:50:32 +0000

As far as I can tell, rackspace cloud is the same thing as slicehost (rackspace acquired slicehost) for about 1/2 the cost