Quarter of Media Temple users suffering 18+ hrs downtime | Hacker News Reader

Quarter of Media Temple users suffering 18+ hrs downtime

weblog.mediatemple.net

24 pointsmootymoots25 comments

The problem that I have with mt is that they sell themselves as the "don't worry, put your site with us and nothing can go wrong" company. And that can never be true. No one can make promises like that, but those promises mean that they don't tell you things like backup strategies, replication, data centers, etc. And so you're just left with a promise that they've taken care of everything.

While I don't expect a company to tell me everything, I much prefer Amazon's take on the issue. For example, they specifically quote an annual failure rate that they expect and offer S3 snapshots (which are replicated across multiple boxes in multiple locations) as a backup. No hype or promises; just the knowledge I need to make sure I don't loose data.

With companies like Slicehost, Linode, or Amazon I know what I'm getting: a Xen instance with a certain amount of RAM. I can then replicate data, take backups, etc. to make sure that my site stays up and running. With Media Temple, I just get the promise of "hundreds of servers working in tandem to power [my] websites". Clearly it isn't 500 servers just working on my site or anything like that. So, what is it? It's a mystery and I find that when companies try to be mysterious, it's not because they're giving you more; it's because they're giving you less.

Glad I switched from them. Such an overhyped company. I've been on Slicehost since February 23rd of last year and have absolutely no complains. On (mt) I got shoddy support and a lazy staff (how the hell can you promise svn without ssh support for that long), for an incredibly overpriced hosting service.

Me too, +1 slicehost

Same! (mt) kept setup their containers to restart if they ran close to their memory limit! We moved to slicehost and haven't looked back.

(mt)'s gs offering has been absolutely terrible for us. Since the summer uptime and support have both been equally awful. The only redeeming quality of the gs service is the insanely large storage (100gb).

We have about 30 sites hosted with them and I have been none too pleased. The service is shoddy and disapointing. Currently, they are running Subversion 1.4.2 on the gridservers packages, which is nearly 3 years old. I've asked if they can upgrade to 1.5.x with no avail.

Our customers constantly complain that their sites are down, even if only for minutes at a time. I know we're only paying $20 a month for about 10 separate accounts, but I get better service from Dreamhost.

I think they're overpriced. I've dealt with tons of shared hosts for much less with better service. It seems like people get suckered in with their pretty website and the fact that a lot of big sites us them without doing much research.

Ouch, that's terrible. I'm glad I switched a couple of months ago -- if I didn't this would've affected me. :/

Anyone else notice how all of the sites on (mt)'s (gs) client page happen to be online every time there's a major outage?

It gets worse for those affected: http://weblog.mediatemple.net/weblog/2009/03/01/cluster-2-im...

Sounds like there's gonna be some financial compensation according to some messages from MT on Twitter to customers voicing their disapproval at the now 24+ hours downtime.

People pay only $20 per month and then complain like they're paying for hosting at rackspace. The service is a great value for what it provides. If what you're doing is that vital, you shouldnt be hosting it on a grid server at media temple

Alternatively just pay $20 to Slicehost or Linode and get better service, oh, and skip the 18 hours downtime; but sure you're right, when you pay $20 you shouldn't complain when you receive a crap service--whatever.

20 hrs now. It's not looking good... So glad I'm on cluster 4. Twitter is full of unhappy campers.

This has been compounded by the outage earlier this week too on the same cluster!

> So glad I'm on cluster 4.

So glad I lease real servers.

I'll second that. Rackspace ain't no $20/month, but as with most things, you get what you pay for.

The only recent downtime was an instance a few months back where I fat-fingered an iptables config. A tech was on the pulling up the console and cleaning up my mess while I was still typing out a support ticket.

Third that as an ex-MTer currently on Slicehost/Mosso. I've personally had sub-five minute resolution times with Slicehost support. If you're reading this, fantastic jobs guys.

Rackspace is on the very high end of the cost spectrum, and the service really isn't that great. But at least their servers are fast and real, if you/your organization doesn't care about the budget, they're probably as good a choice as any as far as quality of infrastructure goes.

Just make sure the person signing the hosting contract actually reads it, also. You might be surprised to find what they try to bury in there.

It's worth noting that they bought out their lower-priced competition at ServerBeach.

>> "and the service really isn't that great."

Not sure what/who you consider "great" then. Rackspace has been stellar for us–always available on the phone; answering tickets fast; alerting me when there are problems.

I've never had a bad cs experience with Rackspace. (I've got 6 servers with them)

Their technicians once left 2 active network interfaces down for 45 minutes on one of our servers when upgrading something and didn't fix it until we called them.

They're just overworked like most NOC guys.

When I was researching server space 2.5 years ago price wasn't an object (ahh, remember those days?).

Anyway, I've never had any reason to complain about Rackspace service. Now, I'm not saying the service is a good value (you definitely pay a premium), but it's been solid for me.

My non-essential sites are down (I didn't notice until I read it on HN). Good thing I'm hosting my emails on Gmail or I'd be getting calls from friends I host for free.

I just cross my fingers hoping this ain't gonna happen to our customer's site (Cluster #: 1 Storage #: 4). Maybe they will become a better hosting after all.

I'd never heard of Bluearc before. Anyone else played with Bluearc storage gear?

I have. If you need a really high performance NAS it's a reasonable choice. Better price:performance than NetApp, but still not priced as low as it should be IMHO.

FPGA powered: http://www.spec.org/sfs97r1/results/sfs97r1.html

We just transitioned a customer off of MT onto some cheap dedicated servers with Amazon S3 storage and not only did their hosting costs go down, the performance went up dramatically.

In the business parlance of today, my stakeholder expectations were not aligned with their soup to nuts value proposition.

The 'grid' has TERRIBLE database performance and always has.

They better think of an interim solution, because databases with schemas are not going away for quite a while, even if they are eventually going to be replaced by giant single key hash tables like everyone is clamoring about these days.

Reply on news.ycombinator.com