Dreamhost is down. All of it. | Hacker News Reader

Dreamhost is down. All of it.

dreamhoststatus.com

33 pointstlrobinson33 comments

Now I know downtime is normal for a host. Yes, even Google.

But when you visit the homepage of their status-blog[1], don't you think that's a bit too many updates a little too frequently?

Or should we commend Dreamhost on being super-transparent in their network/uptime issues?

[1] http://www.dreamhoststatus.com/

DH are mostly just very transparent about downtime - more so than most hosts. They are also very cheap. If you would like more nine's, you will need to pay more. MT, rackspace, etc... are all considerably more expensive.

I always appreciate transparency in support issues. I'm much happier knowing that an issue is being worked on than sitting in the dark wondering if anyone has noticed.

I also see that they host their status blog outside their core network. Good idea :)

They are just very bad. I am so glad I moved away from them a year ago.

what is funny about the status page is that it isn't a day that they don't run into an issue ..

Given the number of domains they host (nearly a million: http://www.webhosting.info/webhosts/reports/total_domains/DR...), I'm shocked they don't have more outages to report. Perhaps they do. They've got enough hardware that system failures are bound to be a daily occurrence. And I'm sure software problems stack right on top of that.

if you look at the issues too there's a lot of "this server had to have an fsck to resolve something" and "theses servers were slow so we fixed them".

99.9% uptime is still something like 8 hours downtime - this big outage was about 40 mins so they've some leeway yet.

If it's of any concern, Dreamhost is also the host for two of my websites.

Semi-unrelated but if you're curious about BGP, this 2004 talk—although absurdly long—talks about some of the issues with BGP (an many other issues with hosting content on the Internet). The speaker is Tom Leighton, one of the founders of Akamai.

http://mitworld.mit.edu/video/199 (BGP at ~10 minutes in)

It looks like our core router lost its BGP session which took down our entire network (as far as we can tell). We have our network administrator heading to the data center now and should be there in 20 minutes or so. More updates to come.

Umm core router?! One? I'm thinking they'll revisit this strategy now.

Significant downtime throughout the entire history of the company doesn't seem to have caused them to revisit any of their strategies. Nothing will change, changing would cost money they clearly don't need to spend, or they wouldn't have retained their customers.

I liked this comment on their status page

"shabooty Says: I wonder if this site is hosted at media temple?"

epic

Good news, my DH hosted site just came back up sometime in the last 2 minutes.

mines up as well

"""We have our network administrator heading to the data center now and should be there in 20 minutes or so."""

Huh. He could not dial in to login to the router? No emergency DSL line for maintenance?

First Microsoft/Danger/Sidekick, then Apple, now this. We need a computing paradigm that does not suffer from single points of failiure, or at least fails more gracefully.

You've got it. Call IBM and ask them about managed hosting. You can use the Dreamhost prices as a reference, just add 3-5 zeroes on the end.

If you buy a $500 car, never service it, and it breaks down unexpectedly, do you call for a new paradigm of personal transportation, or do you say "rats, well it was cheap and served me well while it lasted"?

Is avoiding 33 minutes of downtime worth tens of thousands of dollars to your business? Didn't think so.

I switched from DH, where I had constant outage problems, to Slicehost and Linode, both of which have never had outages (for me), for only $15 more a month...

That's nice, and $15/month probably buys you a lot over DH -- but it's not like Slicehost is immune:

http://status.slicehost.com/2009/9/22/stl-a-network-issues-2

http://status.slicehost.com/2009/3/27/stl-b-network-issues-0...

I recently switched from DH to linode as well. Best decision I've ever made.

I am not very familiar with Dreamhost, although I've heard of them many times, but if they're big enough, shouldn't they have failover for everything?

But even after all those measures, shit happen sometimes. Then the only thing that matters is how you respond to it and how transparent and honest you are about it.

Are you saying they did have failover set up for everything? I'm just curious. I do agree with your point, though.

Quick fix for such a problem. Dreamhost support is pretty impressive, but sites can be a bit slow.

"We have our network administrator heading to the data center now and should be there in 20 minutes or so. More updates to come."

If they host a million domains, why isn't someone on site 24x7?

Anyone with serious uptime requirements avoids DreamHost. It sounds strange considering their size but it probably just isn't really in their budget. I used to have a DH account and it's pretty good for what it is. Now I have a VPS though.

My guess is they only have low to mid level support 24/7.

My site was slow and support told me that this was due to the long travel the packets have to go cross-atlantic. Yeah.

My site's loading...and websites loading is a good thing.

works fine for me

Shit happens. Even Google has outages (Gmail, anyone?). Life goes on.

Other than search and google.com, the reliability of Google's services is truly awful.

Funny that this happens the day I start looking into a hosted Wordpress service.

Reply on news.ycombinator.com