Fremont Data Center

Fremont hasn't had trouble for a while now - but given it's past history of having bad patches, it's probably going to be less reliable than the other datacenters.

I still have servers there for latency reasons though - so it really depends on your priorities.

Thanks for your input.

@jords:

but given it's past history of having bad patches, it's probably going to be less reliable than the other datacenters.

So past hardware failures and power outages can predict future performance?

Any tips on the future powerball numbers?

Yes. A datacenter that has had as many issues as Fremont has, over several different periods, most likely has lax backup and testing measures for their systems. it's highly unlikely based on its performance that their power system is well managed, for example.

The idea that past performance can't predict future performance only applies if the failures are truly random and there is no difference between fremont and the standards at the other datacenters, which is probably not the case.

What exactly does this have to do with lotto? Unlike datacenters, that is a truly random (or at least highly pseudorandom) system.

it's = it is

its = possessive.

All datacenters have problems. Every single one. And they will continue to.

If they do things anything like how we do things, any new problem that gets a solution just makes us better, stronger, faster.

Don't sweat Fremont. Don't jump ship to another DC just because one had a problem. You'll be chasing your own tail for a long time.

-Chris

Fixed :) - My grammar standards have slipped dramatically over the last few years!

That's true, which is one of the reasons I stayed there. But, I do think my point still stands - while it's true that all datacenters have problems, it's not unreasonable to theorize that a datacenter that has been comparatively problematic in the past is more likely to be comparatively problematic in the future.

This is all quite theoretical though, I'm not saying that fremont will have problems or that other datacenters won't, just that if I was picking a datacenter and my only priority was reliablity, I'd pick the one that's had the least serious problems.

@caker:

All datacenters have problems. Every single one. And they will continue to.

If they do things anything like how we do things, any new problem that gets a solution just makes us better, stronger, faster.

Thing is, linode is a small company that's growing massively and learning new things all the time. You invented a large chunk of stuff because you were leading the field.

Datacenter operations, on the other hand, should be a BAU process for a company like HE. To paraphrase Ian Flemming (Goldfinger), once is happenstance; twice is coincidence; three times is incompetence. The sheer number of failures at the same datacenter worried me. After the fourth of fifth outage I eventually requested a move to Dallas. (A totally painless process, I must say!)

I hope the DCops team in Fremont have been replaced and they've now got a competent staff!

@sweh:

After the fourth of fifth outage I eventually requested a move to Dallas. (A totally painless process, I must say!)

I hope the DCops team in Fremont have been replaced and they've now got a competent staff!

Unfortunately it's not completely painless. You still have to change your IP and name servers, which can be a PITA. I suffered through all of the Fremont issues in the past year and I was this close at one point to moving to Dallas, but I didn't because of the hassle.

So far Fremont has been pretty stable for the past six months (knock on wood). The next event will probably push me over the pain threshold and make me move to Dallas…

thanks,

bruce