3 ms·

Quick summary: They have a server shortage. They are adding new servers but this takes time (three days, apparently). Also, they're disabling Cheetah speed. If

Quick summary: They have a server shortage. They are adding new servers but this takes time (three days, apparently). Also, they're disabling Cheetah speed. If only there was a way to start virtual server in less than three days...

ZoFreX 14y ago
They're using EC2, so I'm not sure why it takes them so long to add new servers.
- pan69 14y ago
  Just because you're using EC2 (or any other "cloud" platform) doesn't mean you can just simply add new servers to your infrastructure, your application architecture needs to cater for this. With EC2 you have the "ability" to architecture your application in such a way that you can take advantage of this but if you don't then you have to do everything manual. Cloud is not a silver bullet.
  - kelnos 14y ago
    If they're truly using EC2, and they haven't architected their infrastructure to be able to horizontally scale at the push of a button, then they're incompetent. Period. It's forgivable that it doesn't happen automatically, as automatic provisioning in response to load isn't always an easy problem to solve, but if it's going to take them 3 days to bring up new machines on EC2 to handle load, they're doing it wrong. The ability to do this sort of thing quickly is one of the top few reasons to use something like EC2.
    - rainsford 14y ago
      It does seem safe to say that technically they really dropped the ball. The inadequate estimation of server load combined with the (apparently difficult) task of adjusting to the actual load suggests at least a few groups of people didn't do their jobs properly (or the company denied them the resources to do so).
    - gte910h 14y ago
      They very likely DID do something to bring up servers quickly. I bet they can spin up stuff like NA East 3 rather fast. Just because there is a DB somewhere who can't handle 2304982309482 people trying to slam cheetah mode while 230492340 are editing trade deopots, which is just a test they didn't happen to simulate well enough in load testing, doesn't mean they're utterly incompetent. Server outside EA are pretty new to EA from what I understand The weirdest crap melts sometimes when you apply real numbers of real non-tester peoples to things. Weird crap you cannot test for. Often has little to do with raw power, but instead is small architectural inefficiencies you cannot easily fix.
  - ZoFreX 14y ago
    Yes, I realise that, I just would have ordinarily assumed that a choice to use EC2 also included consideration for scalability, automatic spin up, etc.
- gte910h 14y ago
  They're not necessarily using it for what the bottleneck is. They might do raw compute there, but are locked on a DB query or write that's hosted internally or that just is taking awhile to properly partition.

ZoFreX 14y ago

They're using EC2, so I'm not sure why it takes them so long to add new servers.

pan69 14y ago
Just because you're using EC2 (or any other "cloud" platform) doesn't mean you can just simply add new servers to your infrastructure, your application architecture needs to cater for this. With EC2 you have the "ability" to architecture your application in such a way that you can take advantage of this but if you don't then you have to do everything manual. Cloud is not a silver bullet.
- kelnos 14y ago
  If they're truly using EC2, and they haven't architected their infrastructure to be able to horizontally scale at the push of a button, then they're incompetent. Period. It's forgivable that it doesn't happen automatically, as automatic provisioning in response to load isn't always an easy problem to solve, but if it's going to take them 3 days to bring up new machines on EC2 to handle load, they're doing it wrong. The ability to do this sort of thing quickly is one of the top few reasons to use something like EC2.
  - rainsford 14y ago
    It does seem safe to say that technically they really dropped the ball. The inadequate estimation of server load combined with the (apparently difficult) task of adjusting to the actual load suggests at least a few groups of people didn't do their jobs properly (or the company denied them the resources to do so).
  - gte910h 14y ago
    They very likely DID do something to bring up servers quickly. I bet they can spin up stuff like NA East 3 rather fast. Just because there is a DB somewhere who can't handle 2304982309482 people trying to slam cheetah mode while 230492340 are editing trade deopots, which is just a test they didn't happen to simulate well enough in load testing, doesn't mean they're utterly incompetent. Server outside EA are pretty new to EA from what I understand The weirdest crap melts sometimes when you apply real numbers of real non-tester peoples to things. Weird crap you cannot test for. Often has little to do with raw power, but instead is small architectural inefficiencies you cannot easily fix.
- ZoFreX 14y ago
  Yes, I realise that, I just would have ordinarily assumed that a choice to use EC2 also included consideration for scalability, automatic spin up, etc.
gte910h 14y ago
They're not necessarily using it for what the bottleneck is. They might do raw compute there, but are locked on a DB query or write that's hosted internally or that just is taking awhile to properly partition.