What's new
By:
  • Visit Rebornbuddy
  • Visit Panda Profiles
  • Visit LLamamMagic
  • Visit Resources
  • Visit Downloads
  • Visit Portal

Questions about the HB Backend

Status
Not open for further replies.
E

evilz51

New Member
Joined
Aug 22, 2010
Messages
40
Reaction score
0
Gentlemen,

I know you wont be able to give details, nor would I want the type of details that could compromise your environment, but with all the issues lately it has me curious, What is your infrastructure like? Are your servers virtualized? What type of load balancer are you running (commercial or like a linux based application)?

My last question is about your capacity planning. do you do it? I understand its been trying times, and I am not trying to call you guys out at all, just as an IT person myself I am curious as to how the backend is run, and maybe gleam a little insight into where the failure in the system occurred that allowed this type of extended degraded operation period.

Thanks guys, I do love the product overall! and thanks for the hard work!
 
evilz51 said:
Gentlemen,

I know you wont be able to give details, nor would I want the type of details that could compromise your environment, but with all the issues lately it has me curious, What is your infrastructure like? Are your servers virtualized? What type of load balancer are you running (commercial or like a linux based application)?

My last question is about your capacity planning. do you do it? I understand its been trying times, and I am not trying to call you guys out at all, just as an IT person myself I am curious as to how the backend is run, and maybe gleam a little insight into where the failure in the system occurred that allowed this type of extended degraded operation period.

Thanks guys, I do love the product overall! and thanks for the hard work!
Click to expand...

That would be interesting info. I can tell you they are using iis7 :(
 
I believe for the security of their systems its best if they do not provide actual details of the services and the like they have in place, however I'm pretty sure they'd be happy to answer the rest
 
evilz51 said:
Gentlemen,

I know you wont be able to give details, nor would I want the type of details that could compromise your environment, but with all the issues lately it has me curious, What is your infrastructure like? Are your servers virtualized? What type of load balancer are you running (commercial or like a linux based application)?

My last question is about your capacity planning. do you do it? I understand its been trying times, and I am not trying to call you guys out at all, just as an IT person myself I am curious as to how the backend is run, and maybe gleam a little insight into where the failure in the system occurred that allowed this type of extended degraded operation period.

Thanks guys, I do love the product overall! and thanks for the hard work!
Click to expand...

The current system is under a pseudo-load-balancing system. It's written to failover to another server in the event that one is down (or just not accepting a new request). None of our servers are virtualized, as we use every single bit of resources on the server to run our auth servers.

We do plan for the traffic we get, however, these are only guesses on how much we may get. None of us expected the amount of traffic we had when MoP's pre-patch hit. (It effectively tripled our traffic for a week, which our servers struggled to handle, but somehow did handle)

I'll be posting a more in-depth explanation of what's been going on lately. But that's a very general overview of what we do (which is changing very very soon) and what happened.
 
Apoc said:
The current system is under a pseudo-load-balancing system. It's written to failover to another server in the event that one is down (or just not accepting a new request). None of our servers are virtualized, as we use every single bit of resources on the server to run our auth servers.

We do plan for the traffic we get, however, these are only guesses on how much we may get. None of us expected the amount of traffic we had when MoP's pre-patch hit. (It effectively tripled our traffic for a week, which our servers struggled to handle, but somehow did handle)

I'll be posting a more in-depth explanation of what's been going on lately. But that's a very general overview of what we do (which is changing very very soon) and what happened.
Click to expand...


Thank you Apoc.

The Virtualization idea whould help you out even if you used 100% of the resources on the server. The way a hypervisor would spin up a new server would be of great use to you guys in my opinoin, especially if you work it out with your colo facility to allow you to build and tear down servers in their virtual environment as needed (instantly increasing and decreasing your server count based on load).

as for the load balancer, I can only say that I think a more advanced device like an F5 (even used) would be of TREMENDOUS use to you, allowing you to actually balance the load across servers, not just fail to 1 or the other

just some ideas.. i used to work in online education where we had massive swings in user load, and this is how we handled it.
 
Great info, I suggest posting a thread with the specifics soon as people are wanting to know what exactly was happening with the servers.
 
Status
Not open for further replies.
Back
Top