| asteriskguy | I'd be happy if I can just restore service with in seconds or minutes. |
| sweeper | kavit: openser, and rtp redirect~ maybe rtp proxy as well |
| JT | kavit: you need 2 systems parallel processing everything |
| kavit | JT: i thought that would be the case |
| sweeper | although that won't really solve the problem because the openser server could go down... |
| SwK | you know they do make lock-step CPU servers that solve all that problem |
| sweeper | but maybe you could do some voodoo |
| kavit | hrm |
| SwK | so if one cpu goes out the other just takes over |
| kavit | Cisco refused to support it with callmanager |
| JT | i'm aware telcos use them |
| SwK | yep |
| JT | but that only protects against cpu failure |
| sweeper | is cpu death really that common? :/ |
| SwK | jt: actually it protects again CPU, ram and several other failures |
| sweeper | I think you'll find RAID controller death, and RAM death to be the biggest killers |
| kavit | Sweeper: disaster recovery isnt about it actually happening.... it is about the possibility of it happening |
| SwK | however real telco equipment just detects a failure and moves it to a different media path |
| sweeper | well, power first, but that's easy |
| JT | SwK: sure, but i assume the systems are right next to each other, what if the place catches fire? |
| sweeper | kavit: how often is the cpu gonna die and it not be the whole server? |
| SwK | jt: theres full redundany, and then theres reasonable redundancy |
| sweeper | SwK: multi-home/multi-NOC is well within reasonable |
| kavit | Sweeper: sure but people have money to burn... so why not |
| SwK | its like security... sure I can make build a system thats completely impossible to hack, but it would be completely useless |
| JT | Sweeper: on some telco equipment, 1 cpu card handles tens of thousands of concurrent calls, they can't afford it to go down |
| sweeper | kavit: because you might as well spend that money on something that will actually be effective? |
| kavit | I have worked with Nortel VPS/MPS and Oscar Systems |
| SwK | sweeper: do you realize what a single DS3 port on real telco hardware costs? |
| kavit | Sweeper: they work very well |
| JT | kavit: the financial loss from a major telco switch being down can be over $50k/minute |
| sweeper | SwK: we're talking about asterisk here :P |
| SwK | yep asterisk != carrier grade |
| JT | yeah |
| SwK | asterisk is a PBX |
| sweeper | but we're discussing making asterisk moar reliable, are we not? |