On Mon, Sep 7, 2009 at 12:29 AM, Steve Totaro<stot...@first-notification.com> wrote: > Did you push it past 300 on two year old hardware and software?
old hardware yes. old software no. The servers are more than 3 years old >Core 2 Duo Dell Dimension desktop as proof of concept? are core 2 duo's really two years old already? I guess so. I don't really follow the latest hardware news. I have my lab on server-class gear. > Port mirroring is basic on almost any newer switch. Login, enable port > monitoring, write mem, done. Port mirroring is basic on quality networking gear. I know perfectly well how it works. My point was that replicating ALL traffic on a LAN port seemed a bit like hauling out all the corn plants from the corn field when what you really wanted was just the corn kernels from the ears. That's what I mean by heavy-handed. I've never used the software you've proposed. I realize that replicating all traffic for a port, or in my case, all traffic for a bonded interface is not difficult logically, and is quick to configure. I think it is aesthetically displeasing compared to grabbing the recordings at the place where the calls are already taking place. Personal taste. You're allowed your opinion too, which you've clearly stated. > I build robust and redundant systems, separate server for DB, recording, > gateways, in an all HA configuration. Me too. Again, taste. > Again, how many calls were you able record using RAMdisk? Anywhere 300? As I stated before, this is going to be dependent on how you're manipulating the calls and the gear you're running on. The nice thing about your 'just broadcast the entire LAN to the recording solution' is that the recording service just gets to throw away everything that's not an audio channel, and it doesn't have to do squat to the call. If it COULDN'T do a lot of recordings under these circumstances it wouldn't be worth any money. I don't think I've pushed my solution past 90 simultaneous recordings of MeetMe() mixing, with more than 100 AGI channels running, with assorted ChanSpy() jobs. > Bookmark my post, so when you reach your RAMDisk limit, you can join the big > league. Anything I do as a scaling solution will be price versus performance. So since we're talking about a commercial solution to replace something that asterisk does, I'll have to find out what your commercial solution costs per channel, and compare that against the cost of cloning out an identical server. My solution scales to parallel servers just fine. Is OrecX really $199 per recorded channel? So that 300 channels you're talking about costs $60,000? So I can buy six $10,000 servers, each of which can run circles around my current solution, and still break even. I like my solution better. _______________________________________________ -- Bandwidth and Colocation Provided by http://www.api-digital.com -- AstriCon 2009 - October 13 - 15 Phoenix, Arizona Register Now: http://www.astricon.net asterisk-users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users