Hi Hut. We use a slightly different method to scale James onto multiple servers. Two QMail servers work as a pair (we've used both DNS load balancing using MX records or a load balancer) and then having the QMail boxes deliver to a set of James servers. The James servers then use the QMail boxes as gateways as QMail is better at delivering mail to crappy servers than James at the moment (don't get me started on AOL). This also offloads the DNS processing from the James servers. As far as figures go we never really thrashed this system so I don't have any hard evidence on how good an approach this is.
As a piece of anecdotal evidence we had a HP Proliant DL580 (4 x 2.4 GHz Xenons with 2 stripped disks, mirrored). Due a mistake by yours truly I managed to cause a loop whereby our MLM software delivered just over 500k messages to itself in an hour. Believe it or not, our customer was very impressed! Our standard leak test for new versions of our code & James involves transmitting 1.5 million messages over about 5 days to another server. This is using a 400 Mhz PII as the James server, an identical server using James to receive the mails and an 133Mhz Pentium to send the mails. As a company we have no problems in using James as the core of our MLM system. It works, day in, day out. :) Hope this helps. -- Jason -----Original Message----- From: Hut Carspecken [mailto:[EMAIL PROTECTED] Sent: 16 September 2003 21:06 To: [EMAIL PROTECTED] Subject: Scaling JAMES onto More Than One Server Good Afternoon, I am familar with Noel's capacity numbers; however, I was wondering if anyone has experience scaling JAMES to run on multiple servers or to off load some of it functionality to another server. I am looking at handling several hundred thousand messages a day. Hut Carspecken 972-371-5638 --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
