I recently posted an "Ask Slashcode" question on slashcode.com about
creating a Slash cluster with the ability to handle ~million hits a
month from the outset and be in a position to scale well beyond that.
I didn't get any comments, so I thought I'd see if anyone on this list
might have any pointers. Basically the cluster is all x86/FreeBSD
with the following dedicated machines:
2 webservers (Apache)
1 imageserver (Apache or thttpd)
1 dataserver (MySQL)
+ a few other machines for redundancy
The dataserver is a dual-processor 2.x GHz P4 with multiple SCSI
disks. The others are single-processor 1.2 GHz P4 machines with
single IDE system disk.
Is there anything you'd change? What is the minimum/preferred amount
of memory you'd put in these boxes?