Community technical support mailing list was retired 2010 and replaced with a professional technical support team. For assistance please contact: Pre-sales Technical support via email to sales@march-hare.com.
On Thu, 6 Feb 2003 14:15:36 +0100, "Kim Hansen" <kha at mita-teknik.com> wrote: >I can (sort of) figure out how to fulfill req. 1 and 2 (as indicated in = >the parenthesis), but I am uncertain about 3. I have toyed around a = >little bit with our existing repository (700 MB) on an (ordinary) old = >400 MHz Pentium, and it is my feeling that the load on the server is = >never very high for such a repository and with a single user accessing = >the repository at a time. However, I am very uncertain what happens if = >the repository gets very large with perhaps 3-4 concurrent users (and = >where some of the ,v binaries can have sizes of houndreds of MB).=20 Basically the potential bottlenecks in cvs are disk access and memory. For memory, it's so cheap then 'as much as possible' is all I can say. Disk access will be helped with a decent RAID setup. cvs isn't really very CPU intensive but if you're putting together a nice machine then there's no point it sticking a 486 in it :-) >My feeling is that network bandwidth may the real show-stopper in these = >cases (we have 100 Mbit LAN)? The protocol is actually quite efficient. We have ~30 developers over a 100mbit LAN and have never really seen much slowdown even during peak times. If you think your bandwidth is a problem, enable compression on the clients and it reduces the bandwidth requirement at the expense of more CPU usage on the server and clients. Tony