|
For info, and for people
> [ Sorry for the late reply here - I started writing this reply a few
> days ago, but I've been busy continuing to tweak, monitor and add
> servers into our Pilot LVS based service... (Currently standing at
> 7 servers, combined capacity 350Gb storage, servicing 52 Universities :-) ]
Looking at the stats of the 7 real servers now, + director, and
comparing it to our total traffic load, these 7 servers are now handling
approximate half our traffic - with 47Mbit/s real server traffic during
peak hours, or 53Mbit/s if you include the traffic through the director.
(The traffic through the director though isn't just director traffic since
we're chopping ICP requests off at that point for various reasons)
Put another way this is half our normal service throughput with just one
director, and using about 1/4 of our servers.
Despite the fact that the machines behind the director are "running hot",
the people using this service so far seem to be very happy, with some
commenting that the service is like it is in summer time. (When our
loading is at a minimum, and currently we're at maximum loading)
I've got some leave coming shortly so I'll be documenting all the gotcha's
I encountered in setting up a cache cluster like this since there's
probably common problems in other protocols like this. (specifically the
issues with UDP services were "intriguing" to sort out)
In short, though, the technology works well.
Michael.
--
National & Local Web Cache Support R: G117
Manchester Computing T: 0161 275 7195
University of Manchester F: 0161 275 6040
Manchester UK M13 9PL M: Michael.Sparks@xxxxxxxxxxxxxxx
----------------------------------------------------------------------
LinuxVirtualServer.org mailing list - lvs-users@xxxxxxxxxxxxxxxxxxxxxx
To unsubscribe, e-mail: lvs-users-unsubscribe@xxxxxxxxxxxxxxxxxxxxxx
For additional commands, e-mail: lvs-users-help@xxxxxxxxxxxxxxxxxxxxxx
|
