|
Jeffrey A Schoolcraft wrote:
>
> I'm curious if there are any known DR LVS bottlenecks?
None of the people doing developement have access to high throughput
devices like you have. On 100Mbps ethernet, the director is just
coasting for any reasonable CPU speed. One previous posting from
a person using Gb ethernet found that the director went quite soggy
although I forget the specific problems. There is a set of suggestions
for optimising performance of squid servers on one of the squid pages
(mainly changing a lot of /proc entries), and this person had tried
all of these. This had improved performance noticably, but still
the director was a bottleneck at 1Gbps.
It sounds like you had two LVSs, one with 100Mbps (which had troubles
at 40k connections) and another with 1Gbps.
We should be able to test 40k connections with something like
polygraph (which I don't think anyone has done), but we won't be
able to test the 1Gbps right now.
This is new territory I'm afraid.
Joe
--
Joseph Mack PhD, Senior Systems Engineer, Lockheed Martin
contractor to the National Environmental Supercomputer Center,
mailto:mack.joseph@xxxxxxx ph# 919-541-0007, RTP, NC, USA
|
