|
abhijeet wrote:
>
> * requirement specific load balancing
> i.e. study how different load balancing algorithms address different
> requirements and hence arrive at a algorithm to requirement
> correlation.
this is not a matter of great urgency as far as I can tell. Under most
situations, the current schedulers give good enough results.
> * geographic sensed load balancing.
There is an implementation of this already, although because
of problems getting someone to give you the routing information,
it's not widely deployed.
http://www.linuxvirtualserver.org/Joseph.Mack/HOWTO/LVS-HOWTO-25.html
> * IPv6 and repercussions for lvs.
I'm not sure how soon IPv6 is going to be a requirement for anyone.
People have been talking about it for a long time.
> However we have very outdated hardware on our campus (just 3-4 P-I's at our
> disposal) which makes performance evaluation or benchmarking useless.(??)
not at all. My test realservers are 75MHz pentium classics. With 100Mbps
ethernet, all this means is that I have a very fast network relative to
my CPU speed. In this respect I have a similar setup to people with 750Mhz
machines with GigE ethernet cards.
The time for the tests (to accumulate a certain number of clock cycles say) will
be proportionately longer and your tests may take 1hr instead of 6 mins,
but in a 6month project, the testing time will only be a small part
of your time budget.
> To add to that we have only about the next 5 months in hand to complete the
> project.
The real thing is that you need a project that will be
o interesting to you
o of interest to potential employers when they look at your resume
o (possibly) of interest to users here, so you can point to people have used
your results
o oh, and satisfy your course work too.
> We need help in locating feasible problem definitions in the area.
This is the free software world. Here you have to think of your own
projects. There really aren't great todo lists despite what the HOWTO
says. Most times someone comes up with an idea that no-one had thought
of, that we can't do without once we know about it.
Is this a big project - ie do you have 10hr/week for 5 months?
Can you code? Are you prepared to scrutinize other people's uncommented
inscrutable code to work out what they hell they were trying to do?
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
|
