Re: Multi-head NFS/Gluster

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On 13.01.2014 01:54, Paul Robert Marino wrote:
I have to say its not always a good option and honestly though it may
seem to work on a dumb switch it really doesn't very well at all in
fact a dumb switch will make it worse. The one piece of advice I can
give you is if you do it max out the receive buffers only not the send
buffer on every network card (not the switch) or your re-transmit
numbers may eat more bandwidth than you gain plus the TCP window
scaling issues will compound it.
I spent several weeks trapped in a room doing low level packet
analysis with a group of people because some one used round robin mode
or as it is also called switch assisted load balancing incorrectly and
were killing our core switches. If you really absolutely need the
speed spend the money on 10Gb network but I doubt it will fix your
problem. If I had to take a guess your bandwidth limitation most
likely lies else where.

Thanks, Paul. My problem is the lack of 10 Gbps right now, so I am trying to make up for it with these weird ideas. Maybe the best way to do it is just wait for 10 Gbps budget.

Cheers :)

--
Sent from the Delta quadrant using Borg technology!

Nux!
www.nux.ro
_______________________________________________
Gluster-users mailing list
Gluster-users@xxxxxxxxxxx
http://supercolony.gluster.org/mailman/listinfo/gluster-users




[Index of Archives]     [Gluster Development]     [Linux Filesytems Development]     [Linux ARM Kernel]     [Linux ARM]     [Linux Omap]     [Fedora ARM]     [IETF Annouce]     [Bugtraq]     [Linux OMAP]     [Linux MIPS]     [eCos]     [Asterisk Internet PBX]     [Linux API]

  Powered by Linux