Graham Innes
2015-02-12 02:38:57 UTC
Hi all,
I would like to experiment with different SSHFS mount options to see if
we can improve /home performance on fat clients by sacrificing SSH
encryption strength or compression. Where are these options located?
I benchmarked home directory performance on fat clients using a simple
dd read / write test:
Write: ~60 MB/sec
Read: ~40 MB/sec
This is not bad, but I would have hoped for more. Everything is on a
1Gbit network. The fat clients are reasonably modern Intel Ivy Bridge
CPUs (2012) that score around 2,600 on cpubenchmark.net. The servers
have quad core Xeon 1200v3 CPU with RAID1 Hard Disks serving /home. The
server does not experience significantly increased CPU load when the
tests are run on the fat clients.
However, CPU load on the fat client is significantly affected by the
test; it consumes approximately 100% of one CPU core (or around 50% of
both cores) when it runs. This leads me believe that the (single
threaded) SSH encryption load is causing the bottleneck, and maybe it
can be improved by changing the SSHFS mount options.
Thanks!
Graham
------------------------------------------------------------------------------
Dive into the World of Parallel Programming. The Go Parallel Website,
sponsored by Intel and developed in partnership with Slashdot Media, is your
hub for all things parallel software development, from weekly thought
leadership blogs to news, videos, case studies, tutorials and more. Take a
look and join the conversation now. http://goparallel.sourceforge.net/
_____________________________________________________________________
Ltsp-discuss mailing list. To un-subscribe, or change prefs, goto:
https://lists.sourceforge.net/lists/listinfo/ltsp-discuss
For additional LTSP help, try #ltsp channel on irc.freenode.net
I would like to experiment with different SSHFS mount options to see if
we can improve /home performance on fat clients by sacrificing SSH
encryption strength or compression. Where are these options located?
I benchmarked home directory performance on fat clients using a simple
dd read / write test:
Write: ~60 MB/sec
Read: ~40 MB/sec
This is not bad, but I would have hoped for more. Everything is on a
1Gbit network. The fat clients are reasonably modern Intel Ivy Bridge
CPUs (2012) that score around 2,600 on cpubenchmark.net. The servers
have quad core Xeon 1200v3 CPU with RAID1 Hard Disks serving /home. The
server does not experience significantly increased CPU load when the
tests are run on the fat clients.
However, CPU load on the fat client is significantly affected by the
test; it consumes approximately 100% of one CPU core (or around 50% of
both cores) when it runs. This leads me believe that the (single
threaded) SSH encryption load is causing the bottleneck, and maybe it
can be improved by changing the SSHFS mount options.
Thanks!
Graham
------------------------------------------------------------------------------
Dive into the World of Parallel Programming. The Go Parallel Website,
sponsored by Intel and developed in partnership with Slashdot Media, is your
hub for all things parallel software development, from weekly thought
leadership blogs to news, videos, case studies, tutorials and more. Take a
look and join the conversation now. http://goparallel.sourceforge.net/
_____________________________________________________________________
Ltsp-discuss mailing list. To un-subscribe, or change prefs, goto:
https://lists.sourceforge.net/lists/listinfo/ltsp-discuss
For additional LTSP help, try #ltsp channel on irc.freenode.net