NFS Performance Tuning for HP-UX 11.0 and 11i Systems

nfs performance tuning for hp-ux 11.0 and 11i systems page 124

Notes:

Page 124July 22, 2002

• Default “proto” NFS mount option

(11.0 = UDP 11i = TCP)

• Default “rsize” NFS mount option

(11.0 = 8192 11i = 32768)

• Default “wsize” NFS mount option

(11.0 = 8192 11i = 32768)

• /dev/zero file

(11.0 = mknod 11i = included)

• Filesystem Semaphore Contention

drastically reduced in 11i

• Support for large NFS files

(11.0 – 1TB 11i – 2TB)

• NFS Buffer Cache Management

Redesigned in 11i

• CacheFS – only available in 11i

• Kernel Parameter Differences

• Default number of biod daemons

(11.0 = 4 11i = 16)

• Default number of nfsd daemons

(11.0 = 4 11i = 16)

• Support for AutoFS

(11.0 – patch 11i – included)

• Support for NFS/TCP

(11.0 – patch 11i – included)

summary of nfs

differences between

hp-ux 11.0 and 11i

The default behavior of NFS in HP-UX 11i has changed significantly, providing a

much better “out of the box” experience than HP-UX 11.0.

While 11.0 NFS clients and servers can be configured to mimic some of the

behavioral improvements made in 11i (such as rsize, wsize, and TCP support), HP

has no plans to back-port many of the significant 11i performance improvements

(including reduced filesystem semaphore contention, redesigned buffer cache

management routines, and CacheFS) to HP-UX 11.0.