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

nfs performance tuning for hp-ux 11.0 and 11i systems page 43
Notes:
Page 43July 22, 2002
Copyright 2002 Hewlett- Packard Company
nfsd
How do they work in the READ case?
1. The nfsd checks the server’s buffer cache for the requested data. If
data is present it’s read from cache and returned to the client.
2. If data is not in the cache the nfsd schedules a READ call to the
underlying filesystem (VxFS, HFS, CDFS, etc.) and sleeps waiting for
the data.
3. When the data arrives it is placed in server’s buffer cache. The
sleeping nfsd process is awoken and sends the data to the client.
GOAL: Retrieve the data requested by the NFS clients from the
server’s buffer cache or physical disks as quickly as possible
When a client system needs to read data from an NFS mounted file, it will first
check its local buffer cache to see if the block of data is already present in cache.
If it is then the read is satisfied without generating an NFS request. If the data is
not present then an NFS READ call will be made to retrieve this data.
When the server receives the READ request it first checks it’s buffer cache to see if
the requested data is already present in it’s cache. If so, it sends the data back to
the client without performing a physical disk I/O.
If the requested data is not present in the server’s buffer cache, the nfsd schedules a
READ from the underlying filesystem (i.e. VxFS, HFS, CDFS, etc.) and blocks
waiting for the disk subsystem to retrieve the data. When the data is returned from
the filesystem it is placed in server’s buffer cache. The sleeping nfsd process is
notified that the data is available. The nfsd retrieves the data from cache and
sends it to the waiting NFS client.