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

nfs performance tuning for hp-ux 11.0 and 11i systems page 79

Notes:

Page 79July 22, 2002

cachefs

How does CacheFS work?

• The cfsadmin(1M) command creates a cache on the

client in a local filesystem (referred to as the front

filesystem)

• An NFS filesystem (referred to as the back filesystem) is

mounted referencing the cache directory

• During an NFS read the “front” filesystem is checked. If

the data is resident the request is resolved locally. If not,

it’s retrieved from “back” filesystem and added to cache

• Pools of cachefsd kernel threads are dynamically

launched to manage the cache contents

The cfsadmin(1M) command is used to create the local cache directory and define

the parameters of the cache. By default, the cache is allowed to consume up to

90% of the available disk space of the filesystem in which it resides.

The CacheFS file system caches data read from the NFS filesystem (known as the

"back" filesystem) onto another, local filesystem (known as the "front" filesystem).

A separate pool of cachefsd threads is created to service each cache and each

pool can have a maximum of 5 cachefsd threads associated with it at any time.

The kernel launches new threads as requests for cached data increase.