Statistics (for server)

[David Collier-Brown]

Peter Polkinghorne wrote:
> it is hard to know what your Samba server is doing.
> But I have no measure of how active such clients are.  But what 
> would be good measures?
> 
> I can think of Data transfered, packets exchanged and may be counts of
> operations by type - but that might be too large for SMB (cf nfsstat).

	SMBD is a classical server, which is also dependent
	on other servers.

	The interesting values are
		1) number of connections
		2) latency of a read/write request,
		   from beginning of request to beginning
		   of data transfer
		3) throughput of a read/write request, from
		   the beginning to the end of the transfer
		   only.
		4) latency and throughput of the underlying
		   read/write request, the one that usually
		   is a read(2) from a disk.

	After this come a whole bunch of things which 
	tell you queue backups, latency of individual
	operations and the rest of the nfsstat-like stuff.

	For a performance guru, the first four will get them 
	started, and the rest of the information they'll need
	initially is discoverable from system stats.

	A caveat: all of these are easy to collect, but
	are both hard to aggregate and expensive to printf
	individually to the logs. It makes sense to write 
	a little stats module, and a program to probe it
	and capture/format the data.  That way you can 
	collect the important things and defer doing the
	other stuff until you need to measure an individual
	thing.

--dave (who has ended up a performance analyst several times,
	entirely by accident) c-b
-- 
David Collier-Brown,  | Always do right. This will gratify some people
185 Ellerslie Ave.,   | and astonish the rest.        -- Mark Twain
Willowdale, Ontario   | //www.oreilly.com/catalog/samba/author.html
Work: (905) 415-2849 Home: (416) 223-8968 Email: davecb at canada.sun.com