David_Wallis

Anybody done their BCFP or BCSM??

Any books people would recommend?

David

druddle

I have got some presentations from a BFCP fast track training day I did late last year. Will dig them out for you....

Dave

David_Wallis

bump for a reminder

David_Wallis

bump

J4CKO

iTrader: (1)

Anyone any good with multi pathing, we have a Netapps SAN attached to a SQL server and the preformance is hideous, currently moving the db temporarily to local storage. There are loads of MPIO errors and degraded disk messages, any ideas ?

David_Wallis

Yep.. im pretty good with it.. I Use brocade switches & NetApp FAS6030's

Drop me a mail and I'll try and help you out.. david.wallis@callcredit.co.uk

I also have some good contacts at Netapp

druddle

Sorry David, totally forgot. Will send something to you tomorrow.

Jacko - whats the deal with the LUNs ? What sort of filer is it ? Do you snapshot them, if so how is the database files layed out (user DB, system DB, snapinfo, logs) ? I am assuming its Windows, so what version of Snapdrive are you using, and whats the fractional reserve (and also snapshot reserve) on the volumes ? Best practice for layout from Netapp is to try and put user DB and logs in seperate aggregates, which is not easy unless you have lots of disks or a clustered filer.

Also, how many databases are there and what are they (small/large OLTP, data warehouses, etc) ? How many databases are on the data LUNs ? Depending on your I/O pattern you can change one setting on ONTAP - wafl.nt_admin_priv_map_to_root should be set to on if data access is random.

Also do you use Snapmanager for SQL, Snapmirror or anything like that ?

Looking at the SAN side of things, you can either use ONTAP commands or Filerview to check FCP status of the adapters, and you can also use the MPIO snap in to Computer Manager to check device path status. The Brocade will log if theres things like dropped FC packets or collisions so check if any specific ports are reporting that, it might point at a cable, SFP or interface dragging it all down. Also can you tell me the failover mode of the filers if they are a cluster ? They should be set to single_image.

Have a look at NetApp - Library - Search - search for the SQL TRs, theres alot of good info there.

Drop me a PM if you need some more help, I am Netapp certified and have been designing and installing them for a while. Let me know your company name by PM and I can maybe find someone at Netapp that could look at an autosupport if you have a recent one sent to them, or fancy sending one.

Sorry theres so many questions for this time of night....

Dave

druddle

David - YHM.

Dave

David_Wallis

And the best option they dont tell you is "round robin with subset"

David_Wallis

As for the seperate aggr argument, It depends, we have multiple db's on one box for example and the servers can range from a couple of db's to a couple of thousand, sizes range from mb to >4tb.

Depending on what you are doing sometimes having the spindle count is more beneficial.

you can also play with vol options minra and no_atime_update

However these wont cause MPIO errors..

First things,

Run NetApp certified HBA drivers and bios (see now for compatibility matrix)

Check the cables are 52 (IIRC) micron rather than 62.5.

Check port stats on switch and filer for errors..

Drop to one link and see if you have errors, if not try the other..

Check for dodgy SFP's

Reinstall the MPIO / DSM / Snapdrive

ensure you have the hotfixexs, and if running a proliant ensure you have the latest PSP or at least the latest Smart Array 5/6000 or p4/6/800 drivers otherwise it will bluescreen when experiencing high IO throughput.

David