SAP - on Windos with SQL DB

[former_member296156]

Can any one suggest

Is better to have multiple arrays or multiple partitions within the arrays?

while laying the SQL DB in SAN

following two case will help understand the issue in a better way

Lets assume - the database looks like -- DB1 , DB2, DB3 , DBlog

Case 1

Single Raid 5 array - with 4 LUN as separate drive letter for all each DB and log

Case 2

One Raid 1 & 3 Raid 5 array as 4 individual drive for each DB & Log file

Which option will give better result - consider all other param like FC / SAS / Disk RPM / Size etc same for both the case

Thanks

Arindam

[markus_doehr2]

Having separate LUNs enable you (depending on your SAN) to configuring caching separately. I would go for this way.

Markus

[former_member296156]

thanks for the quick tips , just like to know - if I am going for a LUN modle - that is Single Array with multiple LUN as logical drive - Will disk size play any role ?

like if the total requirement is 2TB - can a RIAD 5 consist of 76GB each will give better performance than a raid 5 consist of 300GB each

or it will not differ much u2013 it itu2019s mater ........in what degree ?

[markus_doehr2]

> like if the total requirement is 2TB - can a RIAD 5 consist of 76GB each will give better performance than a raid 5 consist of 300GB each

> or it will not differ much u2013 it itu2019s mater ........in what degree

The speed of the database is mainly determined by the number of spindles. To enable an optimal parallelism I would create as many data files as there are spindles/disks so for each available physical disk in the SAN one datafile.

Markus

[former_member296156]

Let me explain the situation Iu2019m facing now

I have a storage (SAN) of 300GBX7 SAS disk in Raid 5 array , total array is divided in to 2 LUN

LUN 01 350GB for Data backup , LUN02 u2013 390 GB for SQL DB 1 , DB 2, DB 3, and the log file

We are facing a terrible performance issue u2013 though Disk write Queue is less than 3 but disk read queue is always more than 70 u2013 sometime 100 , and OS06 is showing 100% utilization in most of the time - and finally users are getting very slow response

Can you please through some lights how to overcome the situation u2013 I want to disk read / write queue to come closure to zero.

[markus_doehr2]

If there´s more than one database on the SAN it´s difficult to find out the cause.

I would use the SAN tools to find out, which of those instances is generating so much I/O. If you backup to the same SAN this may also trash the SAN cache and so slowing down the system.

Markus