Is it best to use many Unix Logical Volumes or 1 large

[Tison]

When putting Sybase ASE on a Unix server, is it best to make 1 large Logical Volume (eg 1000K) or better to make many small Local Volumes (eg (500K , 500K) ?

 

[marvin51796]

We have ASE loaded on a HP RP5470 running ver 11.0 of HP-UX, we have created separate volume groups for data,index and logs, that are on separate 18g mirrored hard drives. The data and index LV's are 2g in size, and when they need more disk space i just create another LV and do an alter database, I know that this method was recomended by the Software company and is alot faster than the last time we loaded it use the whole disk method.
Hope this is helpful

 

[sybaseguru]

Sybase database devices best created on 2048MB or below devices. If your O/S is using Veritas Volume manager etc, it will probably use RAID 5 with enough cache so you do not need to worry about split etc. If you are using RAID 1 (striping) then ask your UNIX SA to create 2GB raw partitions for you and stripe them across many disks to give you more spindles. Then you create sybase devices using these partitions. Use alter database to extend the data and/or log segment of your database. If you have two databases and say you have 2 sybase devices namely; device01, device02 each 2GB (very simple example) then you can do:

create database dbase_1 on device01 = 1000 log on device02 = 500

create database dbase_2 on device02 = 1000 log on device01 = 500 etc.

In summary it is best to keep sybase devices to 2GB each. Your UNIX SA should be able to advise you on what he/she is doing

Good luck