XMLA gurus... we're staggering our monthly partitions across several physical
disks. January goes to the D drive, February goes to the E drive, etc. When I
specify a StorageLocation in the Create partition XMLA command, it doesn't
give me the nice directory structure like if I were to go with the default.
It puts the files at:

D:\Program Files\Microsoft SQL
Server\MSSQL.2\OLAP\Data\323204EC-7F45-4874-B3FC-94428C9AF0EC.0.0.0.0

But if I had stuck with the default StorageLocation (omitted that tag) it
would put the files in a nicer directory structure:

C:\Program Files\Microsoft SQL
Server\MSSQL.2\OLAP\Data\MyDB.0.db\MyCube.0.cub\My MeasureGroup.0.det\MyPartition_2000_01.0.prt

Any suggestions for getting a similarly nice directory structure on the
non-default drive? Any way using AMO to look at a partitions parent objects
and figure out the correct path?

Also, if I specify a StorageLocation path which uses a directory which does
not yet exist, I get an error:

Errors in the metadata manager. The 'D:\Program Files\Microsoft SQL
Server\MSSQL.2\OLAP\Data\MyCustomDir' storage location of the
MyPartition_2000_01 partition does not exist, is too long, or contains
characters that are not valid or are reserved.

Any way to tell SSAS to go ahead and automatically create the directory
instead of complaining? The package that builds the partitions runs on a
separate SSIS server so I'd prefer not to have to remotely create directories.

Thanks a million for any suggestions or tips you can pass along to the
community.

Not sure why the structure of folders plays any role in your case.
What is it you trying to do with it?

As for Analysis Server not creating new folder for you. This is done for
security reasons. Allowing AS user to create new folders on server file
system allows malicious user easily bring down your server.

Edward.
--
This posting is provided "AS IS" with no warranties, and confers no rights.


"FurmanGG" <FurmanGG (AT) discussions (DOT) microsoft.com> wrote

You're correct that everything works fine. And your comment about security
concerns did make a lot of sense.

I just would have preferred that the directory structure have helpful text
descriptions instead of GUIDS. If there was a nice folder structure built
automatically for me no matter what drive I put put the partition on, it
would have been very easy to right click the directory and easily get a sense
for how much storage that particular measure group is taking up on this
drive. As it is, I just have to write a script via AMO to loop through the
partitions and find the ones on that drive and sum up the storage size. Not a
big deal... just a nice-to-have to be able to tell what's on that disk by
looking at the directory names.

Thanks for your help confirming I wasn't overlooking something stupid.

I'm assuming you had a hand in this part of SSAS... thanks a million for
your hard work... the new processing and DDL architecture is just fabulous!
Been very pleased with it compared to SSAS2000.

"Edward Melomed [MSFT]" wrote: