devclass

ajd

Active Newcomer
Joined
Nov 16, 2010
Messages
5
Reaction score
0
Points
0
Location
Near Philadelphia, USA
I've been trying to understand devclass.
Why does it exist if it just points to a predefined devtype?
Why not just eliminate it and point to the devtype directly?

Here's my theory.

We define a library.
Then we define a devclass in the library that points to a
recognized devtype (LTO, 8mm, etc.)
Then we define storage pools with a device class.

So apparently devclass
1) associates storage pool and library and
2) prevents the library being one devtype and its storage class being
another. That is, without devclass we'd have to specify the devtype
of library and the devtype of the storage pool,
and the two devtypes might not agree.

Comments? Corrections? Additions?
 
There is no "storage class"
What you do if you have two libraries of same dev type? (eg. LTO)
You need two devclasses.
 
devclass tells TSM how it should be using the device as well. For example, I've had a LTO-4 library configured with two device classes, one set to do drive encryption and one to just write as normal. There would be no way to make this distinction if the library was defined as simply "LTO".
 
Back
Top