I've recently been having a bit of fun(?) solving a problem where one of
our ZFS clusters was allowed to fill all the space allocated to it.

Along the way I (re)discovered in "Distributed File Service zSeries File
System Administration" (SC24-5989):

<quote>
...Assuming that each volume is a 3390 with 3338 cylinders (with 3336
cylinders free), that there are 15 tracks per cylinder and that you can get
6 8K blocks per track (15 x 6 = 90 8K blocks per cylinder), you should get
90 x 3336 = 300240 8K blocks per volume and 10 x 300240 = 3002400 8K blocks
in the aggregate...
</quote>


My question is: "Is there any way to change that 8K block size up to
something larger?"

When defining the linear cluster in the first place, can I, for instance,
specify a CISIZE of, say, 27648 bytes - without being overridden by some
in-built limitation?


The reason I ask is that with 8K blocks, you - as the manual states - get 6
blocks per track. With a track size of 56664 bytes, this means that you use
49152 bytes on each track. In other words, 13% of the allocated disk space
is quite simply being thrown away.
Yes, yes, yes - I _know_ that "disk space is cheap" these days but 13%
waste still strikes me as a bit much to have to swallow. If I could bring
it down to less than 5% or so, I'd feel a lot happier.

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to