[Info-vax] DIRCACHE hit rate.

[Hein RMS van den Heuvel]

On Jan 20, 12:08 pm, Jan-Erik Söderholm <jan-erik.soderh... at telia.com>
wrote:
> Hi.
> I have an 7.3-2 Alpha system where I think that
> the hit rate on the DIR-cache is way to low.
>
> DS20 with 9GB disks in BA356 shelfs.
>
> Here is what it typicaly looks like :
>
>                         CUR     AVE   MIN      MAX
>   Dir Data (Hit %)     3.00    3.20  0.00    30.00
>       (Attempt Rate) 312.00  548.90  0.00  2460.00

That seems supspiciously low.
This statistics is across all drives.
Are there MD (DECram) disks in play?
Are there any disks mounted /nocache ?

If there is such a thing a 'typical' batch job in your environment
then I would
want to run one with SET WATCH FILE/CLA=MAJOR or /CLA=ALL to 'see'
whether any surprises access are attempted.

I also find the SDA extention 'LNM' very 'handy' in tracing file
(attempted) file access.

> Another thing is that all files are timestamped in the filename, so there are normaly only ;1 files.

Not sure how that is good or bad.
Multi-version fiels are MORE efficient that many single-version files,
from a directory space perspective. Can't get any thighter then an
entry in an 8 byte entry to store a fresh version + file-id.

Is the application using a 'nice' timestamp like a leading
yyyymmddhhmm allowing for adding/activity to the end only, or leadin-
names or dd-mmm-yyyy stamps leading to random inserts?

Are temporary files alphabetically mixed with main files in the same
directory?
Ever tried a dedicated (pre-allocated) directory for the temporary
files?
There are multiple disks... are they all being used by all batch jobs
or is it the typicaly silly setup with a directory(s) on a single disk
for each user/application/batch-job?

What tools is creating the (temp) files? DCL $OPEN/WRITE, C create?
The C-RTL adds a lookup to see whether a file with that name exists to
allow attribute inheritance. (nasty feature imho).

> Now, what I'd like to ask, is if anyone knows if
> this part of VMS has had any major improvments in
> the 8.x versions ? We are currently thinking of
> upgrading anyway, and it would be nice to know
> if these hit rates would improve simply by upgrading.

I doubt it, but would certainly encourage going to 8.3 (skip 8.2!).
Lot's of nice stuff there.

Hein.