Reporting true size of dataset on tape

andy deiss · New User Joined: 23 Sep 2011 Posts: 3 Location: usa

Inquiry on any IBM or other supported tools to report on the true size of a dataset dumped to tape. For example, a dataset of a particular size gets dumped to a tape, and is only using a portion of the block. Such as a recfm u type dataset has only 1 track written to a block at a time, and the blocksize for the dataset is 256K. Product such as RMM will only report the dataset size as being the BLOCKSIZE times the NUMBER of BLOCKS. Is there any tool or reporting mechanism that may interpret and report the ACTUAL size of the tape dataset, in other words how much actual tape was used for the dataset? In the above example, a track of data gets written to each block. If 20 tracks are written the RMM report shows 20 * 256K for the dataset size. I am looking for something that would report that actual tape useage size, or that the dataset was 20 tracks large on tape. Thankyou.

dick scherrer · Posted: Fri Sep 23, 2011 3:29 am

Hello and welcome to the forum,

Suggest you talk with your tape storage management people. Some of the tape management systems have an "amount of data" available.

Also suggest you do not care about part of a block being unused on a tape. It is completely insignificant (unless you have something really strange going on).

andy deiss · New User Joined: 23 Sep 2011 Posts: 3 Location: usa

Hello and thankyou for your reply. The reason for this is someone is looking to use this information to bill their customers. One such charge is for tapeuseage when dumping / backing up specific datasets. After change to DSS the blocksize changed( up in size from 64K to 256K) but the amount of data written to each block did not. This part does not need to be discussed as to why, suffice to say that now RMM, which reports the size of the dataset as being BLOCKSIZE * BLOCKS now reports a size about 4 times that which was reported previously. I am looking to find if there are any other tools which may be used to actually report how much tape is used for a dataset on a tape, based on dataset name, etc.. Thankyou.

Bill Woodger · Posted: Fri Sep 23, 2011 3:53 am

If the original reported figure was good enough, why not just divide the current answer by four? If the amount of data in the block is unchanged, the number of blocks will be pretty-much equivalent now as to before.

To put it another way, just continue the calculation with the old blocksize. Or have you been "spotted" and can't go that route?

Seems kind of mean to charge for back-ups. I hope no-one does any over-zealous cost-cutting :-)

andy deiss · New User Joined: 23 Sep 2011 Posts: 3 Location: usa

Understand your thoughts. We had recommended this, but it is a manual item to be done. They wish that there should be the ability to just run a report and be done with it. There is also another workaround that would cause customer to change jcl, but this would need to be changed on too many users jobs per their feedback. We are working at this, but my inquiry is to find out if there is another tool that could report the true tape useage for a dataset, as the RMM computation of dataset size = BLOCKS time BLOCKSIZE is no longer accurate in and of itself. Thanks.

dick scherrer · Posted: Fri Sep 23, 2011 7:18 pm

Hello,

Seems like the "chargeback" is based on a flawed calculation. . .

expat · Posted: Tue Sep 27, 2011 12:45 am

When I used to do chargeback on tape volumes, it was simple. Either the volume was used or it wasn't.

Chargeback was based on the number of tape volumes used.

dick scherrer · Posted: Tue Sep 27, 2011 1:22 am

Hi Expat,

Yup, and there was additional cost for i/o. Writing a million blocks cost more than writing 10 blocks, but in either case the whole tape is "used".

Well, except for when someone got creative and wanted to have a multi-file volume. Don't see this so much any more with applicatoin data, but there were times when someone wanted to know how much of the volume was "used". . .