IBM Mainframe Forum

Hi Guys,

While concatenating datasets having same record length and different block size, is it must to keep the dataset with largest block size at first.
I have tested this keeping dataset with largest block size at the last and not getting any error. I saw many posts where it was saying it is must to keep largest block size at first. Please suggest.

This was true in the past, but is no longer so. However, since many application outsourcers like to run OS/MFT on their 370/195s to save a rupee, we still disseminate this advice.

So, is there any advantage of keeping larger block size at first. because, what we know about block size is that it makes I/O operation faster, but it will be applicable as per the block size of each dataset. It is not like, once system will identify the largest block size and will process all the datasets with that block size only.

Please correct me if I am wrong.

So far as I know, there is no advantage, although someone like Dr. Sorichetti or Mr. Myers, whose knowledge of the internals of the operating system is far in advance of mine, would be a better guide in this.

IIRC, once upon a time the first dataset characteristics where used to build all the control blocks needed for the I/O

I do not remember when it was made available the possibility to specify a BLKSIZE on the first DD
in order to build the control blocks chain for the largest BLKSIZE

and provide for arbitrary concatenations .

PS.
when in doubt I always used to add a DCB=(BLKSIZE=32760) to the first DD of the concatenation of LRECL=80 datasets

Hello,

I do not remember when it was made available the possibility to specify a BLKSIZE on the first DD
in order to build the control blocks chain for the largest BLKSIZE

I also do not remember the date, but it was more than 30 years ago.

It won't affect the performance, the physical block on the physical media for any given dataset is the largest block that can be read for that dataset. no matter what override is specified on the DD. The largest blocksize had to be first, simply so that there was enough space allocated to read a block in to. No longer the case.

These users thanked the author BillyBoyo for the post:

nikesh_rai (Wed Oct 09, 2013 10:39 am)

nikesh_rai conveniently forgot to clearly state if this is for PDS concatenation or sequential data set concatenation.

The largest BLKSIZE first rule was mainly for PDS concatenation. Some years ago (but less than the 30 years ago someone mentioned) IBM changed PDS concatenation so it would automatically use the largest BLKSIZE in the concatenation.

These users thanked the author steve-myers for the post:

nikesh_rai (Wed Oct 09, 2013 10:43 am)

Thanks to all of you for clarification..

nikesh_rai conveniently forgot to clearly state if this is for PDS concatenation or sequential data set concatenation.

I was looking for sequential dataset..