On 09/14/2009 06:15 PM, Neil Duffee wrote: > On 2009-09-11 at 16:14, concerning "stop-x37 type of function in > SMS", jam...gr...@efir...com wrote to IBM-Main: > >> While at SARE a couple weeks ago my manager was told by an IBM'er that >> SMS could handle space abends like a STOP-X37 product. >> >> Has anyone heard this and if so how is it implemented? > > Jim: I've been using something similar for several years now. Check > out the concept 'Overflow Storage Group' as well as 'Extend Storage > Group'. > > In short, I created a StorGrp (appropriately named SPILL) with empty > volumes for overflow. The volumes are Enabled but the StorGrp is > Quiesced. In your StorGrp ACS routine, append the SPILL group to all > SET &STORGRP= statements that you want protected. (there might be > reasons not to) > > The volumes in a Quiesced StorGrp are less preferred than volumes in > a normal pool. I also like to have a quiesced volume (or 3) in each > pool which is preferred over the Quiesced StorGrp. If datasets end > up in the overflow pool on a regular basis, the primary pool likely > needs more volumes. Read "SMS Volume selection for Data Set > allocation" in the Storage Admin Reference manual (mine's SC26-7402- > 01) for a blow-by-blow description. > > Other aids, that I can think of, require the use of a DataClas; > specifically the 'Space Constraint Relief', 'Dynamic Volume Count' > and, potentially, the 'Extended Format/Addressing' attributes. I'll > let you investigate. > > ----------> signature = 6 lines follows <-------------- > Neil Duffee, Joe SysProg, U d'Ottawa, Ottawa, Ont, Canada > telephone:1 613 562 5800 x4585 fax:1 613 562 5161 > mailto:NDuffee of uOttawa.ca http:/ /aix1.uottawa.ca/ ~nduffee ...
For large sequential files where the only problem is free space fragmentation, not lack of total free space, in the storage group, the simpler solution to manage is the DATACLAS route mentioned by Neil; namely, define and use an SMS DATACLAS that forces Extended Format, Space Constraint Relief, and allows for a suitably large Dynamic Volume Count. EF with SCR permits almost an order of magnitude larger number of extents per volume and the ability to get an arbitrary number of extents on an arbitrary number of volumes to satisfy either a primary or secondary allocation. Unless fragmentation is incredibly bad, you pretty much have to be out of space in the Storage Group before an allocation or a dataset extension will fail. After using this for years, we only recently encountered an unexpected restriction that makes this approach not work with arbitrary "&&" temporary datasets: it turns out that && datasets can't be Extended Format and multi-volume. As near as I can tell, this is not documented anywhere except in the error message you receive if you try to force these attributes on an && dataset larger than the 65K single-volume track limit. -- Joel C. Ewing, Fort Smith, AR jremoveccapsew...@acm.org ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
