All, Yes, my final paragraph should read:
" Anyway I have probably bored you to tears. My key point is to go VERTICAL, and avoid HORIZONTAL pooling." That was a helluva transposition after such a long diatribe. Thanks to all those that pointed out my mistake - it confirms you took the time to read my email :-) Ron > -----Original Message----- > From: IBM Mainframe Discussion List [mailto:IBM-MAIN@LISTSERV.UA.EDU] > On Behalf Of Ron Hawkins > Sent: Wednesday, July 04, 2012 3:59 PM > To: IBM-MAIN@LISTSERV.UA.EDU > Subject: Re: [IBM-MAIN] Converting to SMS for the first time > > Lizette, > > Before starting down the SMS path I think there are ways of approaching > your storage groups that will become increasingly important over time. One > is horizontal pooling, and the other is vertical pooling. > > Horizontal pooling is where you group volumes based on some business or > application boundary, like having all the volumes for banking in one storage > group, and all the volumes for credit cards in another. > > Vertical pooling is where you pool volumes based on the volume attributes, > so you may have all your HUR copied DUPLEX volumes in one storage group, > and all the suspended temporary dataset volumes in another storage group. > > I would urge you to consider vertical pooling rather than horizontal pooling. > Horizontal pooling by its nature appeases many of the political problems that > occur with implementing SMS, especially the MVS neophytes that cannot get > their mind beyond the "that other application can steal all my freespace" > mindset. I have seen this turn a site into 120+ storage groups, and I've heard > of more. You don't want to be managing that. > > Horizontal pooling allows you to minimize the things you manage, and > maximize the access to the resources within the pool. I start out with just a > few Storage Group categories that define workload that you may want to > keep separate for usage, performance, or some other reason. As a rule I use: > > TMP: truly temporary datasets, and not transitory datasets passed from job > to job > > DBS: any dataset allocated by a DBMS (IMS, DB2, IDMS, etc) or an OLTP (CICS, > etc). Some datasets may also be used by batch jobs, but if it is part of a > database at any point then it goes here. > > GEN: the general pool which starts out as anything with a TSO users HLQ or a > DSORG=PO. > > SYS: the system software pool. Datasets with a software product HLQ go > here (SYS2, SYS3, CA, BMC, etc). Most of this is software that is not cutover > with a pack copy, or the actual data files used by the software. > > GTD: a guaranteed space pool for anything that requires hand placement or > dedicated volumes. For example volumes shared across plexes with > converted reserves turned off, or a dedicated, custom sized volume for the > TMC. > > BAT: the batch pool. Essentially anything that does not qualify for the first > five categories will be created a by a batch job and drop through into this > pool. It makes ACS coding easy :-) > > If datasets within these categories have special hardware requirements, you > can split the pool and name them with a suffix. For example you have a > mixed vendor shop and most of your batch datasets can use HDS > Shadowimage, but there are a few applications that are built around EMC > Timefinder. You can use SGBAT00 as a generic Storage Group, and have a > SGBAT01 storage group for datasets that require Shadowimage, and > SGBAT02 for datasets that require Timefinder. SGBAT01 only has HDS > volumes, and SGBAT02 only has EMC volumes. When you replace your EMC > with a VSP (VBG) you can merge the three pools back into SGBAT00. > > Once I have poured over DCOLLECT and Type42_6 records to establish how > datasets are used I build a model of the storage group space requirements > and contents. Many years ago I wrote a SMS allocation simulator in SAS using > these SMF records. It helped me plan SMS conversions at a few sites, > especially the pool sizes and Management Class migration potential. It also > provided the basis for a large part of the ACS code. > > One medium sized site in Indonesia - really smart people - implemented a > complete conversion with this approach in a three month window from > project start to end. After seeing this work on production they implemented > this in development with a single storage group, which is what I would usually > recommend for a typical development setup. > > Anyway I have probably bored you to tears. My key point is to go horizontal, > and avoid vertical pooling. > > Ron > > > > -----Original Message----- > > From: IBM Mainframe Discussion List [mailto:IBM- > m...@listserv.ua.edu] > > On Behalf Of Lizette Koehler > > Sent: Tuesday, July 03, 2012 6:27 AM > > To: IBM-MAIN@LISTSERV.UA.EDU > > Subject: [IBM-MAIN] Converting to SMS for the first time > > > > I have a friend who has a shop that is non SMS. I am trying to > > provide a simple check list on how to convert to SMS managed storage. > > > > If anyone has suggestions on best practices for a conversion – please > > let me know. > > > > I have one idea of how to do, but others with more experience might > > have a more elegant approach. > > > > My concept is to create some empty SMS Pools. Change/establish ACS > > routines (using the SAMPLE IBM provides) and slowly take one HLQ after > > another. Then when comfortable with the process and ACS/HSM > processes > > work, go for the big bang. Yes – This is a high level view of what I would > > do. > > So steps are missing. ☺ > > > > If there are better ideas, I thank you for the guidance. > > > > > > Lizette Koehler > > > > ---------------------------------------------------------------------- > > For IBM-MAIN subscribe / signoff / archive access instructions, send > > email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN > > ---------------------------------------------------------------------- > For IBM-MAIN subscribe / signoff / archive access instructions, send email to > lists...@listserv.ua.edu with the message: INFO IBM-MAIN ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN