As I said previously, I believe it's vital to mirror JES. I don't know what that means for JES3, but for JES2 it means both checkpoint (primary or alternate) and all spool volumes. Anything less is useless at the recovery site.
Paging is another issue. There is nothing in a page data set that will be of any use in DR other than the volume itself and the data set allocation. Yet paging can entail very heavy I/O to mirror data that is all overwritten at IPL. It's simpler from a management perspective to mirror absolutely everything, but if performance is a concern, I suggest... 1. Mirror 'transient data' volumes periodically (maybe once a week), then stop mirroring. This keeps the data sets usable for recovery but avoids of the overhead inherent in throw-away data. 2. If you have volumes *dedicated* to temporary files like sort work, consider mirroring them once only at volume initialization. All you need is the VTOC. Be careful here. You don't want to miss cataloged data sets needed by applications. If you can't be sure that all data sets are temporary, then bite the bullet and mirror the whole volume continuously. . . J.O.Skip Robinson Southern California Edison Company Electric Dragon Team Paddler SHARE MVS Program Co-Manager 626-302-7535 Office 323-715-0595 Mobile [email protected] From: "Storr, Lon A CTR USARMY HRC (US)" <[email protected]> To: [email protected], Date: 09/04/2014 07:05 AM Subject: Re: JES2 and IBM Global Mirror (UNCLASSIFIED) Sent by: IBM Mainframe Discussion List <[email protected]> Classification: UNCLASSIFIED Caveats: NONE Thanks to all who responded. Alan -----Original Message----- From: IBM Mainframe Discussion List [mailto:[email protected]] On Behalf Of Scott Chapman Sent: Thursday, September 04, 2014 7:30 AM To: [email protected] Subject: Re: JES2 and IBM Global Mirror (UNCLASSIFIED) We've been mirroring our DASD with Global Mirror for several years (and SRDF before that). Our policy currently is to mirror everything. No problems at all. And it makes it a lot easier when you get to DR--there's no issue of "oh well that dataset we need for production really happens to be on a dev/test volume". IIRC, in the early days (~10 years ago, with SRDF) we did keep page volumes out of mirroring, at least for the two-hop solution where we did sync replication to a local bunker and async from there. But we even let those go now. I'm 100% sure we always did the spool. I'm 95% sure we always did the JES checkpoint, but primary checkpoint is in the CF. Scott ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
