Hi,

In the past, i used to work for a tiny shop with the same distribution you 
indicated. 
Only three Lpars and no Sysplex, no GRS.

At that time, we choose to make all disks available to all Lpars, but there was 
a segregation of Production, Development, and Sysprog volumes done by VOLSER. 
I don't remember the details anymore, but shared disks were labeled as SHR*, 
Production and development disks as PRD* and DEV*, and of course SYSRES, Page, 
spool, etc...

At IPL time, a small program was executed, searching all volumes and issuing V 
OFFLINE to those that do not belong to the appropriated Lpar. This program used 
wildcard masks to select what should remain ONLINE.

And, of course, MVS commands were protected in RACF, so only authorized userids 
can VARY ONLINE a volume.

It worked well for us, in this reality.


Best Regards

Ituriel do Nascimento Neto
z/OS System Programmer






Em sexta-feira, 25 de novembro de 2022 02:38:47 BRT, Joel C. Ewing 
<[email protected]> escreveu: 





But its not just a case of whether you trust they will not intentionally 
damage something, but the ease of accidentally causing integrity 
problems by not knowing when others have touched catalogs, volumes, or 
datasets on DASD that is physically shared but not known to be shared by 
the Operating System.  If many people are involved, the coordination 
procedures involved to prevent damage, assuming such procedures are even 
feasible, are a disaster waiting to happen.

 If volumes are SMS, all datasets must be cataloged and the associated 
catalogs must be accessed from any system that accesses those 
datasets.   If the systems are not in a relationship that enables proper 
catalog sharing, access and possible modification of the catalog from 
multiple systems causes the cached versions of catalog data to become 
out of sync with actual content on the drive when the catalog is altered 
from a different system, and there is a high probability the catalog 
will become corrupted on all systems.

Auditors are justified in being concerned whether independent RACF 
databases on multiple systems will always be in sync to properly protect 
production datasets from unintentional access or unauthorized access if 
test LPARs share access to production volumes.  There should always be 
multiple barriers to doing something bad because accidents happen -- 
like forgetting to change a production dataset name in what was intended 
to be test JCL.

There are just two many bad things that can happen if you try to share 
things that are only designed for sharing within a sysplex. The only 
relatively safe way to do this across independent LPARs is 
non-concurrently:   have a set of volumes and a catalog for HLQ's of 
just the datasets on those volumes that is also located on one of those 
volumes, and only have those volumes on-line to one system at a time and 
close, and deallocate all datasets and the catalog on those volumes 
before taking them offline to move them to a different system.

A much simpler and safer solution is to not share DASD volumes across 
LPARs not in the same sysplex, to maintain a unique copy of datasets on 
systems where they are needed, and to use a high-speed communication 
link between the LPARs to transmit datasets from one system to another 
when there is a need to resync those datasets from a production LPAR.

Joel C Ewing


On 11/24/22 21:38, Farley, Peter wrote:

> Not necessarily true in a software development environment where all members 
> of the team need to share all their data everywhere.  "Zero trust" is 
> anathema in a development environment.
>
> If you don't trust me then fire me.  It's cleaner that way.
>
> Shakespeare was *almost* right.  First get rid of all the auditors, *then* 
> get rid of all the lawyers.
>
> Peter
>
> -----Original Message-----
> From: IBM Mainframe Discussion List <[email protected]> On Behalf Of 
> Lennie Dymoke-Bradshaw
> Sent: Thursday, November 24, 2022 5:24 PM
> To: [email protected]
> Subject: Re: To share or not to share DASD
>
> If you were asking in a security context, I would advise against it in nearly 
> all cases.
> Auditors will not like that a system's data can be accessed without reference 
> to the RACF (or ACF2, or TSS) system that is supposed to protect it.
>
> Lennie Dymoke-Bradshaw
>
> -----Original Message-----
> From: IBM Mainframe Discussion List <[email protected]> On Behalf Of 
> Gord Neill
> Sent: 24 November 2022 20:55
> To: [email protected]
> Subject: To share or not to share DASD
>
> G'day all,
> I've been having discussions with a small shop (single mainframe, 3 separate 
> LPARs, no Sysplex) regarding best practices for DASD sharing.  Their view is 
> to share all DASD volumes across their 3 LPARs (Prod/Dev/Test) so their 
> developers/sysprogs can get access to current datasets, but in order to do 
> that, they'll need to use GRS Ring or MIM with the associated overhead.  I 
> don't know of any other serialization products, and since this is not a 
> Sysplex environment, they can't use GRS Star.  I suggested the idea of no 
> GRS, keeping most DASD volumes isolated to each LPAR, with a "shared string"
> available to all LPARs for copying datasets, but it was not well received.
>
> Just curious as to how other shops are handling this.  TIA!
>
>
> Gord Neill | Senior I/T Consultant | GlassHouse Systems
> --
> ...

-- 
Joel C. Ewing


----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to