But my question is why create a 3390-1 for each of the individual data sets, like RACF db, Jes checkpoint, etc when I can create a 3390-27 (specifically 30015 cylinders) and use PAV's. Shouldn't that give me the same functionality without having to manage a bunch of small volumes?
Brad Wissink Information Technology Services Iowa State University 515-294-3088 -----Original Message----- From: IBM Mainframe Discussion List [mailto:[email protected]] On Behalf Of Lizette Koehler Sent: Tuesday, May 19, 2009 9:39 AM To: [email protected] Subject: Re: System data sets and 3390-27 If I understand DASD correctly, our mainframe dasd is carved out of the EMC storage (or whomever) array. This dasd lays virtually across multiple disks in the storage array depending on its RAID configuration. And you may or may not use BCV or PAV. I am currently working with a DMX4500 storage array running 73GB disks at 15,000RPM. It is transparent to me what it does with the definitions that are my 3390 dasd farm. Once the dasd is carved as a 3390 type device it is still a full volume as far as the mainframe is concerned. The storage array might be doing something with virtualization, I am not sure. But once you make it a 3390-3 or 3390-9 or 3390-27 and so forth it is a full volume of that size to the mainframe. The storage array will select the storage on its disk where that will be setting (hence virtual dasd). Does the storage array actually have that storage all tied up for that device? I am not sure, but on the mainframe side it will look like that type of disk. So we no longer have the old cabinets spinning the round brown disks with 3390 technology. We now have a storage array that pretends to be 3390 dasd but the storage array deals with how the data lays across its disks internal in the device. But for me - I still see it as a 3390-xx. I clip it, format it, move datasets to it just like I use to. I would think that there are still performance issues for these "virtual" disks so that we should consider keeping certain dataset types on their own individual dasd. Which is where storage arrays are great. You can create a 3390-1 which I will define as so many cylinders. It can contain my individual datasets, like RACF DB, JES Ckpt, or others. So now I can have a 10 cyl, 1000 cyl, or any size I like and call it a 3390-1 which I can put those types of dataset upon. I am still learning about this storage array environment. And I always learn something new whenever I work with them. It's a new frontier for me. Lizette > > On Mon, 18 May 2009 16:02:56 -0400, Lizette Koehler > <[email protected]> wrote: > > >However, if you are going to put on RACF Data Base on one Mod27 - nothing > > else on that volume, then some might question the waste of space. I > > am sure there is a balance to those issues. > > > > Not knowing the internals of that SAN thing... But still... I was convinced that > all that is virtualized in such a way that waste of space is actually > not an > issue anymore. As the actual physical spinning platters are only used > when there is actual data to store, allocating a big 3390 and not > writing anything on > it should not consume much capacity, no? > > Mind commenting, those who know (not me...). > > ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
