Marty, This system is going to be around for only 2+ weeks supporting some annual testing that we do. Its workload will be pretty focused and it will not grow (at least this year). The thing I was wondering about is whether CP keeps track of the storage allocations of all users and the sizes of all data spaces and, when that total is equal to the available page space, refuses to create another userid or dataspace. We are doing some reconfiguration to add to the page farm to be used during the tests but may be pretty close to not having the available dasd to support the data spaces if CP does have that type of checking.
Regards, Richard Schuh -----Original Message----- From: The IBM z/VM Operating System [mailto:[EMAIL PROTECTED] On Behalf Of Marty Zimelis Sent: Monday, July 10, 2006 11:44 AM To: [email protected] Subject: Re: Paging Space Question Richard, A quick calculation shows that 5.7M blocks x 4K bytes per block is about 22.8GB. That would *seem* to day that you have enough storage to satisfy your workload's demand. HOWEVER, I would weigh the cost of providing, say, 30GB of page space vs. the cost of losing your system -- even once -- to a PGTnnn abend (or whatever "page space full" is these days)? Marty -----Original Message----- From: The IBM z/VM Operating System [mailto:[EMAIL PROTECTED] On Behalf Of Schuh, Richard Sent: Monday, July 10, 2006 2:30 PM To: [email protected] Subject: Paging Space Question The Situation: A small, special purpose VM that is going to have 12 users accessing data that is contained in DIRC directories that are enabled for data spaces. These directories will have a total of approximately 5.7 million blocks committed. The processes using them will be most of what is running in a machine that has 26GB main and 12GB xstore. The question: Do we need to define sufficient paging space to accommodate all of the dataspaces plus anything else that may be running, or can we get by on fewer because the majority of the files in the dataspaces will be in memory? Regards, Richard Schuh
