> > I am not a DB2 person, but it would seem reasonable to me that IBM > should(in the documentation) make it clear that there are *NEW* major > requirements for DB2 (and what the new requirements are), so the > appropriate people can make the adjustments necessary.
A few people seem to be "piling on" over this change in DB2. We get constant catastrophising about the -potential- for disaster in -any- 64-bit address space. For sure, the sky is a very large place and it might fall on you. And you might be hit by a meteorite on your way home tonight. Who knows? I haven't run the arithmetic in a while, but as recently as a few years ago you couldn't have backed a single fully-populated 64-bit address space with all of the DASD ever shipped! It's really really big! But despite all of that potential for disaster, things tend to go reasonably well most of the time. How can that be if its so flippin dangerous? One way of looking at this is; how much data do you have online? A number in the low TiB range is probably a fair guess for a solid % of the customer base. Now how much of that do you suppose you could, or would, ever want to have resident in memory? 5%? 10%? How much AUX would you need to back that? Hmmm... about the same amount would be a fair number. 2x-3x might be more generous, but you could certainly get away with a whole lot less. Data in memory is all about optimizing access. If you don't want so much in memory, don't over-allocate virtual, or don't fill it up. Keep in mind that all of that virtual storage doesn't populate itself. The contents have to come from somewhere. In the worst case, the storage is referenced by clearing it, but that's just stupid programming. I'm pretty sure DB2 doesn't do that. Open a problem with IBM (or other vendor) if that sort of thing happens. Greg Dyck was a key player in the 64-bit-ification of DB2 and he's one of the smartest cookies out there. Is DB2 perfect? Well hardly, but I am far less concerned about DB2 running amok than I am about joe programmer experimenting with IARV64. Or with sysprogs and/or DBAs doing silly things with the system configuration. There are potential disasters around every corner in life. Most of them are avoidable through prudent planning. This is just another of those. Really. CC ---------------------------------------------------------------------- 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
