Re: Stripe and Mirror Everything?

On Thu, 16 Sep 2004 12:46:00 -0400, David E. Grove wrote: > Thank you for your comments. > > >>For the database as a whole - undoubtedly. However, if any single table >>reaches a certain size, it becomes to big to fit into a single dbspace >>(under 9.40). You must then fragment the table, and (under 9.40 and >>earlier), each fragment must go into a separate dbspa > > I guess I need to go hit the docs... Are you saying that a non-fragmented > table can not fill the dbspace in which it was created? So, lets say I have > a 72GB dbspace, consisting of a single 72GB chunk. What is the larges table > I can put in it? A single partition (read table, fragment, index, or fragment) cannot exceed 16,277,315 pages or just under 32GB on a 2K system (64GB on a 4K page system). So your answer is that an unfragmented table cannot exceed 32GB and a fragmented table on a system with only 2 dbspaces cannot exceed 64GB, etc. Jonthan's point was that 9.50 is planning to remove the restriction that table/index fragments must live on separate dbspaces so you could have a table containing 32TB in 1024 fragments in a single dbspace in 9.50. To do that in 9.40 you would have to create and manage 1024 dbspaces! > Thank you. > >> However, the more practical difficulty on non-RAID systems is that you can >> usually only partition a disk into 8 partitions - one of which is the whole >> disk. So, if you had an 18 GB disk, you could create 7x2GB partitions for >> use by IDS, but the remaining 4GB was useless (2 GB if you were very >> careful). People objected to being unable to use large percentages of >> their disk drives. > > If large chunks hadn't come along, we would have tried to use Veritas or > DiskSuite. But, I guess fewer layers is better. By the way (sidebar)... > I've always treated slice 2 as just another slice. If I was running short > of partitions, and needed another one, I just used slice 2 for whatever set > of cylinders I wanted-- just like any other slice. Was/am I actually > jeopardizing anything? Me too. No you were not sacrificing anything. The ONLY problem would be if a dumb Sun tech decided to archive the drive by just dd'ing partition 2 and then trashed the drive and tried to restore it to a new one.....Boom! Otherwise the slice 2 convention was just that, a convention. > We are comfortable accepting the limitation of <4TB chunk. :-) > > >>assume you lay the partitions out with the >> whole drive starting at offset 0 (of necessity), > > Was I needlessly concerned to always make sure I avoided incorporating any > cylinder 0 into a raw device (to avoid overwriting VTOC). Maybe the devices > "know" to avoid the VTOC, and I didn't really have to worry?? VTOC only exists on Solaris partitions. For any VM based partitions, the partition table is placed on the drive OUTSIDE of the defined partitions. At our site, when they do have to create Solaris partitions (rare) the SA's define slice0 to start on the second cylinder so likewise the VTOC is out of the way. Art S. Kagel > Thanks again for your helpful thoughts. > > DG