gkn@SDS.SDSC.EDU (07/03/87)
I've been watching my file system caches lately, and have noticed that the hit rate on all of them was quite low. So, I increased the sizes of the caches and things worked well for the directory index (ACP_DINDXCACHE), file extent (ACP_EXTCACHE), file ID (ACP_FIDCACHE), file header (ACP_HDRCACHE), and bitmap (ACP_MAPCACHE) caches. The hit rates on the directory FCB (ACP_SYSACC), and the directory data (ACP_DIRCACHE) caches were still low, so I increased the sizes on those again. When I did this, the hit rate went *down*; I mean _through the floor_. So, I've shrunken the caches back a bit, and I'm back to where I was before. Am I misunderstanding something here? In general, the larger a cache is the better the chances that you'll find something in it. This doesn't seem to be the case with at least some of the XQP caches. Does anyone else have these kinds of problems (I know of at least one other site which has similar problems)? If anyone has any clues as to why the hit rate goes down after a certain size and how I can improve the hit rates (ie - optimize the cache size) I'd sure like to hear about it ... I'm running a 4-node VAXcluster (two 785s, a 750 (gag) and an 8200, all of which are way too small for what they're asked to do), and VMS V4.5, by the way. gkn -------------------------------------- Arpa: GKN@SDS.SDSC.EDU Bitnet: GKN@SDSC Span: SDSC::GKN (27.1) USPS: Gerard K. Newman San Diego Supercomputer Center P.O. Box 85608 San Diego, CA 92138 AT&T: 619.534.5076
LLACROIX@carleton.EDU.UUCP (07/07/87)
(Gerard K. Newman recently asked why his directory cache performance decreased when the cache size was _increased_.) There is probably a good, understandable reason for this, and I'll let someone else give the "correct" answer :-) Off the top of my head, I have two comments which may be relevant. First, how controled was your experiment when you were measuring the performance? If you just monitored normal activity, the activity mix may have changed and _that_ could explain the reduced cache performance. Second, and much more obscure, it is known that certain cache algorithms (such as FIFO) _can_ have decreased performance for a fixed sequence of requests when the cache size in increased. I don't know the details but I have seen a small example, considering cache sizes of 3 and 4. Intuitively, I would expect the chances of running into this in the real-life case of your directory cache to be extremely small. Intuitively, I wouldn't have expected the anomaly to exist, either. - Les LaCroix (csnet: llacroix@carleton.edu)