[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference orarep::nomahs::rdb_60

Title:Oracle Rdb - Still a strategic database for DEC on Alpha AXP!
Notice:RDB_60 is archived, please use RDB_70..
Moderator:NOVA::SMITHISON
Created:Fri Mar 18 1994
Last Modified:Fri May 30 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:5118
Total number of notes:28246

5016.0. "zero async writes question" by UKVMS3::SHISCOCK (stand and deliver) Tue Feb 11 1997 07:23

    Hi,
    
    Does global buffering and/or row caching have any impact on
    asynchronous writes?
    
    My customer has two fairly similar databases with fast commit
    and als enabled both using rdb 7.0. One is using local buffers
    and the other has GB and row cache enabled. With the database
    using LB he sees upto 10% aynch writes per area however on
    the other it's nothing. He does see some async activity on
    the root.
    
    Is this coincidence and we should look elsewhere or what?
    
    thanks,
    
    Steve
T.RTitleUserPersonal
Name		DateLines
5016.1NOVA::R_ANDERSONOracle Corporation (603) 881-1935Tue Feb 11 1997 09:118
Well, you are comparing apples and oranges here.

I assume from your description that the local buffer case does NOT have record
cache enabled.  This being probably the case, you cannot compare the I/Os
because of the completely different strategies that record caching uses for
migrating changes to disk...

Rick
5016.2okay andUKVMS3::SHISCOCKstand and deliverTue Feb 11 1997 09:2411
    
    Sure I'm comparing two different environments but it just seemed 
    odd that the database with GB and Row caching performs ZERO
    async writes. 
    So are you implying that because caching uses a different write
    stratergy it may or may not use async writes? Can you expand on
    that further please.
    
    thanks so far,
    
    Steve
5016.3NOVA::R_ANDERSONOracle Corporation (603) 881-1935Wed Feb 12 1997 10:345
It's probably more likely that the record caching async write mechanism has not
yet been instrumented to record async I/Os (the RCS server does them in most
cases).

Rick
5016.4ta muchlyUKVMS3::SHISCOCKstand and deliverWed Feb 12 1997 12:013
    
    
    Thanks. It closed the call with that customer.