Re: Bulk Load - performance degrades

[EMAIL PROTECTED] (Chad) wrote in message news:<[EMAIL PROTECTED]>...
> When we do the second tree it should replace the cache with newly
> created pages until the cache becomes full again, then we should start
> getting a few misses. There is no contention for cache space between
> the first and second tree because each new tree is only created after
> the previous one has been closed off.
>
> While I realize that cache misses cause disk seeks and these have a
> noticeable impact on performance it still doesn't make sense to me
> that the second tree should be slower than the first, the third slower
> than the second...

The second load will definitely be slower than the first load, 
because when the second load needs a page, it will have to write   
a dirty page from the first load in order to make a page free          
for its use.

However, the performance should then balance out, and I would
expect the third load to be no slower than the second load.

Regards,
--keith

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Keith Bostic                          [EMAIL PROTECTED]
Sleepycat Software Inc.         keithbosticim (ymsgid)
118 Tower Rd.                     +1-781-259-3139
Lincoln, MA 01773               http://www.sleepycat.com