---------------
> Campbell,
>
> I have spent a good time with Oracle(as a matter of fact still do ),
but I
> don't have any benchmarks to offer. It was just common sense. Oracle
has a
> mechanism via which you can 'pin' objects ( tables, stored procedures,
> etc. ) in memory at database startup. Then, why use one more level of
> caching, if objects are allready cached in memory ?
> In addition to your observations, what also noticed is that another
> bottleneck can be opening/closing DB connections, so allways use a
> connection pool. And for avoiding roundtrips, allways use views/stored
> procedures, and delegate as much as possible processing to DB server,
but
> you allready said that.
>
> All best,
> Dan
>
To make it clearer. I have about 10 - 15 MB quite static data (may
change once in a month if at all). I doubt that querying a server via
ethernet connection can beat reading local hard disk. There is simply a
latency and bandwith restriction.
The only thing I need to do at startup is querying server for a cache
version id, in order to know if I have to refresh cache. In the
application there will be many queries on this data. Of course some
queries have to go to server because they asking for other than static
data.
Even now I have a simple self written cache that is much faster then
querying server. But the cache itself can be made faster by adding
indizes (maps) for quicker lookup. I noticed that I was going to write
my own little RAM database. Therefor I stopped and was looking for an
existing solution that is fast to implement and sufficient in time. I am
not looking for a hard coded cache that would probably be the fastest
but would require a lot of coding.
Regards,
Stephan
-------------------------------------------------------
This SF.Net email is sponsored by: IBM Linux Tutorials
Free Linux tutorial presented by Daniel Robbins, President and CEO of
GenToo technologies. Learn everything from fundamentals to system
administration.http://ads.osdn.com/?ad_id=1470&alloc_id=3638&op=click
|
| 
