The data should be structured to 1) minimize the search time 2)
minimize the hotsync time

Depending on the comparison function, it could take several seconds
just to check a few thousand records. A linear search just might not
cut it. I've worked on similar large databases with many records. In
order to get decent performance I always added an extra record or two
with sort information. This allowed me to do extremely quick binary
lookups. 

Also, how big (on average) is each record? Will they always be the
same size? If so, you may want to consider packing as many as you can
into the fewest number of records. This will increase your sync speed.

Also, don't forget each record adds 8 bytes of header. My tests suggest that the OS
actually keeps a little more than this. In my experience, even with
small record sizes, I could only get around 30k records on an 8meg
device (event though 30k * (recordsize + 8 bytes of headers) was
about 400-500k less than 8meg. 

-mark



On Thu, Oct 03, 2002 at 12:05:06PM -0700, Lucy wrote:
> I have an option to install the PBD as one large db
> (23,000+ recs) or as a few smaller dbs (5000 recs
> each). My goal is to search against all those records
> according to user's input. I will have to order the db
> (or dbs) by first 2 fields.
> 
> Which way I should go? I know the hotsync is a lot
> faster: 30 mins/db vs 10 mins/multi dbs. What about
> performance, etc? 
> 
> Any input is greatly appreciated.
> 
> __________________________________________________
> Do you Yahoo!?
> New DSL Internet Access from SBC & Yahoo!
> http://sbc.yahoo.com
> 
> -- 
> For information on using the Palm Developer Forums, or to unsubscribe, please see 
>http://www.palmos.com/dev/support/forums/

-- 
For information on using the Palm Developer Forums, or to unsubscribe, please see 
http://www.palmos.com/dev/support/forums/

Reply via email to