Qerying several times.

Question

I need to have different product information from multiple tables for different categories in the server cache or I need to fetch from Database every time. What will be the better way.
Solution 1: I can query upfront from DB and can crate objects in the server(10MB data). Over heads are these created objects takes lot of JVM memory and another one is at initial load going to connect/disconnect(db connections) DB several hundred times. Advantage is whenever I need certain information I can look for the object and can process things fast.
Solution 2: I can run the queries on demand, whenever needed. If transactions grows, performance goes down drastically. Because of querying DB each and every time.
Another though solution is using entity beans; appreciates your input on this.

Answer 1

Hi,
If you are following Solution 1.
1) It the DB size is huge , you cannot cache in the
memory.Of course it depends on the size, but you can cache very large databases in memory.
Prevalyer is basically based on that exact idea... http://www.prevayler.org/wiki.jsp
If you search this forum you will find at least one mention of using 16 gig of memory with a cached data store.
If you are following Solution 2.
2) As you are doing only selects what ever be the
size of the DB, if the query is properly tuned and if
the table is indexed, then the query would execute in
no time. Always use prepareStatement , it is much
faster and if you are going for Vendor specific then
you can write StoredProcedures where you can execute
more than one query at a time and in no time.
Technically for single requests, prepared statements are always slower. The only time that isn't true is if the database doesn't support prepared statements at all. There are of course other reasons for using prepared statements.
You can search the forum for benchmarks that demonstrate this.
Most systems will end up using single queries, even when the query is repeated due to the overhead and complexity of keeping and maintaining prepared statements. Normally the case where this isn't true is when processing a batch (many similar rows at once.)
But the other benefits of using prepared statements will offset the performance difference for most situations.