Is a manual database load mandatory?

Question

hi,
i have a distributed, read write cache which connects to mysql (getters/setters & store implemented:
          <distributed-scheme>
               <scheme-name>distributed-db-user</scheme-name>
               <service-name>DistributedCache</service-name>
               <backing-map-scheme>
                    <read-write-backing-map-scheme>
                         <internal-cache-scheme>
                              <class-scheme>
                                   <class-name>com.tangosol.util.ObservableHashMap</class-name>
                              </class-scheme>
                         </internal-cache-scheme>
                         <cachestore-scheme>
                              <class-scheme>
                                   <class-name>cooee.DAT_UserStore</class-name>
                                   <init-params>
                                        <init-param>
                                             <param-type>java.lang.String</param-type>
                                             <param-value>user</param-value>
                                        </init-param>
                                   </init-params>
                              </class-scheme>
                         </cachestore-scheme>
                         <read-only>false</read-only>
                         <write-delay-seconds>10</write-delay-seconds>
                    </read-write-backing-map-scheme>
               </backing-map-scheme>
               <listener/>
               <autostart>true</autostart>
          </distributed-scheme>
then i query for a set:
                              for(Iterator iter = APP_Server.Cache_Inventory.keySet(new EqualsFilter("getUserID", id)).iterator(); iter.hasNext(); )
                                   Object oKey = iter.next();
                                   DAT_Inventory inv = (DAT_Inventory) APP_Server.Cache_Inventory.get(oKey);
                                   session.write("> INV " + inv.getItemID() + "\n\r");
unless i pre-populate the cache i get an empty set from the cache.
so is a pre-population (which would be a forced load all) mandatory (i assumed this would not be necessary and the queries would trigger some lazy cache-in stuff automatically).
if the pre-population should be mandatory: the loadAll() function of the cache requires a collection of keys, is there a way to force a pre-population without providing the keys?
thanks.
cheers.
Message was edited by:
druid99

Answer 1

hi robert,
thanks for your comments.
"Usually it is best to bulk load them to some JVM (one or more) and put them together to the cache, and also have some ways to signal to the cache that you don't want these puts to be written back to the db."
well, if i do a query to get the needed id's from the db and then just do bulk gets with those id's the cache will read it from the db thus making it available for a cache query which is ok.
with a total prepopulate of all db data to the cache i would circumvent the pattern of
1) get id's from db
2) perform a get with those ids from the cache
this is ok when it is clear where the specific id's have to be cached in for the 1st time, when this it not clear i have to use the above pattern all over the place ... which will put useless load to the db. of course one could optimize the pattern so that you only perform a get with those id's which are not in the cache anyhow etc.
however, i would like to use the cache in a way that normally i do not do any queries to the db directly but use only cache functionality. as far as i can see i have to do a full prepopulation of the cache at some point. the question is how to do this safe and performant on any node no matter if it is the 1st or the n'th node.
cheers.