Bulk cacheStore()? Bulk Db updates

Question

Good Afternoon,
     We have recently implemented a write behind cache for our site that persists to our most active Db table. The write behind behavior has been great at limiting the number of DB transactions by bulking the transactions for a single entity in the cache.
     Our DBA has asked us to take this one step further. The improvement so far only accommodate for each item in the table bulking updates to that one row. he would like updates over many objects/rows to be bulked.
     He would like the updates to the DB bulked using Bulk SQL to perform these updates (i.e. 1000 at a time) via a single transaction. We have had lenghty discussion about how this could be implemented.
     One: Have the original cache store to a second cache. This second cache will be dumped and processed via a JDBC array using Bulk SQL after x amount of time.
     Two: Pipe the cached values into a JMS queue that could process 100 at a time.
     Before we go down this road too much further I have been looking through documentation trying to find if Tangasol has a solution for this scenario already or perhaps even a different approach that I could propose to increase our DB performance and scalability.
     Thanks for any advice!
     -- Grant

Answer 1

Hi Robert,
     Perhaps I am not understanding the write being implementation correctly. I thought that the write-delay specified the amount of inactive time on a specific key/value pair that would then trigger a store() on that SPECIFIC cache entry. Are you stating that after 1 minute ALL entries in the cache are written or is that a different configuration from what i have below?
     You also state that I can "group these writes together into a single batch update ". Is this a Coherence configuration option or are you suggesting I implement the storeAll() to formulate the values into some batch update statement. Which is starting to sound like a good idea as i type it :)
     Thanks again for your input!
     <distributed-scheme>
     <scheme-name>shopping-basket-cache-scheme</scheme-name>
     <service-name>DistributedCache</service-name>
     <backing-map-scheme>
     <read-write-backing-map-scheme>
     <write-delay>1m</write-delay>
     <scheme-name>ShoppingBasketRWBMScheme</scheme-name>
     <internal-cache-scheme>
     <local-scheme>
     <scheme-ref>basket-eviction</scheme-ref>
     </local-scheme>
     </internal-cache-scheme>
     <cachestore-scheme>
     <class-scheme>
     <class-name>com.cache.coherence.shoppingbasket.ShoppingBasketCacheStore</class-name>
     </class-scheme>
     </cachestore-scheme>
     </read-write-backing-map-scheme>
     </backing-map-scheme>
     <autostart>true</autostart>
     </distributed-scheme>
     <local-scheme>
     <scheme-name>basket-eviction</scheme-name>
     <eviction-policy>HYBRID</eviction-policy>
     <high-units>5000</high-units>
     <expiry-delay>60m</expiry-delay>
     <flush-delay>60m</flush-delay>
     </local-scheme>