How does the coherence handle reads when coherence read-write-backing-map-scheme using write-behind queue behavior when batch not ripe yet?
(Doc ID 2427750.1)
Last updated on NOVEMBER 16, 2020
Applies to:Oracle Coherence - Version 126.96.36.199.0 and later
Information in this document applies to any platform.
The customer needs clarification on the behavior of Coherence read-write-backing-map-scheme using write-behind queue setting with the following two use cases:
Use Case 1: Coherence cache set up with read-write-backing-map-scheme to a DB store. It's write-behind queue set up with 5-second delay (write-delay) value. At some time (T), entry value for some key is updated in the cache. Data grid node member with key-value pair must process. The key-value pair is queued. The cache access for the node is not blocked. After T + 2 seconds, query made of cache for same key. So this is before SQL update completed. Does the node:
- Return the old value?
- Return the updated value?
- Something else?
Use Case 2: Coherence cache set up with read-write-backing-map-scheme to a DB store. It's write-behind queue not set up so all cache updates cause grid nodes to block during the update. At some time T, cache value for some key is updated in the cache. Grid node member with key-value pair must process. While DB operation completes, node member blocks. During blocking, new requests come in for a search operation followed by an update of another key-value pair. DB update completes. Does the node:
- Prioritize the update, block and process it?
- Prioritize the search, completes it, then does the update?
- Something else?
To view full details, sign in with your My Oracle Support account.
Don't have a My Oracle Support account? Click to get started!
In this Document