- Query response time
- Clustering, data distribution, data affinity
Coherence has improved response times for queries of sizeable data sets. Also by allowing for data distribution and replication through clustering, it improves the reliability of information systems.
I've been using it for three years
We had one instance when we experienced intermittent network failure. This issue was not reproducible for obvious reasons. Coherence failed to live up to its SLA by not being able to recover but getting into a state where new nodes were created when the old ones were still there but for some reason no longer recognized as being part of the cluster. The Oracle support was not something to write home about, i.e. there was a constant request for more info (logs, timelines, etc. – which were provided) and never a feeling that the problem was understood or at least that there was any serious attempt at investigating or reproducing on Oracle’s side.
Medium to Good. Sometimes prompt competent responses, at other times support was lacking.
It was a company decision as this is a commercial product with guarantee of support.
It was complex. There are a multitude of configuration files and shell scripts, most of which could be copy and pasted. No uniformity of approach or tool to allow for proper management of configuration.
ROI is reasonably good, since no cheaper alternative satisfying company requirements was identified.
The product is considered expensive, hence the company will be on the lookout for a replacement if feasible. I'm not involved in licensing discussion.
My advice would be for Oracle to prepare a database with already existing configurations from clients. This would help future clients to have templates for various solutions already instead of reinventing the wheel. Generally Oracle fails at this chapter as opposed to open source solutions. It is very painful to start from scratch with little or no concrete solutions posted online (full solutions with commercial value).