Recently, I've seen a lot of interest at our clients in
Enterprise Data Fabric. Two of our major finserv clients have purchased
GemFire and we've started getting some hands-on experience with it. We're helping to build a Market Data System which uses GemFire as the transport layer and the caching layer which sits between the clients and the database. Several things strike me from my early experiences with this product:
- It's totally distributed. I can distribute GBs and even TBs of data across nodes.
- Since it supports seamless failover, it's possible to design an application so that every single call for data hits the cache and does not hit the DB.
- If I was a messaging middleware provider (Tibco/Reuters, MQ Series, etc), I'd be very, very worried.
- This technology relegates the database to the role of 'permanent data storage'. Once you've got GemFire setup to cache your data, you can drastically reduce the number of DBs deployed across your enterprise.
You should definitely expect to see more raves (and the occasional rant) in this space about our experiences with GemFire in the coming weeks as we get deeper into our MDS implementation.