Aim of membase
Mon, 07/12/2010 - 01:17
Hi, is membase aiming to be a "persistent" memcached, or can we expect new operations similar to redis or couchdb?
Having a persistent key&value store is nice, since i do not have to worry about filling the cache after a reboot, but getting more operations (list management for example) would allow me to consolidate caching&data into one layer.
Also, is there a way to make the buckets larger than the available physical memory (per node)? Or the SSD option is doing this?
Thanks for the inquiry Attila. Yes, we are planning on adding querying and indexing operations by the end of the year. Membase is designed to compete directly with the other NoSQL databases out there. It may be more useful to think of it as a database rather than a better memcached. The link to memcached is that the protocol is exactly the same which makes learning membase much easier.
As far as making the buckets larger than RAM, we're going to be releasing that feature VERY soon. The idea of SSD's in conjunction with regular spinning disk is to be able to have "tiered" storage where objects will be persisted first to SSD and then overflow into disk. All access into and out of the database will still be through DRAM for the best performance, and we will bring items into and out of DRAM as necessary based upon access.
Hope that helps, let me know what else I can answer for you.
Perry
Forum support is great for free but sometimes you need a guaranteed response time and dedicated resources for your questions or issues.
Consider purchasing enterprise-level support from Membase: http://www.membase.com/products-and-services/overview
Call or email "sales -at- membase -dot- com" today!