Good article about In-Memory Data Grids on Cubrid’s blog by Ki Sun Song.

The features of IMDG can be summarized as follows:

• Data is distributed and stored in multiple servers.
• Each server operates in the active mode.
• A data model is usually object-oriented (serialized) and non-relational.
• According to the necessity, you often need to add or reduce servers.

Even if you don’t read it all, but plan to use an IMDG solution, the first two questions you want to ask your vendor are: what the approach you are proposing to deal with the limited memory capacity and what’s the strategy for reliability. You’ll get good answers from well established products, but these answers are not necessarily the ones that provide the exact requirements your solution will need.

Original title and link: Main Features of In-Memory Data Grids (NoSQL database©myNoSQL)

via: http://www.cubrid.org/blog/dev-platform/introduction-to-in-memory-data-grid-main-features/

Nikita Ivanov (GridGrain):

Look, 90% of NoSQL usage comes from the same crowd as a typical memcached users: non-critical, “moms-n-pops” websites. 90% of IMDG/IMCG usage comes from mission critical systems.

Different customers => different requirements => different products…

Unfortunately neither my mom nor my dad are behind non-critical websites like Amazon, Facebook, Google, or LinkedIn. What about yours?

Original title and link: NoSQL Is for Moms-N-Pops Websites (NoSQL database©myNoSQL)

via: http://java.dzone.com/articles/why-memory-data-grids-are-more

A new in-memory data grid, always consistent and suporting high :

And that’s why we’ve begun to build Galaxy, a new kind of IMDG that does not use hashing (or keys at all) for partitioning, but rather a software implementation of a cache-coherence protocol, very similar to the one used by CPU L1 caches. As objects move around in space, our database would use Galaxy to migrate them from one node to another, and because nearby objects are often accessed at the same time, keeping them on the same node would do away with network I/O for the common case, and would offset the higher cost of an initial object lookup.

The creators say the origin of Galaxy (and SpaceBase—the commercial product built on top of Galaxy) is in real-time spatial data stores for MMO games, location-based services and military services. My knowledge in these fields is zero, thus I’m having a difficult time understanding its applicability. But the Hacker News community has already dissected it.

Original title and link: Distributed Memory Systems: Introducing Galaxy In-Memory Data Grid (NoSQL database©myNoSQL)

via: http://blog.paralleluniverse.co/post/26909672264/on-distributed-memory