Has anyone tried Hazelcast?

[#1]

Hi,

Did anyone use Hazelcast? It is alternative for Cassandra, couchDB, etc.

[#2]

@Rushikesh Deshmukh

Alternate to in memory solution like ecache or grid grain

Not sure about Cassandra or couch db (couchbase)

[#3]

It's not part of HDP stack and solution for datasets that can fit in memory

[#4 · Accepted solution]

Hi @Rushikesh Deshmukh, Hazelcast is somewhat related to CouchDB since they both provide a kind of distributed cache, Hazelcast is more Java oriented, while CouchDB is a kind of a document store (like MongoDB). Cassandra is not in that group, it's a column-family oriented key-value store similar to HBase. Here is a great and pretty exhaustive list of NoSql storage solutions, more than 225 of them right now! It's a great reading, but at the end of the day you have to select one, or a few for your application.

[#5]

Just wanted to point out that Hazelcast is an in-memory data grid not a data store. With standard configuration, the data is entirely in volatile memory. You can use a backing store to ensure that data is persisted between restarts but the purpose of Hazelcast and IMDGs in general is for application acceleration not data storage. IMDGs are also capable of recieveing and distributing instruction sets across the cluster (send compute to data) similar to Hadoop. IMDGs can also execute instructions on every individual get/put/delete operation that hits the cluster. At the moment, IMDGs are not designed to scale past several TB and so would generally be used to augment a big data architecture, not replace it. However, the potential acceleration provided by and IMDG to an OLTP use case can be in the n^x realm.