Hazelcast Platform integrates data grid and event streaming
Hazelcast's new technology will provide a unified SQL query engine on a single cluster that will allow users to query both real-time data and data at rest.
Hazelcast on July 14 unveiled a new technical direction, bringing together the vendor's In-Memory Data Grid database and Jet event streaming technologies into a single platform.
The new Hazelcast Platform 5.0 is now out in beta, with general availability expected for August. Hazelcast IMDG has long been used alongside Jet, but the two technologies until now had been separate and distinct platforms.
By integrating them in the new Hazelcast Platform, the in-memory data grid vendor's goal is to provide organizations with the capability to rapidly ingest event streaming data and then use it for applications, including both analytical and transactional workloads.
The new Hazelcast release combines the best of both instantaneous data and instantaneous analytics, said Mike Gualtieri, an analyst at Forrester Research.
"Modern applications and especially AI applications need to react in real time," Gualtieri said. "With Hazelcast, enterprises can get a durable in-memory data store and analytics. It greatly simplifies real-time architectures."
Hazelcast Platform accelerates data for use
John DesJardins, CTO at Hazelcast, explained that historically organizations have been able to use IMDG and Jet and have them work together.
What's new about the Hazelcast Platform is that the two technologies are integrated in one cluster that is optimized to ensure that the data grid workloads and the streaming workloads can coexist, DesJardins said. That integration provides a unified runtime that scales up and down in an optimized approach, providing better overall performance.
"What that means for the users is they now only have one cluster to manage," DesJardins said.
With the combined Hazelcast Platform users can query real-time data as well as data at rest in a unified approach.
Hazelcast CEO Kelly Herrell said that previously, with two separate systems, users had to move data from one system to the other as needed in order to execute different types of data queries.
That's not the case with Hazelcast Platform, which has a common query language for both event streaming and in-memory data. Herrell noted that unifying event streaming and in-memory also reduces latency so organizations can use data in real time in analytics and other applications.
Mike GualtieriAnalyst, Forrester Research
In the past, with the separate IMDG and Jet platforms, users had to deal with two different SQL query platforms when querying data, with different syntax.
"With our common query language in Hazelcast Platform, it doesn't matter if you're querying the real-time data or the stored data, it's the same SQL syntax," Herrell said.
Next move for Hazelcast Platform is more persistence
Alongside the integration of Jet and IMDG, the new Hazelcast Platform will begin to expand beyond its in-memory roots to provide data persistence options.
DesJardins said that alongside IMDG, users have been deploying a database, with some users telling the vendor the only reason they have the separate database is that IMDG does not write data to disk, as IMDG is an in-memory data store.
With Hazelcast Platform, DesJardins noted there is a new durable persistence capability.
That means that users can have data available even after they restart or stop a cluster. The initial phase of durable persistence stores the same data that is available in-memory to disk.
In the future, DesJardins Hazelcast expects to develop more persistence features, including the ability to work with tiered storage.
Confluent builds out Kafka event streaming platform