With GridGain 3.5 is just around the corner I want to take time and preview some of the changes that are coming.
We expect to push Enterprise Edition GA release of GridGain 3.5 in the next 30-45 days. Community Edition will be released later on with possibly some changes being back ported by that time – in which case we’ll release it the same day.
As we noted in our monthly newsletter GridGain 3.5 is the first release since GridGain 2.0 came out where we will be removing most of the deprecated APIs. In most cases deprecations were for non-uniformed naming that we wanted to fix but there will be some API changes. Specifically, JEXL-based predicates and Collision SPI will be changed.
Some of the older 3rd party integrations that we no longer support or can recommend will also be removed (that including JBoss Cache, JGroups communication, Mule and some of JMS providers). In all cases – our updated TCP/IP discovery and communication SPI provide much superior functionality, performance and stability for GridGain.
For all our Enterprise Clients we will provide close hands-on support in cases when non-trivial changes are required during migration.
One of the key themes for GridGain 3.5 is cumulative performance & stability improvements. We’ve worked with several of our customers this last quarter that utilized GridGain is rather edge scenarios that allowed us to really fine tune the performance characteristic. We’ll be introducing number of new configuration properties as well as design changes to GridGain such as changes to Collision an SwapSpace SPIs.
It is interesting to see how edge cases start popping when our software being used in real-time scenarios on 1000s of nodes. Achieving linear scalability is becoming progressively hard in these cases but I’m proud that GridGain is bucking this trend.
We’ll also introduce enhancements to our current APIs.
First of all, we’ll clean up the monadic support in Scalar, our Scala-based DSL so that Scalar projections can be used in standard Scala for-loop comprehensions.
We are also adding API-level support for affinity-based co-location. You can already use simple annotations to provide automatic affinity-based co-location – but the new APIs will make it all but effortless. You will be able to co-locate any closure or a task with any set of key in the in-memory data grid in a single call. I’ll detail it in one of the blogs later on.
Overall – GridGain 3.5 is looking to be an important milestone for GridGain eco-system.