Facilitating the Spread of Knowledge and Innovation in Professional Software Development

Write for InfoQ


Choose your language

InfoQ Homepage News Hazelcast Release Managed Cloud Offering

Hazelcast Release Managed Cloud Offering


Hazelcast, a vendor of distributed compute technology, including an open-source in-memory data grid (IMDG), recently revealed the general availability of a new, fully managed, cloud-based offering called Hazelcast Cloud.

As is common with managed services, Hazelcast Cloud removes the need to manage servers and other required infrastructure. Developers and other personnel can deploy new, pre-configured, instances of Hazelcast by simply using the browser-based console. Behind the scenes, Hazelcast Cloud makes use of the enterprise-edition of Hazelcast, called Hazelcast IMDG Enterprise HD, which adds additional features such as data persistence, recoverability, and rolling upgrades.

Hazelcast Cloud comes in three different offerings. The first two offerings, one of which is free, are hosted on Hazelcast-operated infrastructure. The free version comes with limits and the paid version increases those limits, adds support, and higher availability. The third offering, Hazelcast Cloud Dedicated, is hosted within a user’s own AWS VPC but configured and managed by Hazelcast. Support for Microsoft Azure and Google Cloud is also planned.

InfoQ spoke with Greg Luck, CTO of Hazelcast, about the release of Hazelcast Cloud.

InfoQ: What led to the creation of the Hazelcast Cloud offering?

Greg Luck: We know that enterprise developers are now moving to the cloud as a large movement. As they do, many want the convenience and consumption model of the cloud, rather than just managing Hazelcast themselves. We have also been talking to our customers, the vanguard of which would have wanted us to have done this a few years ago.

InfoQ: What’s the ideal use case or workload for Hazelcast Cloud? What would be a perfect fit?

Luck: In short, anything you use Hazelcast for today. Enterprises can lift and ship their applications to the cloud. The main use cases stay the same: cache or operational in-memory data store.
But, there are some new use cases in the cloud we think we are a perfect fit for. We can be the low latency, operational store for serverless functions, such as AWS Lambda.

InfoQ: Does Hazelcast Cloud support all of the same features as Hazelcast IMDG, such as the distributed ExecutorService? If not, what features are missing?

Luck: It is Hazelcast Enterprise HD, our highest edition, under the covers. So anything you can do with it you will be able to do with Cloud, including WAN replication to multi-cloud and hybrid cloud. So distributed ExecutorService, EntryProcessors, the full panoply is all there.

Today, we provide web-based configuration for Hazelcast iMap, our most popular data structure. You can create new data structures as you have always been able to via the programmatic API. We will provide more web-based configuration of the other data structures in the future.
In Hazelcast Cloud, when you allocate memory today it is all off-heap memory. We will also add heap configuration so that you can have more control over memory for the larger heap-based structures. But everything works today.

InfoQ: How are concerns of data privacy addressed with Hazelcast Cloud? Does it offer data encryption?

Luck: TLS encryption is a checkbox when you create a cluster. This is TLS with both a client-side certificate and a server-side certificate, for mutual authentication. We are unique in providing this today as others just do server-side certificates.

InfoQ: What is the pricing model for Hazelcast Cloud?

Luck: We charge 10 cents per GB per hour. Clusters can be checked as persistent, in which case they are saved to S3 when the cluster is suspended and then reloaded when resumed. That storage is free.

InfoQ: Hazelcast Cloud Dedicated seems different than most managed offerings since it’s deployed in the customer’s own VPC but managed by the Hazelcast team. What were the motivations behind this?

Luck: Actually, there are two approaches: create a cluster in a VPC unique to that customer and second, create the cluster in one of their existing VPCs.
Creating a cluster in a VPC unique to that customer is a common approach used by others, including MongoDB Atlas and Snowflake. You then setup VPC peering to the customer’s VPC.
Creating a cluster in the customer’s existing VPC requires additional security components such as a security gateway to be developed. Monitoring software vendors like AppDynamics use this approach.
All versions of Hazelcast Cloud Dedicated will have HA features for multi-region, multi-cloud, and hybrid cloud, all leveraging technology that already exists in Hazelcast Enterprise HD.

To learn more about Hazelcast Cloud, visit the getting started guide or the sign up page.

Rate this Article