Hazelcast Aims to Democratize Real-Time Data with Serverless
Hazelcast has a vision of democratizing real-time data processing and management, and the key to this strategy is the company’s Viridian Serverless offering.
Launched in beta at the end of July, Viridian Serverless is a self-service provisioning process where the cluster grows and shrinks automatically based on the workload, said Manish Devgan, chief product officer at Hazelcast. The service provides users with quick access to a horizontally scalable real-time data platform that includes a high-speed data store and stream processing capabilities.
Perfect for Hazelcast
Thus users can speed up app development, simplifying provisioning, and enabling flexible integration of real-time data into applications, Devgan said. With the service handling all underlying hardware configuration and any operational complexity, users can deploy a functional cluster in seconds. Moreover, the service helps organizations avoid over- and under-provisioning, and users can get started immediately with the free-for-life tier. At launch, Hazelcast said it would provide users with a limited-time offer of up to 2 Gibibytes (GiB) of data storage at no cost. This offer is still valid.
“Serverless is perfect for Hazelcast since they support extremely low-latency workloads that often oscillate wildly from a trickle to a thunderous rush of event data,” Mike Gualtieri, an analyst at Forrester Research, told The New Stack.
Without serverless the only possible way is to build a cluster with enough headroom to handle the largest expected load — or manually reconfigure clusters which is labor intensive, he noted.
“The serverless approach enables always right-sized compute, storage, and network resources without wasting electricity, cooling and thus cost,” Gualtieri said. “Serverless is not new or novel, but it is a particularly good architecture for Hazelcast Viridian because of the latency and throughput workload variability.”
While large enterprises with lots of developers, DevOps teams and SREs are equipped to stand up their own platform on a public cloud or in their own data center. “But not everybody can do that,” Devgan said. “Not everybody has 20 developers working on a real-time application and having to manage all this. So for those customers, we basically are targeting this offering where it’s a service offering. We basically want to democratize this real-time aspect of the economy.”
Getting Unpredictable
However, Hazelcast Viridian Dedicated is a managed, single-tenant cluster designed for organizations that have clear, predictable usage models and want precision control over their own clusters. Meanwhile, users can choose the serverless model for development and unpredictable production workloads, though customers requiring an enterprise-grade solution for production and predictable workloads will likely choose Viridian Dedicated for their company’s needs.
“If your workload is unpredictable, that’s when you would use Viridian,” Devgan told The New Stack. “And you want to build your application as quickly as possible. You don’t have to decide upfront, how many nodes in a cluster you need, you basically answer a few questions, and we will do the sizing under the hood for you.”
The pay-as-you-go model also benefits users who want to get started quickly and use the resources they need.
“When the serverless approach started a few years ago, it was mainly limited to stateless applications. However, we now see an increasing number of data management tools adopting this model,” said Sanjeev Mohan, principal and founder at SanjMo. “There are many benefits, such as optimized resource usage, decreased complexity of configurations, pay-per-use pricing, and scalability. We see this trend continuing as more databases announce their serverless capabilities.”
What’s Inside
Features include:
- Self-service sign-up with a free-forever tier that will include a limited-time offer of 2GiB of in-memory storage
- Familiar, declarative API for building applications that leverage real-time streaming data
- Support for streaming SQL to enable a large base of developers to run queries on real-time data
- Out-of-the-box connectors to multiple cloud data sources, and an API for building custom connectors to any other data source
- Integration with cloud deployments of popular data technologies, such as Apache Kafka
- WAN replication capabilities that enable data integration across cloud deployments to support geo-distributed systems and disaster recovery strategies
- Change data capture (CDC) technology to allow real-time responsiveness to data updates
The Hazelcast Platform
“Real-time applications are hard to build on traditional data platforms, forcing businesses to compromise on their strategic aspirations,” Devgan said in a statement. “Hazelcast Viridian Serverless is the next step in providing a truly real-time cloud, all while making it even easier to develop, configure and deploy innovative applications.”
Hazelcast competes with the likes of Redis and Aerospike, which are data platforms that provide very similar, low latency access to data. But they don’t they don’t do streaming.
The Hazelcast Viridian products are powered by the Hazelcast Platform, a real-time data platform and integrated runtime that combines distributed stream processing and real-time data management to automatically act on discovered patterns, trends, and anomalies, the company said. This eliminates functions that create bottlenecks in other architectures, including database writes, batch processing, and the need for human intervention.
“The days of nightly batch jobs are numbered,” Mohan told The New Stack. “A huge need in the market is to make decisions on the most recent data. To enable real-time analytics, we not only need the right data platform, but also a serverless data infrastructure. Hence, Hazelcast’s serverless option will help its customers increase development agility and deployment of their real-time applications.”
Moreover, Hazelcast has two popular open source projects. One is in the data management space — the company’s eponymous in-memory data grid. And the other big project is the Jet project, which is a streaming analytics engine.
“Imagine you’re writing a real-time application, and you’re able to do like a join of the data at rest, and also data in motion,” Devgan said. “There are a lot of use cases like this around — being able to do real-time analytics while the data is flowing in, and while doing that also pull contextual information for that event which is happening in real time.”
Use Cases, Availability
A typical use case for Viridian Serverless would be a bank extending an offer or coupon while a customer is banking or detecting a fraudulent event, Devgan said.
Indeed, from banking to retail and an increasing number of other industries, the adoption of serverless architectures continues to grow. In fact, 40% of enterprises using cloud native technologies are also using serverless, according to Cisco’s “2022 Global Hybrid Cloud Trends Report.”
According to Hazelcast, the Viridian Serverless offering is applicable to digital integration hub pattern architectures, enabling companies to capture data from numerous sources and correlate the data immediately to create enhanced context for end users. With this architecture, companies can gain a 360-degree customer view, track assets in real-time, prevent fraudulent transactions, deliver personalized recommendations, create real-time promotional offers, and more, the company said.
Also, the Hazelcast Serverless service is applicable to stream processing environments that are the foundation for payment processing, Internet of Things analytics, and microservices messaging.
The service is available on AWS, with availability on Google Compute Platform (GCP) coming soon. Hazelcast Viridian Dedicated is available on Microsoft Azure, but the serverless version is not yet available, Devgan said.
