We know the different types of node types available in Pega.

BATCH, STREAM, WEBUSER, BIX, BACKGROUNDPROCESSING, SEARCH, and yes ofcourse, UNIVERSAL!

Typical production-level systems may have multiple nodes of these single-node types, especially WebUser, and BackgroundProcessing to handle User traffic and API operations. But, Do we know how Pega internally handles these communications across those nodes?

Pega uses the Hazelcast external library to provide distributed caching and messaging capabilities in a Pega cluster. Hazelcast is an open-source in-memory data grid platform that provides a set of APIs for distributed caching and messaging.

When a Pega cluster is deployed, it consists of multiple nodes that work together to process incoming requests and provide a seamless user experience. To ensure that these nodes can communicate with each other and maintain consistency across the cluster, Pega uses Hazelcast as a distributed cache and messaging layer.

Hazelcast provides a distributed cache that can be used to store frequently accessed data in memory, reducing the need for repeated database access. Pega uses Hazelcast caching to store various data such as rule cache, data pages, session data, and lock manager. Each node in the cluster can access this cache to retrieve or update data as needed.

In addition to caching, Hazelcast also provides a messaging layer that allows nodes in the cluster to communicate with each other in real-time. Pega uses Hazelcast messaging to coordinate and manage various cluster-level activities such as distributed lock management, session replication, and cluster membership management.

When a node in the cluster updates a piece of data, it sends a message to other nodes in the cluster using Hazelcast messaging. The other nodes in the cluster then receive this message and update their own cache accordingly. This ensures that all nodes in the cluster are kept up-to-date and that there is consistency across the cluster.

Currently, Hazlecast comes as an embedded service with Pega deployment but please be aware that from Pega 8.8 Infinity versions, Pega started deprecating the support for embedded Hazlecast Pega deployments.