Hazelcast, Inc., the company that enables instant action on all data, is announcing the latest iteration of its unified real-time data platform, aiming to empower organizations to rise above workload complexity in the age of AI with unparalleled consistency, performance, and scale. To enable enterprises to move past the “Day 2 problem”—or the data-related challenges the surface after AI model design—Hazelcast Platform 5.4 offers three core innovations: the advanced CP subsystem, the thread-per-core (TPC) architecture, and tiered storage.
Hazelcast puts a spotlight on data consistency in this latest platform release, acknowledging the needs of some of the biggest payment processing workloads in the world that it supports, according to the company.
The trouble with a consistent, distributed architecture is its impact on performance and resilience; with the platform’s latest update, Hazelcast continues to ensure strong consistency while simultaneously delivering high-performance, reliable workload support. Hazelcast Platform 5.4 protects the hard work developers and engineers put into creating their applications, preventing them from falling victim to “Day 2 problems,” and, instead, keeps costs under control while maintaining high performance at scale, according to the vendor.
The advanced CP subsystem makes a significant contribution to ensuring data consistency, delivering an accurate, continuously fresh view of data for all client requests for key/value data structures (“maps”) in a distributed system. As a feature driven by customer requirement, the advanced CP subsystem delivers an effective, predictable approach to data consistency and performance at scale, according to Hazelcast.
“There's constant pressure from the industry to cut down your costs, reduce your technology footprint. The advanced CP subsystem essentially allows you that capability without having to rebuild your infrastructure or go to a different vendor to do this… [you can do it all within] Hazelcast,” explained Karthic Thope, head of cloud engineering at Hazelcast.
Hazelcast’s new TPC architecture is designed to fully utilize all the cores in a modern CPU, increasing the Hazelcast Platform’s throughput by an additional 30% on large workloads with a high number of clients. This allows customers to fully exploit the computing power of its network, enabling them to process large data loads in sub-milliseconds with improved hardware optimization.
Tiered storage focuses on empowering highly scalable storage processing for intensive AI/machine learning (ML) workloads, all from a single, integrated environment. With larger workloads, tiered storage acts as a “spill-over disk” that when any data doesn’t fit into a customer’s machine’s memory, it spills over to the storage disk to ensure the same performance and efficiency expected of the Hazelcast Platform.
“This whole concept of being able to scale dynamically to your workloads fits in perfectly…[in] a prevalent thread across industries now,” said Thope. “Your volume is always spiky—[now,] you can predict how to handle bigger workloads, versus a period where there isn’t that big a workload going through the system. In any situation, whether it's retail finance, it sort of fits perfectly there.”
To learn more about Hazelcast Platform 5.4, please visit https://hazelcast.com/.