Rockset, the real-time analytics database built for the cloud, is launching compute-compute separation for real-time analytics, which splits streaming ingestion from low latency queries.
While real-time analytics has grown both in popularity and necessity, its challenges in compute intensity have become realized, where resource over-provisioning and high operational burdens plague traditional databases and engineering teams alike. Relying on replicas to maintain a semblance of efficiency, the demand for real-time analytics compounded with the use of traditional databases stages a significant compute issue, according to the company.
This compute-compute separation allows streaming ingest or query serving to be scaled up and down independently, additionally authorizing multiple isolated applications to exist on shared real-time data.
The Rockset database, armed with compute-compute separation, is now equipped to further drive application efficiency while maintaining low latency and providing predictable performance.
“As a cloud-native real-time analytics database, Rockset already has compute-storage separation, but we found that was necessary, not sufficient for streaming architectures. Compute isolation is a critical innovation for breaking down the barriers of cost and complexity in processing high-velocity streaming data and low latency queries,” says Venkat Venkataramani, co-founder and CEO of Rockset. “Rockset is the industry’s first to extend cloud-native data architectures, delivering compute-compute separation for real-time analytics, and enabling multiple applications on shared real-time data, without any operational burden.”
The Rockset database release with compute-compute separation eliminates the need for replicas and allows for fast concurrency scaling. By alleviating the workloads of engineers and introducing independent scaling, the Rockset database delivers more for less at cloud scale, according to the company.
“Rockset delivered true real-time ingestion and queries with sub-50 millisecond end-to-end latency, that didn’t just match Elasticsearch, but did so at much lower operational effort and cost, while handling a much higher volume and variety of data,” said Emmanuel Fuentes, head of machine learning and data platforms at Whatnot.
To learn more about Rockset’s compute-compute separation capability, please visit https://rockset.com/.