Rockset, the data platform that positions itself as a high performance indexing tier for analytic queries, has taken in an additional $40 million round of B funding that brings its total raised to $61.5 million. And the provider, which offers its indexing platform exclusively as a cloud-based service, is adding support for running its platform inside a customer's own VPC in the AWS cloud.
As we wrote back in 2018 when we introduced the company, Rockset was founded by a Facebook alum who built the social graph. It's a cloud-based service that offers a high performance, serverless indexing tier for offloading complex queries from data warehouses and search platforms. Built on RocksDB, an open source, log-based key-value store with an engine written in C++, Rockset incorporates what it terms a "converged index" that combines properties of columnar and search indexes. In the Rockset engine, every column is indexed, both in a columnar index and an adjoining search index, with the platform's SQL query optimizer running in a C++ engine making the choice of which one to use.
The result is a platform that performs a role akin to Redis, which adds a high performance in-memory tier to offload queries to hot data. The differences, of course, are that when Redis used as a caching tier, it is better suited for lookups rather than analytic queries involving complex SQL logic. (As a database, it is more suited for apps taking advantage of Redis Modules.)
Rockset in not intended for deployment with petabyte-scale data lakes. Instead, it is aimed at speeding up complex queries on well-bounded data sets. And so, it has seen significant uptake in the past year Snowflake, MongoDB, and DynamoDB customers. In these cases, it doesn't replace these databases, but instead offloads selected workloads. The typical use case for NoSQL platforms like DynamoDB and MongoDB is providing an analytic querying tier. For Snowflake customers, it is speeding up and lowering the cost of workloads that would otherwise require more EC2 consumption. In each case, Rockset populates indexes using change-data-capture streaming feeds from source databases.
As a cloud service that operates in AWS, data is typically stored in multitenant S3 storage. That's where the just-announced VPC feature comes in; it provides a way for the Rockset service to operate on data within the customer's own AWS VPC. It's targeted for SaaS services that build in Rockset's features, or enterprises with sensitive PII-class data.
With the new venture round, the 30-person company is looking to roughly quadruple in staff over the next 18 months, tripling sales and marketing teams and doubling the product engineering group.