Reading Time: 2 minutes

SAN MATEO, Calif., Feb. 24, 2020 (GLOBE NEWSWIRE) — Rockset, the leading real-time database built for the cloud, today announced it is hosting the first RocksDB Meetup of 2020, taking place Wednesday, Feb. 26, 2020. The event will showcase talks from industry experts focused on the latest research, emerging trends, and interesting use cases in the RocksDB community.

Created in 2012 by Rockset co-founder and CTO Dhruba Borthakur, RocksDB is an embeddable persistent key-value store for fast storage, currently running some of the most demanding database workloads in the world. The RocksDB Meetup will feature cutting-edge research from an all-star lineup of speakers from LinkedIn and Facebook, located at the Rockset headquarters in San Mateo, Calif.


  • What: RocksDB Meetup
  • Who: LinkedIn’s Abhishek Shivanna (senior software engineer) and Ray Matharu (software engineer); Facebook’s Siying Dong (software engineer) and Zhichao Cao (Ph.D. candidate, research assistant at University of Minnesota)
  • When: Wednesday, Feb. 26, 2020 from 6:00 – 9:00 p.m. PST
  • Where: Rockset’s headquarters, located at 100 S. Ellsworth Ave., Suite 100, San Mateo, Calif. 94401
  • Register: Attendees can register for the RocksDB Meetup here
  • Watch: The event will also stream live, and a recorded version will be available in the following days

Session Descriptions

“Stateful Stream Processing with Samza and RocksDB”

Stateful stream processing powers thousands of applications at LinkedIn, ranging from notifications, abuse and trust, service monitoring and diagnostics, and many more. Samza is run as a managed stream processing service at LinkedIn and uses RocksDB to back its state storage abstractions. Presented by Abhishek Shivanna and Ray Matharu, this session details the systems and enhancements LinkedIn has built to provide consistency, fault tolerance, availability, and operability for its managed stream processing service—and the lessons learned in doing so.

“Characterizing, Modeling, and Benchmarking RocksDB Key-Value Workloads at Facebook”

Hosted by Facebook’s Siying Dong and Zhichao Cao, this session is part of a paper accepted into FAST 2020, the 18th USENIX Conference on File and Storage Technologies, taking place the same week. In order to make debugging, performance tuning, and workload analysis easier for RocksDB developers and users, Cao and Dong developed a set of workload tracing, replaying, and analyzing tools for RocksDB. In this talk, the detailed characterization of three RocksDB production use cases at Facebook will be introduced. These characterizations reveal several interesting findings such as strong key-space locality, key/value size distributions, and special temporal patterns. Moreover, a new benchmark that follows the distribution models from the real-world workloads is developed in db_bench, which will also be presented and discussed in this talk.

Supporting Resources

About Rockset

Rockset is a real-time database in the cloud, built by a team of industry veterans with decades of experience in web-scale data management and distributed systems at companies including Facebook, Yahoo, Google, Oracle and VMware. Rockset is backed by Greylock Partners and Sequoia Capital. For more information, go to or follow @RocksetCloud.

Media Contact
Marta Debski
Offleash for Rockset
[email protected]
(810) 956-4501