April 25, 2018
Jason D. Rowley is a venture capital and technology reporter based in Chicago.
share

Behind almost every piece of software that you use, there is at least one database system to store and retrieve the information you need. And demand for ever more specialized and performant data storage has increased.

Follow Crunchbase News on Twitter & Facebook

ScyllaDB, a Tel Aviv-based company that develops and supports a database of the same name, announced that it has closed a $10 million Series C funding round led by TLV Partners. Prior investors like Bessemer Venture Partners, Qualcomm Ventures, Magma Venture Partners, and Western Digital participated in the round. Shahar Tzafrir, managing partner of TLV Partners, will be joining the company’s board.

The company was founded by Dor Laor and Avi Kivity. Both co-founders worked together at a company called Qumranet which was later acquired by Red Hat. At Qumranet and later at Red Hat, Kivity worked as the lead maintainer for the Linux Kernel Virtual Machine (KVM). Laor was a manager of the KVM project and briefly served as Red Hat’s director of platform virtualization before both he and Kivity left in late 2012 to start OSv. OSv developed the cloud computing operating system OSv, and the server-side application framework Seastar. The company rebranded to ScyllaDB in 2015.

To date, including this most recent round of funding, the company has raised $35 million in total equity funding.

The Technology

When working with large datasets, especially in cases where speed is critically important, it’s important to use a database that delivers high throughput with low latency. In other words: use a system that can move a lot of stuff very quickly.

ScyllaDB is an open-source NoSQL database that can serve as a drop-in replacement for Apache Cassandra. Cassandra bills itself as a distributed, high-performance, and scalable NoSQL database. A NoSQL database allows software engineers to input, store, and read data that doesn’t fit into traditional tabular formats used in databases like MySQL or PostgreSQL.

According to the company, it can deliver performance of one million input-output operations per second (IOPS) per node, can scale out to hundreds of nodes, and has a 99 percent tail latency of less than one millisecond.

“Beyond its distinct performance advantages, Scylla is unique in its ability to fully exploit available hardware resources, leading to a dramatic reduction in [total cost of ownership],” Shahar Tzafrir said in a statement.

So what makes ScyllaDB so fast, at least when compared to Cassandra? ScyllaDB does away with Java and the performance limitations of the JVM. Instead, ScyllaDB uses Seastar, an open-source C++ framework developed by ScyllaDB’s team for building high-performance server-side applications. Using Seastar, ScyllaDB shards processing requests across multiple engines, each running its own memory and network interface controller on individual CPU cores.

These close-to-the-hardware improvements deliver higher scalability and stability, and lower latency compared to Cassandra and other NoSQL database solutions.

The Market

According to a ranking released by DB-Engines, an information resource covering relational and NoSQL databases, Cassandra is the eighth most popular database system worldwide as of February 2018.

ScyllaDB boasts full compatibility with Cassandra, but it’s been used by companies with other needs. Extant implementations of ScyllaDB involve it being used for key-value storage, time series databases, large blob storage, and as a backend to graph databases, among other applications.

Shahar Tzafrir said in a statement that customers using Cassandra, Hbase, MongoDB, Redis, and even MySQL are switching to ScyllaDB. ScyllaDB’s current users include AppNexus, AdGear, CERN, Discord, IBM, Nauto, Ola Cabs, Samsung, and Ytel.

In addition to its open source offering, ScyllaDB offers an enterprise version as well. CEO Dor Laor said in a statement that “later this year we will launch our database-as-a-service offering as the next logical step for our high-performance NoSQL database.”