Can the world handle yet another open source NoSQL database?
The world of the database is one of those areas that sees lots of people obsessing over details that to outside observers would seem trivial. Graph, NoSQL, SQL, distributed—so many choices.
So, when ScyllaDB told me about a funding round that they’d raised and their stated intention to replace Apache Cassandra, I was interested—if slightly skeptical. Not skeptical because of anything I know about ScyllaDB per se, but simply because of the busy-ness of the space.
+ Also on Network World: Google’s new cloud service is a unique take on a database +
The launch only a couple of weeks ago of Google’s Cloud Spanner database, an offering developed from the internal tools that Google itself uses, certainly upped the database ante. Google’s assertion that Cloud Spanner gives users all the benefits of both regular relational and NoSQL databases put all other database offerings on guard.
I’m always up for a challenge, however, and thought it worthwhile to at least talk to the company and find out what it’s about.
ScyllaDB was launched in 2015 and since that time, seems to have gained small but committed traction with the Apache Cassandra community. ScyllaDB has seen companies such as Arista, Eniro and mParticle use its offering in production, and those users explain that ScyllaDB offers higher levels of consistency, performance and availability than other options. ScyllaDB was created by the team behind Qumranet, the company that itself developed the KVM hypervisor.
CEO Dor Laor is quick to take aim at the offering in his sights:
“Quite simply, ScyllaDB redefined Apache Cassandra. We invite the Apache Cassandra community to join us in our mission to be the number one NoSQL datastore in performance and availability.”
Why Scylla?
Of course, ScyllaDB could foresee that many would question the need for yet another open source NoSQL database. It has done its homework and asserts that while Apache Cassandra has succeeded in creating a viable community for its high-availability use case, the innate limitations by low-performing Java and external pressure from Redis, Hadoop and others have left the community a number of questions relating to performance requirements, the time and financial impacts of yet another database, and concerns about community support for Cassandra.
The company has, therefore, come out strongly to justify why an Apache Cassandra user would consider using Scylla in production. They suggest that Scylla enables developers to build mission-critical applications with high levels of throughput and ultra-low latency. The angle here is that Scylla leverages the best from Apache Cassandra—in particular high availability, fault tolerance and its rich ecosystem—but that Scylla offers developers higher performance and resource efficiency.
It’s funders seem convinced; the company is announcing a $16 Million Series B funding round led by Western Digital, Samsung, Magma Ventures and Qualcomm Ventures, with participation from Bessemer Venture Partners.
My POV
Apache Cassandra certainly has a loyal following, Scylla is therefore smart to articulate the “simple drop-in replacement” that it offers Cassandra users. The dual promises of larger scale and resource impacts that it resolves is a compelling proposition.
Of course, all of this needs to be viewed within the context of an increasingly busy database space (both open source and proprietary). Whether Scylla can really make a dent before the next entrant attempts to knock it of its perch is anyone’s guess.
