Follow Datanami:
March 6, 2019

Fauna Announces Official Results of Safety Study of Its Software

SAN FRANCISCO, Calif., March 6, 2019 – Fauna, provider of the distributed OLTP database FaunaDB, has announced the official results of an exhaustive safety study conducted on its software by independent testing firm Jepsen. The report conclusively demonstrates that FaunaDB joins the elite class of databases that offer strictly serializable transactions in geographically distributed environments. But unlike other databases, FaunaDB does so without placing limits on the number of keys or regions, and is robust against cloud chaos such as clock skews or intermittent node outages — resilience that is critical for delivering the highest quality customer experience in today’s competitive digital business arena.

The FaunaDB study was conducted by Kyle Kingsbury, the creator and maintainer of Jepsen, a leading independent proving ground for database technologies, under a thorough and aggressive test plan to validate FaunaDB’s architecture and functionality and investigate its stability and enterprise-readiness.

The Jepsen report details evidence for the safety and reliability of FaunaDB and concludes that it offers strict serializability across multi-region transactions, the “gold standard” in data consistency for concurrent systems. Key observations from Jepsen, as highlighted in the report, include:

  • “We expect to observe snapshot isolation at the minimum, and where desired, we can promote SI or serializable transactions to strict serializability: the gold standard for concurrent systems.”
  • “FaunaDB’s composable query language, temporal queries, and support for transactional consistency models ranging from snapshot isolation to strict serializability are welcome choices, and they work together well. We’re pleased to see these ideas brought together in FaunaDB.”
  • “Membership changes are notoriously difficult to get right, especially in consensus systems, and we appreciate Fauna’s efforts on behalf of their users.”

In today’s competitive digital age, companies need global database architectures that support classic RDBMS primitives, such as ACID transactions, relations and constraints, while scaling horizontally and performing effectively across private, public, hybrid, and serverless clouds. Otherwise, business applications are vulnerable to severe bugs and outages that can have a significant impact on customers and weaken a company’s brand.

“The ongoing enterprise adoption of the cloud is driving new requirements for databases,” said James Curtis, Senior Analyst, Data, AI & Analytics for 451 Research. “Emerging vendors such as Fauna are driving operational databases to address enterprises with global needs, featuring data guarantees in distributed cloud environments, to address a key need of companies engaged in the digital transformation journey.”

The primary difference between FaunaDB and other distributed OLTP databases is its ability to meet these requirements in cloud environments, combining established technologies into a single system that delivers on three important business values: productivity, safety and agility.

To accomplish this goal, FaunaDB is the world’s first implementation of Calvin, a protocol for fast distributed transactions for partitioned database systems. Formulated by leading academic Daniel J. Abadi and a team of researchers at Yale University, Calvin uses a deterministic ordering guarantee to significantly reduce the normally prohibitive contention costs associated with distributed ACID transactions.

“When we started building FaunaDB, our goal was to deliver a database that offered global data consistency without compromising operational scalability in today’s distributed environments – something others in the industry said was impossible,” said Evan Weaver, Founder and CEO of Fauna. “For that reason, we chose Calvin as the underlying transaction protocol. The results of this study are a testament to the hard work by our engineers to achieve a reliable, strongly consistent globally distributed database.”

“We believe FaunaDB’s approach is fundamentally sound. We suspect that Calvin-based systems like FaunaDB could play an important future role in the distributed database landscape,” said Kyle Kingsbury at Jepsen.

FaunaDB’s Jepsen report conclusively puts FaunaDB at the top of the list of vendors offering global transactional databases:

  • FaunaDB is the only distributed database that offers cloud-agnostic strict serializability in geographically distributed deployments.
  • FaunaDB offers this capability without specialized hardware and without clock synchronization dependences outside normal operational practice.
  • Other databases in its class sacrifice consistency, scalability, or both, in their efforts towards multi-datacenter distribution.

The full Jepsen Report is available at www.fauna.com.


Source: Fauna

Datanami