FoundationDB Raises $17M For Database Similar To Next Generation Google Technology Used To Run Adwords

FoundationDB has raised $17 million for its industrial strength database with a data model similar to Google’s F1, a new scalable relational database management system used to run AdWords. The Series A round was led by Sutter Hill Ventures with additional investment from CrunchFund (which is owned by TechCrunch founder Michael Arrington) and existing angel investors. The round follows a significant seed round and investment by the founders and other investors.

The FoundationDB technology is scalable, fault tolerant and can scale across a distributed infrastructure without any single point of failure, an inherent issue with most NoSQL databases now on the market. The database strength comes from its transaction core, a feature that, for the most part, NoSQL database providers have not integrated into their technologies. It’s what FoundationDB calls “YesACID,” based on ACID, which was first developed in the 1970s. ACID is s set of properties that guarantees database transactions are processed reliably.

CEO and Co-Founder David Rosenthal said in an interview yesterday that FoundationDB’s transactional model has made it possible for the startup’s team to create layers that behave like MongoDB, SQL or a key-value store. The company is starting to open-source all the layers for general use. Rosenthal said the tradeoffs with the approach come with the complexity in building a database that can manage these multiple layers and cross-replicate across datacenters.

There are also the latency impacts that have to be mitigated with a model that works on a distributed infrastructure. Google runs F1 on top of Spanner, a global distributed database. It is similar to the way FoundationDB manages its layers by providing synchronous, cross-datacenter replication and strong consistency. According to a FoundationDB blog post, the replication implies higher commit latency but is mitigated by almost unlimited bandwidth and scale. It may take longer, but the overall throughput of the system allows for queries to be run in parallel at a very fast rate.

FoundationDB’s technology represents a new category of databases that are more than just pure SQL or NoSQL technology. They have qualities that make them usable for Internet-scale applications or for small teams that need the reliability that comes with ACID. FoundationDB points to a change that we can expect to see in the database market as more customers look for the reliability of SQL but also the scaling that comes with distributed technologies.

(Feature image courtesy of Waqas Mustafeez via Creative Commons on Flickr.)