NuoDB Takes the Wraps Off Blackbirds Database
NewSQL database vendor NuoDB today unveiled the version 2.0 release of its in-memory relational database. Codenamed Blackbirds, the updated database’s main new feature is the capability to support geo-distributed deployments, which allows customers to deploy a single database across multiple locations to boost availability and scalability.
NuoDB, like other NoSQL and NewSQL startups, is chasing the market for new database management systems that are flexible and scalable enough to handle the new generation of applications that traditional databases (Oracle is the 800-pound gorilla in the room here) have a hard time supporting.
If version 1.0 was about getting the database off the ground, version 2.0 is about making it fly farther and faster. NuoDB CTO Seth Proctor recently explained the significance of Blackbirds in a telephone interview with Datanami.
“We’ve worked very hard in this release to make it be something that feel like something very familiar, that has increased number of data types, functions, language features, and management capabilities… to really get the type of maturity that people want from a relational database,” he says.
Support for geo-distributed deployments will give NuoDB customers the capability to run a single Blackbirds database in multiple data centers, thereby giving the applications that run on it higher availability and lower latency, Proctor says. The challenge was delivering this capability as an out-of-the-box feature, without forcing customers to micro-manage the database.
“If you have a web application and you’ve deployed it on the East Coast of the U.S. and the West Coast, you’ve done that because you want your customers on the East Coast and the West Coast to experience a low-latency website,” he says. “You want your database to follow suit. You don’t want to think about how to do sharding. You don’t want to think about some kind of explosive replication that introduces delays or breaks the rules of ACID. You don’t want to think about how to do replication between two sides of the countries, where some things are masters and others are read-only replicas or slaves or passive copies.
“What you want,” he continues, “is a single logical database that’s always active everywhere, that’s always consistent, that’s low latency, and that provides you a simple, familiar programming experience, so you can just scale across those two data centers. So when you decide to expand into Europe, you can deploy your web application in Europe and your database just expands into the new data center.”
Building a database that can do that is not a trivial exercise. Database vendors like Oracle and IBM have spent decades making their databases bullet-proof and able to withstand the demanding pressures that its users place on them. The challenge, of course–and the factor that’s driving such widespread interest in formerly exotic database technologies like NoSQL, in-memory systems, column-oriented databases, and others–is that customers are finding traditional relational databases are now too inflexible to support new data types and data volumes.
“Essentially what we’re talking about are very complicated distributed systems,” Proctor says. “We have companies running dozens of machines in multiple regions, or on Amazon. Things get very complicated very quickly. What we’re doing really, as far as I’m concerned, is what’s required to be able to support customers that want to do those kinds of deployments.”
Blackbirds is the culmination of years of work for the Cambridge, Massachusetts software company, which was founded five years ago by Jim Starkey. A veteran the relational database wars, Starkey got his start in 1975 at DEC, where he led the group that created the Rdb/ELN database. He eventually went on to found Interbase Software in 1984, which was acquired by Borland in 2000, the same year he founded Netfrastructure. Netfrastructure was subsequently acquired in 2006 by MySQL, where it became a transactional engine. Starkey worked at MySQL until it was acquired by Sun Microsystems, which was subsequently bought by Oracle.
In NuoDB, Starkey aimed to create a database that didn’t face the same kinds of constraints and bottlenecks around disk, memory, and I/O that have held traditional relational databases back. The NuoDB team “re-architected a view of a relational database that doesn’t have the traditional database locking, that doesn’t have the same kind of traditional I/O dependencies that intentionally breaks the transaction from the storage tier,” Proctor says.
One company that will be using Blackbirds is Fathom Voice, which developed an Internet voice communications application that’s hosted on Amazon. NuoDB was the only vendor that could deliver a single logical database that could be shared across multiple Amazon AWS servers in different geographies, while providing real-time updating and elastic scalability, according to Fathom Voice co-founder and CEO Cameron Weeks.
NuoDB doesn’t have a lot of customers yet, but it has garnered praise from analysts for its approach. Gartner labeled NuoDB a “cool vendor,” while Robin Bloor, principal analyst at Bloor Research, said NuoDB is “the most elegant database I’ve seen in a long time.”
Blackbirds follows Starlings, the version 1.0 release. The company uses the bird theme to reflect the ability of flocks of birds to fly and navigate as a group. No single bird controls the entire flock, which reflects NuoDB’s approach to eliminating single points of failures.