Basho brings Apache Solr search to Riak in enterprise overhaul

A new version of the Riak Enterprise NoSQL distributed database is now available, with better search and more data types designed to broaden its appeal.
Written by Toby Wolpe, Contributor
Stuart McCaul: Making Riak very accessible to general software engineers. Image: Basho

Database firm Basho says newly-available Riak Enterprise 2.0 will open the distributed NoSQL key-value data store to a wider audience and to new use cases, with extra data types and integrated Apache Solr search.

Additional distributed data types in the new enterprise-supported version of the database, which Basho created and continues to develop as an open-source project, now include sets, registers, flags and maps, which are designed to simplify application development.

The company has also replaced Riak search with the Apache Solr engine, which it says will allow integration with a wider range of external software through APIs.

"These new Riak data types for the first time really make Riak very accessible to general software engineers," Basho EMEA managing director Stuart McCaul said.

"Previously, software engineers have had to think about distributed systems and eventual consistency to get the benefit of the high availability, scalability and fault-tolerance of Riak.

"Now if they use the Riak data types, we've taken that extra step away. They no longer need to think about eventual consistency to take advantage of the database."

McCaul said the introduction of Solr is also drawing more people to Riak.

"We've replaced our own home-grown Riak search — which frankly wasn't as scalable as our customers required — with Apache Solr, which is very widely known, widely liked, widely used," McCaul said.

"Using Riak as a distributed framework for Apache Solr, we've lent an already powerful search tool the benefits of fault-tolerance and scalability and high-availability that Riak has.

"That new capability in Riak has really changed the kinds of things that people will think about using Riak for and make it much easier to reason about how Riak would fit with a lot more use cases and much more widely applicable."

McCaul said the combination of the new data types and Solr remove the need for Riak users to write their own document structures, using secondary indexes or MapReduce for querying.

"They can take advantage of the map. It will synchronise itself across the distributed system, even in the event of a partial system failure, and it will automatically be indexed for them by Apache Solr in Riak, so it's immediately searchable," he said. "We've taken away a lot of the manual effort, a lot of the difficulty of really using Riak."

Riak Enterprise 2.0 now also offers authorisation and authentication for the first time, enabling firms to verify a user's identity before they are allowed to connect to the database and to verify that a user or group of users has rights to access to particular data and perform certain functions.

"In Riak previously our best advice regarding security would have been to speak to your security officer. It was really about putting it behind a firewall: 'Don't expose Riak to the internet and then implement your standard corporate policies around the database'," McCaul said.

"We weren't offering anything previously in the way of security internally to the database. Now we have two primary features."

Consistent with Basho's aim of simplifying Riak adoption, Enterprise 2.0 offers revamped configuration management.

"What we've had previously is a database that's been very easy to operate. It installs in a few minutes. If you downloaded it now, you could follow the documentation and actually be reading and writing data to Riak in five to 10 minutes' time. And we know that's true because even our sales people do it," McCaul said.

"What we've added in Riak 2.0 the ability for generalist sys admins in the enterprise to do advanced configuration management in Riak. That's important for us because increasingly the developers are choosing the right tools for their job. Then it becomes the issue for the ops team to take care of whatever the developers have chosen. We want to make sure there's no friction."

Also included in the new version is tiered storage so some frequently accessed parts of the data store can sit on flash drives, with the rest on less expensive disk storage.

Earlier this month the UK health service announced it had migrated to a Riak Enterprise-based Spine 2 patient record platform, which is being used by 21,000 organisations.

"In Europe the NHS story is the validation of Riak in enterprise use cases. It says if you're running a safety-critical or revenue-critical system, you shouldn't be afraid to take the leap," McCaul said. "Until now, we've still been at maybe the early-adopter stage. I'd like to see this technology go mainstream now. In the next 12 months we might see that."

More on databases

Editorial standards