Basho aims for enterprise operational simplicity with new Data Platform

 

NoSQL database provider Basho Technologies has announced the launch of its new Riak Data Platform, in an attempt to make it easier for enterprises to integrate with other complementary tools and to introduce greater simplicity to complex operational environments.

Equally, the announcement reflects a market that recognises the need to mature and introduce better data management and automation tools, if it wants to continue to gain broader traction in the enterprise.

Competitors DataStax and MongoDB are attempting to do similar things, having announced either similar integrations or enterprise focused consulting arms – but Basho's attempt to deliver a 'platform' on top of its powerful Riak NoSQL technology will likely sit well with enterprise buyers.

When I last spoke to new CEO Adam Wray, it was clear that this was his agenda. He was aiming to redefine Riak in the eyes of the business buyer, wanting it to not be seen as simply a tool that got developers excited. Back in January he said:

One of the things we have not done very well is listening to our clients and community in how it drives our specific road map. People look at us and have a hard time figuring out what we want to be when we grow up.

The developers all know who we are, but if you talk to the business people it’s hit-and-miss. One of the things we have to do is simplify the product so devops can engage with it even faster and understand the power of what it can mean to use our solution set.

The Basho Technologies Data Platform includes a number of integrations that enterprise customers are already trying to make work with NoSQL databases themselves and consolidates it into an easier to manage, somewhat automated system. It includes:

• Apache Spark Connector – Allowing for real-time, in-memory analytics. It also includes Spark cluster management without having to use Zookeeper (an additional management tool that is often used by companies). DataStax has already announced integrations with Spark, so it's good that Basho is now offering the same capabilities.
• Integrated caching with Redis – This allows for enhanced read performance with integrated Redis caching, which lets users easily replicate and sync data from Riak KV to Redis Cache. Basho claims that the Data Platform will simplify the current manual coding needed for applications and can improve read performance. Again, this essentially aims to improve the stability and speed of operational performance for customers.
• Riak Search with Apache Solr – Solr is one of the most popular enterprise search engines around and Basho's Data Platform now allows for full-text search functionality across the distributed database technology. Basho says that this will provide transparent indexing and querying of Riak data values.
• Orchestration services – The Basho Data Platform cluster management also enables download and deployment of instances of Riak KV, Apache Spark and Redis. It allows customers to auto-detect and auto-restart failed instances to again help optimise operations and improve availability.
• Simplified development using data replication and synchronization – This decreases the chance of data loss should an instance of Redis or Spark become unavailable due to network, system or other connectivity failures. It automates synchronization of Solr and Riak KV indexes and replicates and synchronizes data between Riak KV and Redis Cache.

Basho said that as part of a recent survey of its users, which was conducted by TechValidate, it found that more than 50 percent of respondents indicated that they were using Riak across multiple applications – implying that a data platform would help to reduce this complexity. Equally it found that more than 40 percent have already deployed Redis and an additional 42 percent are using Apache Solr. So easier integrations will be welcomed, I'm sure.

CEO Adam Wray released some canned comment along with the announcement, which indicates that this is a big leap for Basho, but also is just the start of further broadening of the platform. He said:

This is a true turning point for the database industry, consolidating a variety of critical but previously disparate services to greatly simplify the operational requirements for IT teams working to scale applications with active workloads. The impact it will have on our users, and on the use of integrated data services more broadly, will be significant. We look forward to working closely with our community and the broader industry to further develop the Basho Data Platform.

Director of EMEA

Prior to the public announcement of the platform, I got the chance to have a chat with Basho's director of EMEA, Manu Marchal. He reiterated Wray's sentiments and said that the Data Platform was pretty much a no-brainer for Basho on this route to simplicity for customers, as it pretty much consolidates a lot of their efforts into an easier to use system. Marchal said:

If you think of the new generation of databases, which our marketing people like to call NoSQL, they all started as single purpose databases. Riak was a key value store, others were document, or graph etc. As these products evolved they became multi-model, so they were able to do more than just being a single-purpose database.

So for Riak, we have added secondary indices so that people can type data and search it very efficiently. We incorporated Solr, the most used search engine in the world for enterprise use cases. We have also incorporated MapReduce from Hadoop into Riak. And so last year we became a fully multi-model database that is being used for mission critical systems.

Now a lot of customers are relying on Riak for running their mission critical applications and from there they want to do more stuff that goes into Riak, because Riak becomes the core platform for all of their operational data. The other things they want to do are, for example, to cache part of the workload that goes into Riak into a cluster. Sometimes they want much faster analytics than they can do with MapReduce, through something like Spark. So integrating all these products is quite challenging because they each have their own technology, they each have their own clustering and high availability system. That's where the data platform comes in.

Marchal was keen to emphasise that although Riak's competitors in the NoSQL space also rely on

integrations with tools like Spark, he believes that Basho will win out in the enterprise because of its strong operational performance. He argues that other tools may be winning the interests of developers, but the key to long-term success will be making sure that NoSQL databases are stable and easy to use in operational environments.

Marchal refers to customers in Europe, such as the National Health Service in the UK and betting giant Bet365, which are using Riak for this exact reason. He said:

That's why Riak is quite unique in the NoSQL space, it brings predictable performance at deliverable operational cost. The integration with Spark is about delivering that same level of operational simplicity that people have been used to. Yes Cassandra has a Spark integration, but Cassandra is nowhere near what Riak delivers in terms of operational simplicity. We get many customers migrating from Cassandra to Riak because they got so tired of trying to keep their Cassandra clusters up. They migrate to Riak to benefit from operational simplicity and probably double performance.

The road to simplification is a never ending road. But talking to our customers, we are by far the most advanced in operational simplicity. I think operational simplicity is where the focus on enterprise is very important. A lot of people think that Mongo is easy to use because it is developer-friendly. But operational simplicity for enterprise is key – how predictable will the developments remain under the worst conditions? How much support do you need to spend with nodes start going down? Riak is by far the best technology to handle all of these things automatically and increase operational simplicity.

But there is still a lot of work to make this happen. To me, the ideal scenario is when the number of actions required by the ops staff to keep the service delivering is reduced to zero. Are we at zero

yet? No. We are certainly better than anyone else and that's why larger scaling companies are relying on Riak.

My take

The NoSQL market is growing rapidly and attracting the attention of many. But when you see announcements like these, you realise that enterprises need management and automation tools to make wide-scale NoSQL deployments realistic – the Data Platform goes some way to helping with that and it is a sign of a market maturing. It is a market that is shifting from a developer focus to a business/CIO/CDO focus.

Are we completely there yet? Nowhere near. Will there be one winner here? Probably not. And we are almost certainly going to see some market consolidation. But in the meantime, Basho's Data Platform is good news for its enterprise customers.