Interview - Basho CEO takes aim at IoT and edge computing
- Summary:
- NoSQL continues to grow in relevance amongst enterprise buyers and Basho wants to target those making IoT investments.
As more and more enterprise buyers continue to invest in scalable Internet-based applications, the need for NoSQL databases, which cater to complex data models and web-based architectures, becomes increasingly important. It’s a market we have been following closely at diginomica, with the leading vendors fighting it out for market share.
Basho is one of these vendors, with its Riak database attracting the likes of Uber, the NHS and Bet365 as customers.
We got the chance to speak to Basho CEO Adam Wray just before the Christmas break about his priorities for 2017, where he is hoping to make Riak the database of choice for buyers that are investing in IoT deployments and are looking to push more of their data management out to the edge.
Coupled with this, investments in functionality that will attract developers to the platform are also central to the strategy. Basho has traditionally focused on large enterprise customers that needed stability from their distributed systems, but hasn’t got as much of a developer following as say, MongoDB.
In our recent interview with MongoDB CEO Dev Ittycheria, for example, he outlined how Mongo has focused on making the lives of developers easier and thus claimed to have ‘won developer’s hearts away from Oracle’.
Basho’s recent open source release of its Time Series database should go some way to helping compete with MongoDB on this front, as well as attract buyers interested in IoT, according to Wray. He said:
Our big bet, which we made available on open source in the summer, was our time series offering. If you think of Riak, we’ve traditionally played in the key value. So we are in large scale clients that value stability and resiliency, in a distributed environment above all things. NHS medical records, Uber with their dispatch process etc.
That’s all built off Riak key value, based on Riak Core, which is really a distributed routing engine that we can do about anything with. So we built a time series, we beta tested it, and we let it loose in open source in the summer.
Our approach in 2016 was to focus on core scale and performance before features. So we knew that this would have a detrimental effect on incremental initial use. But keep in mind the roots of our company are that we appeal to the high-scale enterprise first, versus front-line developers. So with that in mind, by bringing a purpose built time series database, we are in a position to deliver performance and scale better than anything in the marketplace.
Wray provided an example of a company that Basho has recently worked with, which he didn’t want me to name, but is a large enterprise, where he said that the Riak Time Series offering was able to offer 10x performance on a sixth of the footprint compared to that of competitor database Cassandra. Wray said:
More efficient operational footprint from a CPU/RAM perspective, coupled up with performance, was the first real big thing we wanted to nail in 2016. And we did it in spades.
Now you are going to see a real push walking into 2017 for feature functionality, so that we can appeal to frontline developers. What we are really after is owning the IoT centric portion of the time series marketplace, versus metrics and other areas. If we can nail the IoT, it opens up a lot of partnering opportunities, particularly in one major trend which we are keeping an eye on and are moving towards - edge computing.
IoT and the edge
We are beginning to see a lot of vendors, and some buyers, talking about the growing importance of edge computing in a world where all devices are connected - as there is often a need to place processing and analytics close to the device e.g. machinery, smart cars etc. This model differs from the solely cloud computing world where everything is centralised, but in reality it is likely to be bi-modal and a mixture of the two approaches.
This is where Basho believes it has a sweet spot, as it can cater to both Time Series data, commonly produced in IoT deployments, as well as more static key value stores.
Wray said:
We see the opportunity in edge computing in Time Series as the next big wave. Our clients are not looking only looking at Riak Time Series, but Riak Key Value to help address that. So with Time Series you keep all the IoT data, which is linear in nature and sequential, then on the flip side you need to keep your state of device, profile status etc, which is much more appropriate in a Key Value engine.
We want to break out with a point of view that is not just that we are good for the top 100 to 200 most large scale enterprises, but more importantly, that if you’re thinking IoT, that Basho is a central part of that. And if you’re thinking about the challenges of distributed data at scale and edge computing, then we are definitely going to be within the centre of gravity of that discussion.
We are in another cyclical cycle, where there is way too much compute that needs to happen in the field, if not right on the device itself. So how do we get the best of cloud computing where it is a centralised core offering and push stuff out to the edge? Well, you’re going to need someone that is lightweight, distributed in nature, purpose built for the type of data you’re trying to tackle so that it can actually be processed and managed. That’s what we nail in spades.
I can’t speak for what the rest of the industry is going to do. But I can tell you that we are going to have our piece of that.
Competing with MongoDB
Whenever I speak to one of the main NoSQL vendors, the main competitor they refer to is MongoDB, as it appears to be the one that is gaining the most traction. And that in part is because Mongo has attracted a lot of attention from developers in the enterprise - and because of its capital investments.To compete with MongoDB going forward, Wray is going to improve Basho’s functionality for the developer with Time Series as the focus and is also going to hone in on Basho’s message of being the database for IoT. Wray said:
I think Dev’s strategy for MongoDB is a good one. He has a much broader view and candidly is significantly better capitalised than I am. And so, my strategy had to be - where is my strength? And our strength is in the high end. Let’s leverage that in the first few years, but now we are in a position with time series to finally have a point of view that can be bring mass appeal to a developer population by making it simpler and simpler.
And we want to accelerate that. But I can’t appeal to every vertical like MongoDB can. But we do believe we can nail the front-end developer with regards to making this an appealing option of first choice.
As a smaller company in the tens of millions, versus MongoDB which is $100m-$200m - there is always the risk of being outmarketed. Because I just don’t have the capital to make the noise, or the ecosystem to make the noise. To me, that’s our biggest risk. How we are addressing that is looking to have a megaphone around a singular point of view, IoT for time series and what it means for edge, to break out of that noise.
So instead of having having to spend dollar for dollar, we might be able to spend a dollar for every ten dollars they spend, to be able to own an amount of surface area.
My take
Basho’s niche is that it can now cater to a number of use cases. It’s play for IoT is a savvy one, but it will face stiff competition from MongoDB and Datastax. That being said, it has proven its worth with the likes of Uber and the NHS and has a number of big name referrals that it can play into.
Whether it can make as much noise as Mongo, remains to be seen. Wray does seem to have a firm handle on the technical requirements of what enterprise buyers need at the moment, however, which should take the company a long way.