Why NoSQL hasn't become the mainstream technology that Couchbase CEO Bob Wiederhold predicted - yet
Wiederhold says 2016 will be the year of 're-platforming' but are NoSQL vendors struggling to convince end users to take the plunge?
Making predictions - particularly in IT - is always a risky business. If they don't come true, it can make an educated guess look like a haphazard statement.
That's not to say Couchbase CEO Bob Wiederhold's prediction that 2015 would be the year that NoSQL left the labs and test benches and emerged as a mainstream technology was completely wide of the mark, but Wiederhold has now had to move the timing back a bit.
"We think it's going to happen in the first half of next year now - it's always hard to predict," he tells Computing.
Back in January, Wiederhold told Computing that going mainstream would mean that the firm's biggest customers would start a broad "re-platforming", that is, moving away from the relational systems that have for years supported enterprise applications, meaning SQL databases such as Oracle, Microsoft SQL Server, MySQL and the dedicated hardware on which they run, and onto the new solutions that in many cases are able to do the job as well or better on cheaper commodity hardware.
So why hasn't the re-platforming started in 2015?
Wiederhold suggests that it is a pretty big deal for organisations to take the plunge with NoSQL and that some patience is required.
Back in January, he admitted that Couchbase needed to work on a number of features in order to meet customer demands, but once these were in place customers would begin re-platforming.
"They're telling us, ‘you need to have these 12 or 14 features before we can [deploy NoSQL to support mission-critical applications]," he said at the time.
Wiederhold believes that Couchbase's latest release, 4.0, which is set to launch next week at Couchbase Live New York, will meet many of those demands, but he accepts that the firm can't satisfy every possible customer requirement.
"The first few customers that entered phase two [where enterprises start to adopt NoSQL for mission-critical business applications] with us are the first few to enter phase three [re-platforming] with us - we understand their needs very well and we think this 4.0 release gives them what they need to broadly deploy - but of course this won't happen overnight," he says.
"[Our customers] are saying things like ‘we're spending $40m with Oracle today, we want to take that down to $10m and broadly deploy NoSQL technologies'," he adds.
Another area that the firm has been working on is its service level agreements (SLAs), which it aims to improve in two main ways.
"If [a customer] calls you up at 2am and they're having problems with their app or their database, and none of our support or engineering people know how they're using Couchbase then we can't move as fast - so we have built up expertise in the customers' applications.
"Secondly, if at that point it is going to take another five hours to get the log files, to get the information that we need to get to de-bugging the problem, then that will also slow things down - so we're making it clearer for the customer about what information they need to give us so we can start to work on it," Wiederhold explains.
Wiederhold says open source NoSQL technology offers end users a model that they aren't tied into, as opposed to proprietary technology from the likes of Oracle. However, Couchbase has competitors within the NoSQL space, too, with none bigger than MongoDB, a company that is valued upwards of $1bn (£642m), at least by those with industry knowledge.
The firms have been going head-to-head for several years now, and back in June, MongoDB revealed the results of tests using the Yahoo Cloud Serving Benchmark (YCSB) at MongoDB World. The tests found that its latest version, 3.0.1 with WiredTiger, was ahead of both Couchbase and a third competitor in the space, Cassandra, for scalability. (One of the main criticisms levelled at MongoDB was that it struggled at scale.)
But Couchbase immediately fired back with a test of its own, in which benchmarks carried out by independent firm Avalon LLC showed that Couchbase Server outperformed both Cassandra and MongoDB.
The general perception coming out of MongoDB World this summer was that the acquisition of storage engine WiredTiger has eradicated many of the issues that MongoDB had with scalability. But Wiederhold doesn't believe it has made a substantial difference.
"We think they've taken a small step forward, and we are still moving forward with our scalability and performance very rapidly, and so the fundamental differentiation between our products hasn't changed in any meaningful way," he says.
Wiederhold says there is no argument that the most popular NoSQL database is MongoDB and that it has the biggest community as a result, but he claims that The Washington Post didn't engage with Couchbase and therefore can't have properly evaluated the product.
"I read the article. We were not involved in any evaluation there, so they must have done evaluation on their own, or not evaluated Couchbase very deeply," he says.
"If you don't require much scale or high performance, and so if the developer features and community are your criteria, then you may go for MongoDB," he adds.
Wiederhold puts much of MongoDB's success down to its focus on developers who use the product at home and then bring the solution to the workplace for lightweight projects, rather than for mission-critical applications. He calls this "phase one".
"Once phase two hit, which was in 2013, the criteria for success changed dramatically as you had to have scalability and performance, and that's why Couchbase took off and why Mongo really hit a wall. Their growth dramatically slowed, they had massive changes in their management and their growth levelled off because they didn't have scalability and performance," he says.
Wiederhold believes that MongoDB has problems that are "fundamentally architectural" rather than any specific features.
"We have had a lot of customers that started out at Mongo that switched to Couchbase, and DataStax have had a lot of customers who have switched from Mongo to Cassandra because of scalability and performance," he says.
"So far they have done one thing, which is to buy WiredTiger - that's still not their default storage engine - that's a very critical piece. Their scalability problems are a different set of problems that they have done nothing I am aware of to fix," he adds, describing the changes MongoDB has made as "cosmetic".
"What they've done is set up an admin console and provided some admin operational things to try and hide some of their complexities and problems," he claims.
Both vendors, along with DataStax, are aiming to be become better at monetising their products. MongoDB CEO Dev Ittycheria told Computing that he wants to make a fifth of MongoDB's users paid-for - an ambitious figure, considering that at the moment it's less than five per cent.
Teaming up for 'more than just logos on the website'
Wiederhold says Couchbase's partner ecosystem is an increasingly important facet of its business. The firm has started to work with a number of the big systems integrators, and is also working with other software vendors.
It has partnerships in the Hadoop area with Cloudera, Hortonworks and MapR, in the cloud services area with Amazon, Microsoft and IBM SoftLayer, and with business intelligence firms Tableau and Informatica, among many others.
Wiederhold admits that Couchbase hadn't fully focused on building deeper relationships with other software vendors before, but that it is now concentrating on ensuring that it has these relationships because it is yet another aspect that end users find important.
This, along with the release of 4.0 and improved SLAs, builds into where Couchbase wants its customers to be - what Wiederhold calls phase three. That is the phase that he now expects to begin in 2016.
But the fact that phase three is apparently kicking off a year later than he expected suggests that NoSQL end users aren't yet fully convinced that the technology can indeed replace the relational systems that are still holding strong after decades of use. Whether or not the likes of Couchbase, DataStax and MongoDB can convince customers otherwise will be intriguing to see. If they can, then they will become multi-billion dollar companies in their own right, although it is unlikely that all three will succeed in the same space.