GigaOm is definitely stoking the NoSQL discussion by asking whether scalable SQL databases have taken the momentum out of the NoSQL movement. I interpret recent market events somewhat differently however, and want to offer an alternate perspective.
First, while there have been some who hailed NoSQL as the “replace-all” for relational databases, it’s not clear this was ever a widely held view. In any event, the fact that you hear fewer radical claims about NoSQL as a panacea shouldn’t be read as a sign of slowing momentum, but rather a growing understanding of the role of NoSQL in an overarching data management strategy.
Second, while it’s true that new relational-based approaches are springing up to address RDBMS scalability and performance problems, I haven’t seen any evidence that they are damping NoSQL momentum. There are advantages and disadvantages on both sides of the relational equation, but how much traction these ‘scalable SQL’ databases get remains to be seen, particularly for web applications where NoSQL solutions are building the greatest momentum today. Certainly we haven’t seen any change in the interest and enthusiasm for Membase technology in spite of the availability of Clustrix, VoltDB, Xeround and their like. (And if advisory board involvement is to be taken as an indicator of traction in any given technology approach, we too have representation from Facebook on our advisory board.)
Finally, as highlighted in the GigaOm article, recent NoSQL-related outages are regrettable, but probably not totally unexpected given the newness of NoSQL solutions. As with any new technology, the software is evolving quickly and best practices are still developing. Nonetheless, even in the face of these growing pains, customers continue to pursue NoSQL solutions, leading me to believe that the pain of a non-scalable RDBMS database is much greater than the pain of working with less-proven – but much more scalable – NoSQL technology. Maybe that’s why, despite the publicity around recent outages, none of the affected companies have abandoned their NoSQL projects.
We’re still in the early stages of rapid growth in the NoSQL market. Some of the bold statements from the fringe have subsided, new RDBMS capabilities are being offered by RDBMS companies to address their sizable deficiencies, and there will no doubt be louder voices from the relational camp as NoSQL continues to rapidly get traction. But none of this is an indication of a slowdown in NoSQL momentum. In fact, it’s more likely these are all just signs of the significant change that is afoot in the data management market.