No to SQL? Anti-database movement gains steam
But can enterprises take open-source alternatives Hadoop, Voldemort seriously? — Computerworld - The meet-up in San Francisco last month had a whiff of revolution about it, like a latter-day techie version of the American Patriots planning the Boston Tea Party.
Context & Ripple Effects
The challenge to the relational incumbents has been building for years: Business Week profiled the startups taking on the database giants back in February 2006, and in January 2008 the establishment struck back intellectually when database luminaries publicly criticized MapReduce in the 'database gods' dust-up. What changed with this story is venue: a San Francisco meet-up written up by Computerworld means the anti-database movement has moved from engineering blogs into mainstream IT press.
That mainstreaming matters because it drags enterprise buyers into a debate that until now lived among web operators. The open question the article flags — whether Hadoop and Voldemort are serious options or hobbyist tooling — is precisely the evaluation enterprises were never asked to make before.
First-order effects
- Enterprises evaluating Hadoop and Voldemort now have to weigh missing vendor support, mature tooling, and transactional guarantees against scale-out economics that relational products were not designed around.
- Web-scale operators running these stores gain legitimacy: a Computerworld write-up converts their internal architecture choices into a visible industry position.
Second-order effects
- Relational database vendors face pressure to justify per-seat licensing and single-node scaling assumptions in deals where buyers now have a credible non-SQL alternative on the shortlist.
- An integration and support ecosystem forms around the open-source alternatives, since enterprise adoption stalls without paid support — shifting services revenue toward whoever productizes Hadoop and Voldemort first.
Third-order effects
- If the pattern holds, enterprise data architecture bifurcates by workload: relational systems retain transactional systems of record while distributed non-SQL stores absorb high-volume, schema-flexible workloads, eroding the default of buying one database from one vendor.
- Procurement criteria shift from feature checklists toward operational maturity and community velocity, weakening the lock-in that licensed database vendors have historically priced on.
The trend: Enterprise data infrastructure is fragmenting beyond the relational monoculture as web-scale workloads normalize non-SQL storage and force traditional vendors to defend their defaults.