SASC report too blunt an instrument to address Accumulo


Language in the Senate Armed Service Committee's fiscal 2013 national defense authorization bill  report regarding Defense Department utilization of an open source NoSQL database may have unintentional bad side effects.

The report would require the Defense Department chief information officer to certify that the National Security Agency-developed  Accumulo database is either unique or a successful open source project--and prohibit DoD components from utilizing it if neither condition is met.

The language, its authors stress, is well-intentioned. There is a legitimate worry that by not utilizing other Big Table-derived open source databases such as HBase or Cassandra at the center of active open source ecosystems, the intelligence community will lock itself into a government-unique solution that's isolated from advances made elsewhere.

The fact of Accumulo code being available as open source isn't by itself sufficient, they say, since unless supported by an outside community of users, it would remain specialized, and therefore de-facto expensive government-developed software.

But, those with technical know-how have lately pressed their case that the unique features that make Accumulo valuable to the intelligence community – particularly, it's cell-level security, permitting genuine multilevel security – couldn't be added to existing Big Table open source projects without considerable difficulty.

However, with the "big gun being pointed at Accumulo" that is the report language (as a member of the Apache Foundation Accumulo Project management committee put it), potential Accumulo users may be now avoiding it. The report language, which meant to ensure that DoD didn't lock it itself into a lower user-base solution, may be creating that very condition by casting doubt on the long-term viability of Accumulo – in effect, creating a negative feedback loop.

True, the report language would permit the DoD CIO to remove all doubt by making a simple affirmation that Accumulo has unique features and is therefore needed. But that stricture, while simple-seeming enough, injects enough politics into the act of using Accumulo as to make DoD employees nervous, thereby perpetuating the negative feedback loop. The condition of uncertainty is worsened by the fact that it's an election year, and that passage of the national defense authorization act is in doubt this year. A final bill, if it emerges, may very well exclude that language from the final report – but it's very possible that there will be no final bill this year.

By observing nature, we change it, it's often noted – and in this case the Senate committee, by wondering aloud whether Accumulo is indeed the best solution may have doomed it not to be. - Dave