Stephen Downes

The Canvas hack, writes Tim Klapdor, was inevitable. "When the decree from management increasingly mandates that all core systems must be off-the-shelf products from established vendors (a policy that sounds like sensible risk mitigation) the result is that all your vendors share the same infrastructural single point of failure. When Canvas went down, it took every system routed through it with it." This argument has been made many times in these pages - a distributed and decentralized system is much more resilient. The push toward optimization and efficiency, if taken too far, greatly increases fragility.