Using high-availability and multi-datacenter clustering and has had 99.9% percent uptime since 2009. Typically, issues are related to third-party network providers or code deployments that had to be reverted. In addition to server reliability, system-wide latency for our core optimization engine is averaging 15 milliseconds and the fully loaded time to plan new routes is approximately 150 milliseconds, which includes writing all route data persistently to multiple databases and search clusters.