PingMesh + NetBouncer: Fine-grained path and link monitoring for data centers
Albert Greenberg set the tone for network operations @scale by describing the PingMesh and NetBouncer systems used across the Microsoft network. Finding a network problem across millions of hosts and many global data centers is like the proverbial “needle in a global, networked haystack.” Albert even showed how the NetBouncer system takes the solution to the next level, with automatic remediation of the problem found by PingMesh. This is critical to allow applications to function properly across the network in spite of ever-present, individual network issues. The underlying motivation and theory behind PingMesh are available in a SIGCOMM paper by Microsoft.