IMPROVING NETWORK DIAGNOSTICS USING PROGRAMMABLE NETWORKS

Abstract

Network Diagnostics (monitoring, debugging and testing) in data centers has always been difficult. The problem is only getting more challenging with link speeds reaching 400 Gbps, the number of endpoints crossing 100K and data center topologies getting more complex. Recent studies have noted that network faults are extremely hard to diagnose due to their transient nature and in-ability to correlate events. Recent advances in programmable networking have led to better control, management of networks in the control-plane and data-plane. In this thesis, we study how network diagnostics of data center networks can be enhanced by leveraging programmable networks. We propose: 1) DPTP: a time-synchronization protocol to enable synchronized measurements in the data-plane, 2) SyNDB: a framework which leverages DPTP to provide synchronized packet-level network-wide recording of events at data-plane to perform debugging, and 3) BNV: a scalable framework of creating arbitrary topologies for network experimentation and testing.