Summary
Storage I/O performance monitoring provides advanced insights into network traffic, and these insights can be used to accurately solve network congestion. Cisco SAN Analytics, which takes a network-centric approach to storage I/O performance monitoring, provides end-to-end visibility into I/O operations between virtual machines, initiators, targets, and LUNs/namespaces. The per-flow performance metrics from SAN Analytics help in determining network traffic patterns. For example, the throughput on a port can be predicted by using the I/O size of the read and write operations. Also, most throughput due to read I/O operations is in the direction from storage (target) to hosts (initiators), whereas most throughput due to write I/O operations is in the direction from hosts to storage. Although the read and write I/O data frames make the most of the traffic, these data frames are just a consequence of the read and write I/O command frames that are sent from the hosts to the target. These details help in detecting and predicting congestion issues, and they also help in preventing them by using mechanisms like Dynamic Ingress Rate Limiting, as explained in Chapter 6.
This chapter explains the practical usage of SAN Analytics via four case studies. The steps explained in these case studies can be reused in other environments for detecting and predicting congestion issues.
Finally, storage I/O performance monitoring and SAN Analytics are detailed subjects, and these tools can achieve a lot more than detecting and predicting congestion in storage networks. We recommend continuing your education on this topic outside this book.