All-Flash arrays enable applications to deliver real-time, instantaneous results to the applications using them. But to make sure those applications are getting the full benefit of the all-flash array requires an infrastructure that is also performing optimally. Real-time analysis is critical in diagnosing performance problems before they occur to make sure those using those applications never notice a problem.
The challenge with monitoring an infrastructure supporting all-flash is every bit of data has to be managed all the time. Many network monitoring tools only sample data at specific intervals. The problem is the flash array is so fast that only monitoring at specific intervals means the problem could have come and gone before the next interval period, even if that interval period is every 10 to 20 seconds. External tools that poll the switches for a given set of conditions are at a severe disadvantage.
One way to capture this real-time information is to capture and analyze the data on the switch itself. Of course, continuously monitoring the infrastructure creates its own challenge; how to perform this ‘on switch’, data capture/monitoring without negatively impacting the latency and performance of the network itself. Avoiding latency and performance impact requires dedicated silicon in the switch itself.
Cisco, as we discuss in our ChalkTalk Video, “Managing The All-Flash Infrastructure for Optimal Performance”, is integrating this monitoring and analysis capability into its entire switch portfolio, along with dedicated silicon to make sure that it does not impact latency and performance. The result is a simple-to-use analytics platform that enables IT to diagnose problems in real-time, before they impact application response times.