Briefing Note: Brocade Analytics Monitoring Platform
Flash has solved a lot of the storage performance challenges that the modern data center has but it is not a cure-all. The problem is that most organizations are ill-equipped to resolve the issue beyond just throwing flash into the storage infrastructure and hope it resolves the issue. The good news is that putting flash into the mix does help but the question remains, “are you getting the most out of that investment?”.
What’s needed for the infrastructure to provide feedback on I/O activity so that maximum value can be derived from the infrastructure. Brocade recently announced its Analytics Monitoring Platform (AMP) to arm IT professionals with the data they need to extract maximum value from storage investments.
The Trouble with Troubleshooting
One of the reasons why the “throw flash at it” mantra has been so popular is that diagnosing storage network performance is complex and extremely time-consuming. It is often a manual, methodical process that tries to diagnose something that is changing in real-time. Accurate diagnoses require end to end monitoring including components that are abstracted thanks to virtualization. Lack of end-to-end visibility and real-time data gathering means that the IT professional is trying to complete a 1,000 piece jigsaw puzzle with half the pieces missing.
Solving the Network Monitoring Challenge
Brocade’s new Analytics Monitoring Platform (AMP) is deployed with pre-configured policies and rules that can be adjusted as needed. It can automatically detect and recover from errors, and it can validate infrastructure moves, adds and changes before they are implemented. Most importantly, it deploys non-disruptively into the storage fabric in minutes.
The AMP is a 2U form factor platform that has 24 Gen 5 fibre channel ports and processors dedicated to analyzing the data being collected. It runs Brocade’s Fabric OS and is integrated into Brocade’s Network Advisor. It analyzes all network connected devices and provides I/O performance and traffic statistics, both current and historical for each device. Most importantly it provides visibility into virtualized environments for complete end to end troubleshooting and optimization.
Brocade’s AMP collects a wide variety of metrics including:
- Read/write I/O latency and first response times
(including both average and maximum values)
- Read/write IOPS transfer rate
- Other command latency stats (reserve, release)
- I/O queue depth (pending I/Os)
- Protocol error stats (partial list)
- Check conditions
- I/O aborts
- SCSI reserves, releases, reserve conflicts
- Fabric latency
The key is knowing what to do with these metrics as you get them, and Brocade’s AMP as well as their professional services team are there to help. Also, Storage Switzerland will be creating a series of blogs that will discuss each of these metrics, what they mean and what they are trying to tell you about your environment.
Randomly throwing flash at storage performance problems is only a temporary fix. Eventually users will want even better response time than the blind implementation that flash can deliver and many data centers are experiencing that now. We’ll have to return to end-to-end performance analysis, solutions like Brocade’s AMP go a long way toward helping IT professionals navigate the sea of performance statistics to provide real-time answers to storage performance challenges.