Interacting with customers has led me to realize they spend significant amount of time (investigations by internal IT department or by 3rd party storage performance consultants) diagnosing storage performance issues just to understand what the root cause of the issue is. More often customers end up with weeks of diagnosis in vain with no tangible result and are forced to discontinue diagnosis to tend to other priority items or concentrate on running their business. This leads to multifaceted problems that not only complicate growth of data but storage management as well. So the question is - how do I identify and root cause storage performance issues?
For a long time we’ve had early insights into exactly how our customers consumed 3PAR Storage because sensor data was part of the day one design. Earlier this year we started applying the power of HPE InfoSight AI to all of this data, redefining customer experiences by bringing the same revolutionary capabilities HPE Nimble Storage customers have grown accustomed to. To date, HPE InfoSight has already predicted and auto-resolved 85% of more than 1,500 complex, priority cases across the HPE 3PAR installed base.
Now, we are adding even more AI-driven insights for 3PAR Storage for an advanced analytical approach to identify and root cause storage performance issues. HPE Performance Insight for HPE 3PAR eliminate the guessing and the time-consuming work involved with diagnosing bottlenecks and optimizing application performance. Instead of spending time tuning infrastructure with painful trial-and-error, customers are provided automated details that show the root cause to complex anomalies with application-aware insights.
Available in HPE 3PAR StoreServ Management Console (SSMC) 3.4, HPE 3PAR Performance Insights introduces a new kind of storage management with adaptive analytic software that learns your storage operations and adapts to reporting anomalies accordingly. HPE 3PAR Performance Insights offers machine-learned algorithms that are trained in the cloud and deployed on premise. This capability enables a fast time-to-response and extends HPE Infosight support for dark sites. Following are benefits of HPE 3PAR Performance Insights:
- Identify when performance issue is due to saturation
- Plan workloads better with knowledge of available headroom
- Identify root cause to complex anomalies with application-aware insights automatically.
To help detect performance issues, SSMC calculates saturation levels and headroom availability with respect to contributing factors and reports in easy to interpret graphs in real-time. Figure below provides information about saturation and performance associated with a given HPE 3PAR Storage System. Performance chart (heat map) illustrates periods of time when the hosts must have experienced high storage latency. This information includes community wisdom by comparing performance against other similar systems from the connected 3PAR installed base.
The next generation performance analytics allows you to gauge the system performance in dimensions of Storage Saturation, Storage system performance, and other analytics. These analytics mainly help diagnose and root- cause performance issues in your environment, and also suggest potential remediation actions. For example, recommending moving the workload to another array if an array is saturated and can’t perform with the available capacity.
To support Analytics, SSMC 3.4 has added 5 additional optional panels to the dashboard that helps assess what is happening across storage systems within seconds. New AI-driven dashboard enhancements provide easy to identify insights into most resource-utilized systems and performance-related issues. The Figure below is the enhanced dashboard with new panels.
Following table lists panels that have been introduced and their associated explanation.
|
Panel |
Screenshot |
Description |
|
Performance and Saturation |
 |
This panel shows performance and saturation scores for a single system at a 5-minute granularity over the past 24 hours or the user selected time period. A color coded chart (bolder the color – higher the expected latency) to identify where higher saturation causes some performance issues. |
|
Top Systems by Saturation |
 |
This panel lists the top storage systems ranked by their saturation levels over the past 24 hours or user selected time period. It helps identify a storage system that is closer to the saturation. |
|
Top Performance Outlier |
 |
This panel lists the top storage systems ranked according to their performance severity scores over the past 24 hours or the user selected time period. Red or closer to red indicates performance anomalies detected on the storage system. |
|
Top Volume Hotspots by IOPs |
 |
This panel gives a listing of volumes ranked by IOPs at each 5-minute granularity. Volumes are ranked by weighted average based on number of times they experience high IOPs. It helps identify the volumes that are saturating the storage system. |
|
Top Volume Hotspots by Latency |
 |
This panel gives a listing of volumes ranked by latency at each 5-minute granularity. Volumes are ranked by weighted average based on number of times they experience high latency. Volume at the top of ranking list indicates that it has been experiencing an unusual performance penalty compared to rest of the volumes in the storage system. |
Latency score is a comparison of system configuration, workload characteristics and community wisdom (systems that have called home with performance data)
As Taneja Group says HPE 3PAR Performance Insights analyzes globally to help you act locally and transform complexity into simplicity with newly introduced AI driven dashboard panels.
But wait there’s more…….We give you insights as to where the performance issue comes from and identify the impact in much more detail when I am back next week.
For more information on SSMC 3.4 please refer to Release Notes, Admin Guide and User Guide. For supported HPE 3PAR StoreServ systems please refer to SPOCK.
BhaskarGM
Product Manager, HPE 3PAR SSMC
