Around the Storage Block
1748019 Members
4902 Online
108757 Solutions
New Article
BhaskarGM

HPE InfoSight at the Edge

Are you constantly tuning your storage infrastructure to fix performance issues? Consequently, you try to make sense of various reports, trying to map components (VMs, hosts, volumes and storage system) arrive at a possible/probable cause of the issue to pinpoint where the issue is without success? And for all of this, you have no systematic way of doing it again when things go wrong in the future? Look no further!

Continuing to deliver HPE InfoSight at the edge, HPE 3PAR SSMC is introducing the following features to get ahead of performance bottlenecks, plan workload smarter and troubleshoot issues:

1) Enhancements to Performance Insights

2) Workload Insights

3) Topology Insights

Let me explain how each of these features helps identify root-cause performance issues and plan workloads at the click of a button.

1) Enhancements to Performance Insights:

Back in 2018 HPE introduced AI-driven capabilities called Performance Insights for 3PAR Storage that takes an advanced analytical approach to identify and root cause storage performance issues.  It includes community wisdom by comparing performance against other similar systems from the connected 3PAR installed base. This helps our customers to identify what is the current load on the array, factors that affect performance, identify performance outliers and how to fix them.

We are enhancing it further to provide clear insights in cases why saturation is high but performance is not impacted and in cases why performance is impacted but saturation is intact. You can stop guessing on the root cause as these enhancements provide clear insights on factors causing performance issues. 

And here is how this actually looks like in the management console.

Performance InsightsPerformance Insights

Quite similar, we show you the saturation graph, a performance impact band and tell you which volumes have unusually high latencies in comparison to all your volumes in the storage system. When you mouse over the volume, it will tell you why we highlight it at that point in time and what the issue might be. We show you this information based on YOUR system configuration. Every customer’s configuration is different, the number of drives, nodes, hosts, etc.

We take into consideration data that we learned from thousands of systems that send us telemetry data every day and adjust our algorithms to know what your configuration can provide with regards to performance and where its limits are.

Not only that we go one step further with AI-driven capabilities built into Performance Insights. You can now forecast array saturation for the current workload on the array and understand available headroom in the array.

Saturation ForecastingSaturation Forecasting

The figure above shows the saturation forecast for a storage system by analyzing past data. This shows how much headroom is available in terms of saturation. This also allows us to stack rank systems and that gives you the ability to know which system should take on more workloads and which system is already at its limit. Not just from a capacity point of view but actual resource utilization.

2) Workload Insights:

Workload Insights brings in application awareness and application intent to storage Infrastructure and storage management. This application awareness can be leveraged to monitor and optimize storage Infrastructure for an application instance. Workload Insights implements this through application tagging resource in SSMC. Workload Insights help monitor, optimize and place workloads on the best fit system for an application instance. Following are benefits of Workload Insights:

  • View Performance charts at application instance level: time-series charts of IOPS, service time and bandwidth

Workload InsightsWorkload Insights

The figure above shows a time-series chart for the workload and in this case, it is SAP HANA on HPE Primera system.

  • Top volume hotspots by IOPS and latency panels in dashboard offer information about application instance
APP Volume Set Latency.png APP Volume Set Througput.png
  • End-to-end map view of application instance associated with volumes, system, and hosts
  • Additionally, application tagging through Workload Insights allows Performance monitoring and reporting at the application instance level, associated volumes, and hosts. This helps in measuring how the application is performing.

Workload Insights also helps with workload placement on the best-fit array. It does this by providing a ranked list of systems based on usable capacity and saturation forecast. This helps the user to choose the best-fit array for an application instance. This takes away the guesswork in finding "which is the best fit array for a workload helping you to plan your workloads better”.

Best fit storage for a workloadBest fit storage for a workload

Above figure shows a “Create App volume Set” dialog box filled in and when selecting system a new screen on the right displays stacked rank storage systems based on saturation forecasting (from Performance Insights) and free space. This helps you to place your workloads/applications intelligently on the best-fit array.  

3) Topology Insights:

Our customers find troubleshooting performance issues involving VMs and storage infrastructure in a VMware environment as a grueling task. One possible solution is to map storage components, corresponding performance metrics and figure out where the issue is. Sounds simple but things can get out of control if you want to do this exercise over an extended period of time involving more than one VM and arriving at critical insights is laborious and time-consuming.

Knowing how your system impacts virtual machine operations is the next level of intelligence that we are bringing to you through Topology Insights – on-premise cross stack analytics for 3PAR storage. We deploy the latest machine-learned algorithms to the 3PAR management console and give you insights as to what makes your virtual machines perform better on 3PAR.

Topology Insights provides a bird's eye view of the entire VM stack to help quickly root cause a critical problem. Here is a typical layout of how it looks like.

Topology InsightsTopology Insights

The figure above shows a horizontal hierarchical view of virtualization to storage over a period of time having the following components, associated metrics, and connectivity:

Component Displayed Information
VM

CPU and memory used by the VM

VMDK

Read/Write latency and throughput in ms

Datastore

Device, kernel, and queue read/write latency and throughput in ms

Host

CPU usage % and active memory %

Virtual Volumes

Health status, capacity used % read/write throughput (exported) and latency (exported) and a whole lot of information. 

Storage System

Health status, firmware version, model, capacity allocated %, CPU usage % and latest saturation % as derived by the advanced analytics (Performance Insights for All flash storage system)

Topology Insights also includes a color-coded time slider at the top of the view. Color coding of the time slider is derived based on the health state of storage components, virtual volume performance outlier by latency and system saturation score. You can move the time slider to the desired timestamp and view various components’ performance at that time. This resembles a time machine to go back in time, visualize the complete picture of the topology in terms of performance parameters and connections between components at that time. If a critical application is not performing well, you can go to this view use time slider to figure out what is the issue, since when the issue has been there, what configuration changes have happened which triggered this problem, etc. all of this at the click of a button. How cool is that!

From the "Topology Insights" figure above, we can see that the time slider is positioned at “03 June 19:35 CDT” where it is color coded to red. And the horizontal hierarchical view has one of the components highlighted in red. In this case, the virtual volume needs attention as it is experiencing higher latency as displayed in the figure. As illustrated Topology Insights help root cause performance issues beyond Storage.

Enhancements to Performance Insights and Workload Insights are available for HPE 3PAR to be leveraged today.  Topology Insights will be delivered through SSMC 3.6 in August 2019 supporting HPE Primera and 3PAR systems. Good News! All of the AI-driven features Performance Insights and Workload Insights in SSMC will support HPE Primera systems on Day1 for you to leverage.

In closing these features provide an intelligent mechanism to plan workloads better and resolve performance issues faster that extends beyond storage system. Furthermore, these AI-driven features eliminate the guesswork and time spent on diagnosing bottlenecks and optimize performance. Use them and let us know how it helped you. Remember, all of these HPE InfoSight at the Edge features are delivered through HPE SSMC free of cost and works on-premise (dark site).


Bhaskar GM
Hewlett Packard Enterprise

twitter.com/HPE_Storage
linkedin.com/showcase/hpestorage/
hpe.com/storage

0 Kudos
About the Author

BhaskarGM

Product Manager, HPE 3PAR SSMC

Comments