Around the Storage Block
1745829 Members
3955 Online
108722 Solutions
New Article ๎ฅ‚
ENicholson

Tech Preview for Resource Planner, the newest HPE InfoSight Labs feature for HPE Nimble Storage

Overview

One of the most unique and cutting-edge sections of HPE InfoSight is the Labs section. Here, customers and partners can experiment with the next generation of intelligent storage insights. At HPE, data scientists are able to leverage diverse historical configuration data and performance telemetry from HPE storage arrays. This provides an opportunity for thorough analysis of storage usage and trending as well as workload identification. Weโ€™ve been able to hone this process by utilizing the largest available infrastructure data lake in the industry. 

Coming soon, Resource Planner, one of two brand-new HPE InfoSight Labs features, helps to determine optimal workload placement based on simulated impact to array resources and neighboring workloads. Capacity, cache, and CPU needs can be predicted by using the toolโ€™s drop-down menus, which represent curated application workload signatures.  It helps to answer these types of questions often faced by IT staff and decision makers:

 
โ€ข What infrastructure would be needed if I want to expand my application? Will it fit on our existing array and if not, which array should I buy?

โ€ข What infrastructure would be needed if I want to deploy a new application?

โ€ข What will be the impact on current performance if I add a certain workload on the array?

โ€ข How much does each workload contribute to total workload on the array? Which ones are the heavy hitters? 


Now with the addition of Resource Planner, HPE InfoSight can be used to answer these โ€œwhat ifsโ€. Letโ€™s take a closer look at some of the ways this tool can be used.

 

Existing workload predictions

By analyzing similar workloads across many customer environments, HPE has been able to identify certain cause-and-effect relationships. For instance, we know what effect deduplication has on CPU usage across different array models. Using this type of determination, we can easily simulate changes to predict future needs for a specific set of arrays. With Resource Planner, you can easily modify your existing workload by selecting a specific timeframe for analysis, by adding a multiplier to workload or capacity, or by throttling deduplication on or off. 

Planner_screenshot_01.png

This ability to experiment with workload modifications is helpful when trying to predict whether your current storage resources could handle twice the load or capacity. If there are seasonal spikes in storage consumption or needs, for example, you can select that timeframe for the workload modeling, allowing you to more accurately predict that same spike in the future on existing or new hardware. 

 

Application specific analysis

Our time-tested data science models analyze CPU and cache usage on arrays, enabling you to tag and save application workload characteristics. For instance, we know that virtual desktop environments are more easily deduped and compressed than Microsoft SQL Serverยฎ environments. We know that Microsoft Exchange applications tend to have a higher percentage of sequential (as opposed to random) reads than virtual desktops have. Resource Planner incorporates this kind of metadata and modeling into the application categories, which are available in a convenient drop-down menu. 

Here we are going to see what the effects will be of adding an Exchange workload on top of the existing workloads for a particular Array...

Planner_screenshot_02.png

 

Other customizable options when adding new application workloads are total dataset size, max IOPS, bursty or sustained workload, and the ability to toggle deduplication on or off. These modifications help more accurately predict capacity as well as CPU and cache needs based on characteristics provided by the HPE InfoSight analytics engine. 


Planner_screenshot_03.png

 

You can simulate the effects of adding or modifying applications and workloads on capacity, CPU, and cache as seen across different models in the HPE Nimble Storage portfolio. 

Planner_screenshot_04.png

  Planner_screenshot_06.png 

Planner_screenshot_05.png

 

Workload distribution

Different workloads on the array contribute to the total load. Load is calculated as an array-wide metric. Suppose that the array is running a mix of VDI and SQL Server and suppose that the load is 70%. How much of that 70% is from SQL Server? Here we see the effects of adding two different applications on an existing array. In this example, we added a VDI and a SQL Server application workload. 

Capacity needs after adding one new application...

Planner_screenshot_08.png

 

Capacity needs after adding two new applications...

Planner_screenshot_07.png

 

Cache needs after adding one new application...

Planner_screenshot_10.png

 

 Cache needs after adding two new applications...

Planner_screenshot_06.png

 

 CPU needs after adding one new application...

Planner_screenshot_12.png

 

 CPU needs after adding two new applications...

Planner_screenshot_11.png

 

 

Summary

HPE InfoSight Resource Planner is a powerful tool. It helps customers to optimize their environments and determine whether they can add new workloads or applications to their arrays based on existing workloads. Its predictive modeling can also help Sales accurately understand needs and properly size new purchases or upgrades. The strength of this feature comes from three powerful HPE resources:
โ€ข Intelligent use of the storage industryโ€™s largest data lake
โ€ข Strong data collection software on infrastructure components
โ€ข Constant improvement of data science modeling based on many years of exposure to a wide range of scenarios and environments

 

 

 

 

 

 

 

 

0 Kudos
About the Author

ENicholson

Evans specializes in HPE InfoSight and HPE Nimble storage technologies. He has extensive experience with the Microsoft, VMware, and Citrix product portfolios both as an Infrastructure Engineer and in Customer Support and QA Engineering.