Re-Platforming an Application? Start by Modeling Performance

As discussed in a prior blog, NetApp OnCommand Insight can be used to oversee all the storage resources within an enterprise environment—a control tower for your infrastructure, if you will. This time, I’ll review how OnCommand Insight can be used to analyze I/O performance so you can determine if money can be saved by re-platforming an enterprise application.


There are several scenarios where re-platforming makes sense. For example, it can be used to reduce database license costs. Databases often employ a pricing model based on the server CPU type and the number of cores utilized. Re-platforming to newer, less expensive servers, such as moving from RISC to an x86 architecture, can result in substantial savings on software licenses. Similar savings can be achieved by deploying all-flash storage to optimize the I/O path and reduce the number of cores required.


For some applications, the cloud provides another option for reducing costs. But how do you know which applications have the right performance characteristics for the cloud? When re-platforming an application, it’s always better to start by modeling performance rather than taking a trial-and-error approach.


Blog - Enterprise Application Performance.jpg


Application Performance Modeling with OnCommand Insight

OnCommand Insight is an agentless infrastructure monitoring solution that models application, compute, networking and storage performance by capturing workflow data and identifying baselines and trends across your entire data center ecosystem. Since no agents are required, your applications are not adversely affected during the data-gathering process.


The first step in performance modeling is to isolate and monitor application workflows. Applications can be selected for analysis by annotating their associated LUNs, volumes and/or VM’s and hosts via the OnCommand Insight user interface. As assets are discovered, a built-in collection engine begins mapping the end to end service paths, performance and efficiency attributes using a variety of collection methods, including:


  • API calls & CLI commands
  • Log file interrogation
  • SMI-S information
  • Virtual Machine hypervisor information
  • Information from other element managers, such as HP Command View & EMC Unisphere

In addition to complete I/O characterization, the collection engine also logs capacity, CPU and memory utilization, swap rates, network throughput and zoning details —all correlated to individual applications. Once the performance information has been collected, the data is made accessible from the OnCommand Insight client via popular HTML 5 compliant browsers and a fully published Rest API. Performance data is stored in a scale-out distributed Cassandra database that is optimized for big data growth and the rapid retrieval of data sets.


OnCommand Insight’s correlation analytics quickly isolate and identify bottlenecks and datacenter contention for quick root cause analysis. For example, if an application is running slowly, the correlation engine can quickly determine if the storage is over utilized, analyze contention between VMs, and report whether a particular VM or switch port is bogging down overall datacenter performance. Not only are workload victims and bullies easily identified, granular metrics are provided that clearly identify the extent of the victimization or bullying. In addition, multiple workflows can also be combined or stacked to analyze complex interactions as performance ebbs and flows across both physical and virtual constructs - expediting remediation efforts and identifying alternative resources with suitable headroom.


Re-platforming Across a Data Fabric

With the information obtained by OnCommand Insight and the ability to understand the rate of change and I/O density over time, administrators can identify the appropriate server, storage and network resources and apply the correct tier service quality required for any application. But identifying a re-platforming candidate is one thing, and actual migration quite another. The risk of downtime associated with re-platforming critical applications often outweighs the potential benefits, forcing IT teams to maintain the status quo and miss opportunities for financial return.


Fortunately, a NetApp-powered Data Fabric facilitates the easy movement of application data. If an analysis indicates a need for re-platforming, a production application can be seamlessly migrated to flash, disk, or cloud-based storage platforms, as determined by its performance characteristics.


Within a data fabric, NetApp facilitates application migration using the VolMove capability (inside a NetApp cluster) or with SnapMirror (outside of a NetApp cluster). The combination of OnCommand Insight and a data fabric built on NetApp gives organizations the freedom to match application operating costs with their business requirements.


The following video demonstrates how applications can be seamlessly migrated using NetApp OnCommand Workflow Automation and VolMove. To learn more, join us at the annual NetApp Insight conference or contact your local representative about our performance consulting services.



+Added new video demonstrating how applications are seamlessly managed and migrated using OnCommand Workflow Automation