Flying Through Clouds: Managing a Mixed Application Cloud on Clustered Data ONTAP for Performance

By Dan Chilton and Bhavik Desai, NetApp Solutions Performance Engineers


Thanks to Kristina and team for letting us do a guest blog series. Bhavik (follow @bhavik_desai) and I (follow me @dan_chilton) are Solutions Performance Engineers at NetApp. He’s originally from India and I am from Louisiana, which means we both come from hot places and like spicy food. Home for us now is in the RTP, North Carolina ( area.  We love getting deep into the weeds in performance analysis, but for this series, we hope to take complex things and make them simple.  Check out a recent YouTube video we did on storage performance basics, What is an IOP?, to see what we mean.






In this blog mini-series we hope to demonstrate the value of building a private cloud on clustered Data ONTAP with a strong focus on performance management through the data life cycle.  The information we share will be practical, taken from NetApp field guys, performance gurus, and IT that have helped customers deploy and manage clustered ONTAP coupled with our own experience in deploying a private cloud in our lab.  We plan to showcase good methodology and the potency of tools like OnCommand Insight 7 to manage all of this in an effective way.


So what is the data life cycle? 

  1. Purchasing storage - Start with a performance management philosophy and with the controller design at the datacenter and rack!
  2. Architecture - Application Service Level Objectives (Tier 0 – Tier 4) map to Storage Service Level Objectives (Tier 0 – Tier 4) that map to Storage Architectures.
  3. Provisioning - Aggregates, volumes, quality of service (QOS)
  4. Deployment - Applications, VMs, existing data and new data is deployed first to the incubator nodes
  5. DR (Snapmirror) – Designing to provide adequate protection and mitigate performance impact.
  6. Application workload - Multiple concurrent workloads servicing the tenants having zero operational downtime.
  7. Application workload redistribution - On-going performance monitoring, bottleneck identification, and leveraging the secret sauce "vol move" to redistribute the workload to existing and new nodes.
  8. Adding Nodes – Adding new storage for both performance and capacity.
  9. ONTAP upgrade – Non-disruptive operations to provide seamless controller software upgrades. 
  10. Decommission - Removing nodes from the cluster.

Now this all might seem familiar to the point of being boring to you.

So what makes deploying a cloud on clustered ONTAP different than other storage solutions?  

Matt Watts captures it well in his recent vblog “Seamless Scalability, Proven Efficiency, and most importantly Non-Disruptive Operations”.



It is these concepts that we hope to demonstrate in this blog series in a very practical way.

We’d love to hear your feedback, so give us a shout.



Hi Dan and Bhavik,

Looking forward to some interesting view points specially around Performance as a framework for our systems!



dchilton Netapp Alumni

Hi Bino,

Thanks for the enthusiasm, if you have questions along the way, please let us know.


Thanks Dan Performance sizing has always been the best part of being in a Pre-sales role for me ( most challenging as well ! ) and hopefully time and again I might come up with a few questions around architecture and elements on how to get the best out of our system's..