Set in a NetApp Data Center
Sunil – the IT Storage architect
Joe – Storage Administrator
Don - the HR IT liaison
Steve the R&D application manager
Meet Sunil, a tenured IT professional, doing what he does best, performance! It’s a bright sunny day in Palo Alto, but little did Sunil know that dark clouds were on the horizon. After months of working with cross-functional teams to define service level objectives, planning the transition from 7-Mode to clustered Data ONTAP, taking classes on OnCommand Insight (OCI), clustered ONTAP 8.2, and deploying a private cloud, things are really all in place now. It has been a busy year, but now things are running like clockwork.
Six months ago Sunil and the storage IT team migrated a database for the HR department from 7-Mode to Clustered ONTAP. At the time of deployment, based upon performance monitoring, the workload was deployed with a target service level objective (SLO) of:
IOPS: 300- 400 IOs/sec
Latency: < 6 ms.
The HR database was virtualized on ESX, housed on VM#2 and utilized storage on Service Platform “C”, which resides on node storm-03. Sunil just talked to Don the other day and she confirmed that her users were very happy. Sunil keeps a close eye on performance, checking OCI weekly and verifying that the application is performing well from VM to storage.
Adding Flights to the Cloud
Last week Joe, another member of the storage team, added a new workload to the same cluster that houses the HR database. The Research and Development (R&D) arm of the company is developing a new patient imaging system. The new application is unique in that it will allow image collection from multiple imaging devices found in the hospital, from multiple locations. It can also keep all of this data within the same database, instead of needing several different database systems. Nurses and doctors will now be able to view images from multiple devices both locally and remote, at the same time! Before deploying the app, Joe discussed the requirements with Steve the R&D application manager. Steve mentioned that the application is experimental and they do not have a large budget currently to put towards its development. Steve, being an engineer, stated the requirements as such:
IOPS: 300 IOs/sec
Latency: < 20 ms
Joe discussed the options for service level objectives (SLOs) and aligned them to Service Platform “C” (SPC). Joe then deployed the new application on VM#30. He used the concept of incubation to test out the new workload on the SPC node, before fully deploying the workload onto the cluster. Leveraging OCI, he checked which available SPC nodes had free capacity in terms of system utilization, disk utilization, and disk storage capacity. Joe then used vol move to deploy the application on node storm-03, non-disruptively!
All was well for a few days, until today…
A new workload brings a storm to the cloud!
Stay tuned for Part II.