Flying Through Clouds 3: Establishing a Flight Plan

By Dan Chilton and Bhavik Desai, NetApp Solutions Performance Engineers

 

As I think about flying through clouds, establishing the right performance methodology is similar to a pilot completing flight school before taking to the air.  Continuing with this analogy, every pilot must complete a flight plan and file it with the local Aviation Authority (FAA in the USA) before each flight. The flight plan must include departure points, arrival points, estimated times, alternate airports, and the number of people on board.  As you plan your private cloud there will be many flights (application workloads) flying through your cloud.  Application workloads map back to the business customers and the impact of delays or outages is different depending upon how business critical the application is. Each application needs its own flight plan.

 

We wanted to share NetApp IT’s approach as we think this model might be really helpful to your Enterprise as you build clouds with clustered Data ONTAP. Thanks to Ken Socko and Stetson Webster for sharing this model.  It’s working great for us so far. 

 

  1. Perform a business impact analysis with all stakeholders including business and IT.  For public cloud providers, this is done with your customers.
  2. Complete a survey to evaluate the impact to revenue, reputation, customer impact, compliance and operational exposure to company of the business application.  
  3. Define Business Continuity Tiers.  “For example Tier 0 is Business Critical, supports direct delivery of goods or services which if interrupted could cause an immediate, major, direct and indirect impact to employee safety, compliance, revenue generation, brand name, and or customer satisfaction” We have five defined continuity tiers (Tier 0 through 4).
  4. IT Service Tiers are then defined with Service Level Objectives to meet the Business Continuity Tier needs.  The Operational resiliency defines in detail what happens after an unplanned outage and resumption within the Production site.  Disaster recovery plans for disaster and defines expected service level at the alternate site.

 

 

 

OTO =  Max downtime before production back up

OPO = Max time period of data loss allowed from an outage

RTO =  Max downtime when DR site is activated

RPO = Max time period of data loss allowed when DR site is activated

 

 

See the NetApp 360 blog NetApp on NetApp to see NetApp’s approach to running a business on NetApp.

 

The nice thing about this model is the flight plan is in place before the flight takes off.  The business unit customers agree to the expected cost, times and delivery of service and then IT has the freedom to provide that service level with whatever means available.  It also maps costs to service levels.  This keeps business units out of the hardware worries, making ITs life much more pleasant. If the business unit needs higher service (first class seating), they can pay for it and IT can provide it.  I always want first class seating, but unfortunately my wallet just doesn’t agree with the cost of it.

 

Now here’s where we get to the beauty of clustered Data ONTAP for your private cloud solution.  We can meet all of these needs in one cluster, with one operating system, with multiple storage architectures to provide the right level to each tier.  That’s right, one solution!  Finally, let’s look quickly at the Service Platforms.

 

  • Sp A –     FAS with all SSDs (All Flash FAS)
  • Sp B -     FAS with Flash Pool with SAS drives
  • Sp C -     FAS with Flash Cache with SAS drives
  • Sp D -     FAS with Flash Cache with SATA drives
  • Sp E -     FAS with SATA drives

 

Build out your cluster with some of each service platform and watch your flights cruise through the clouds!  How do you build your private cloud for performance?