Establishing an Application Driven Infrastructure for Cloud Ops

When I interviewed at NetApp three and a half years ago, I told George Kurian, our CEO that application developers would rule the world! I maintain this is a fact today.

The public clouds liberated application developers from traditional IT, platforms burdened with slow and expensive processes and costly infrastructure and democratized innovation. Infrastructure decisions on the public clouds were made by application developers and their supporting cloud operations dev/ops or Site Reliability Engineer (SRE) teams.

Before the availability of the public clouds any application idea had to be big! We had to promise massive savings or big revenues measured in millions of dollars, basically any idea had to change the world to get funded. To get our ideas approved we often over promised and more often we under delivered on that promise. It was too hard and took too long to innovate. We operated in an era of Infrastructure driven applications.

Application developers today build and deploy fast on the backbone of continuous integration and continuous deployment (CI/CD) with microservice architectures. They deploy their code through these pipelines to clouds selecting the underlying resources they need through API’s. The Public Clouds have become the default platforms for all new application development.

All of this sounds fantastic, but for the fact that application developers don’t want to have to understand infrastructure details to be able to develop and deploy their code. And really, why should they?  Why shouldn’t they be able to develop their application or service, specify its requirements, and have the infrastructure figure out how to meet those requirements?  Why shouldn’t the infrastructure be able to determine how to optimize performance, availability and cost, even as demands change and evolve? Instead, what happens is infrastructure and money are simply thrown at the problem--in other words, storage and compute are overprovisioned, creating huge amounts of wasted resources. Even on the public clouds, infrastructure drives the application!

We need a solution to this problem that’s not an analytic of your cloud bill, we have to enable Cloud Operations to inject ‘optimization’ and ‘data management’ into the infrastructure decisions, an API that manages the infrastructure resources. Compute and storage account for 70% of the cost of infrastructure today and must be aware of the application’s needs more intelligently. The Cloud Operations team need an Application Driven Infrastructure (ADI). The ADI translates the applications workload patterns and automatically delivers optimal performance and availability while minimizing the cost for storage and compute, all done while maintaining the contracted SLA and SLO. We need to extend the CI/CD model to include Continuous Optimization, completing the process by establishing CI/CD/CO.

For the last three years my work at NetApp has been to drive an ‘ADI’ with storage that accounts for as much as 45% of cloud spend. My team established NetApp on the three public clouds and as the market leader in shared storage, sold by Microsoft and resold by Google. Today our cloud volumes deliver highly optimized IOPS for block and file taking advantage of NetApp’s core IP on all three public clouds running primary and secondary workloads. We provide a rich set of services to the volumes including compression, deduplication, caching, tiering, backups, monitoring and compliance as a service delivering unmatched performance at the lowest possible cost. With Cloud Volumes and Azure NetApp Files we invented ‘Volume Shaping’ an API that changes both the capacity and service tier at run time, matching the IOPS requirements to the workload’s peaks and valleys. We uniquely offer customers the ability right size their storage at run time, both capacity and IOPS as the application needs.

Today I am excited to announce our intention to acquire Spot and its fantastic team led by its CEO and founder, Amiram Shachar. Together, we share the goal to deliver an Application Driven Infrastructure solution critical to the CloudOps teams. Spot manages and optimizes the compute needed for workloads while maintaining its contracted SLA and SLO. It does so by using a blend of compute types: on-demand, reserved and spot instances driven by years of monitoring and AI based decisioning.

Upon the closing of the transaction, Spot and NetApp will deliver savings of up to 90% on compute and storage which account for more than 70% of the public cloud cost. Optimized API’s for storage and compute on the public cloud you choose.

This combination will empower application developers to build, and Cloud Ops to deploy applications faster to the public clouds knowing we at NetApp will continuously optimize and protect them through our Application Driven Infrastructure.