Is your enterprise cloud ready for the big leagues? Start with Triple A.
So your organization wants to be more agile—aspiring to spin up new test environments in minutes, speed development processes, and scale infrastructure with ease. Your reflex is to move applications to public cloud to gain agility, but that isn’t always the right option. Sure, some cloud-native applications are well suited to public cloud, but other enterprise applications are not. The cost of recoding, refactoring, and building resiliency into those applications before migration can be prohibitive.
But those applications (which are often business critical) demand agility too.
That’s why many organizations are pursuing an enterprise cloud model. This platform uses a building block approach similar to public cloud, allowing it to snap into existing infrastructure, add web services, and scale quickly.
For example, one of our customers has built an enterprise cloud platform from which it spins up 120,000 VMs every month, and manages over 3 petabytes across eight data centers in just four hours per week. Most organizations might staff 10 people for such an operation, but with the right building block architecture, this operation uses just one-tenth of an employee’s time—thus freeing up the team to support a total reinvention of the organization’s business model.
So, how do you build an enterprise cloud platform ready for the big leagues? To use a baseball metaphor, you need to start with “Triple A,” otherwise known as the three pillars of enterprise cloud:
1. Autonomous operation. If you want agility, you can’t depend on constant manual intervention. That means the enterprise cloud platform you build has to work autonomously. Stated simply, you should be able to plug it in, and it works.
Now, most solutions tout this level of ease—so how do you put the claim to the test? One simple way is to check if there are LUNs or volumes involved.
If so, you may want to reconsider. They’re not the currency of cloud, and require you to lump together multiple applications into one LUN or volume. Those applications must share resources, and occasionally fight over them. The result is unpredictable performance—and when latency appears, you need experts that can dive in and manually manipulate the environment, shuffling applications between LUNs or volumes to try and restore performance. That doesn’t scale.
Instead, you need a solution that keeps every application separated in its own lane. That way, each application gets the resources it needs without conflict, and you can grow your footprint while guaranteeing performance—and it all happens autonomously.
2. Automation. With your current environment, there are tasks that you need to complete on a regular basis, from provisioning new virtual machines to applying protection policies. Rather than manually manage all these tasks, it’s possible to write scripts that automate most or all steps.
But how easy is it to create that automation? It depends in part on the availability of RESTful APIs. If your platform has a set of RESTful APIs, you can connect the system to tools like VMware’s vRealize Orchestrator or vRealize Automation, and Microsoft SCOM or SCVMM. Plus, you can leverage PowerShell, Python, and other purpose-built technologies to set up and simplify automation.
Putting this automation in place early ensures that the processes and policies you establish for your first 500 VMs make it easy to grow to 5,000 or 50,000 later—without having to add more staff (or hours in the day) to keep a handle on it all.
3. Analytics. This is how one person keeps tabs on a large and growing footprint. Your enterprise cloud platform needs to offer analytics that deliver in two different dimensions.
First, they need to span your infrastructure—that means creating an in-the-moment picture of every application’s performance across compute, network, and storage. That way, if you have a latency issue, you can get to the root cause in a single click.
Second, analytics need to be predictive. Your platform should use all the data collected about historical application performance to forecast future needs. That can also translate into what-if scenario modeling, allowing you to assess the impact of potential changes. And all these analytics must be free of averages and correlations—just real data about any individual application.
Finding an enterprise cloud platform that addresses all three pillars can be difficult, but there are options out there. Look for platforms that assign every application its own lane, have RESTful APIs, and application-level analytics. The best platforms should make it easy to tie them in with the broader cloud ecosystem, and forecast your needs for capacity and performance at a moment’s glance.
Enterprise cloud is a chance to knock it out of the park for your organization. The right platform allows you to manage petabytes in minutes per day, to spin up new applications, and to scale in parallel with business growth. That’s why you’ve got to start with Triple A.
If you’re interested in a conversation about enterprise cloud, we’d be glad to take a swing at it.
Seth Moskowitz is the Technical Partner Sales Manager of Tintri with over 20 years of experience designing, building, and selling infrastructure technology. He has worked in the public sector and then spent over 14 years working for a large financial organization. Seth currently manages the Tintri/SHI relationship nationwide.