It’s been said that a cloud without automation is like a racecar without a steering wheel: All that performance just goes to waste unless you have a way to control it.
But even with the cloud, the data environment is still so complex and intricate that a fully automated stack is rather elusive, and it is questionable whether complete automation is really the best thing for the enterprise.
To be sure, the list of cloud automation or orchestration (call it what you will) platforms is growing, and for good reason. With the kind of elasticity and self-service provisioning going on, it would take an army of technicians to provide the functionality that modern knowledge workers expect these days. HP’s new Orchestrated Datacenter platform, for example, is supposed to run the gamut from physical resources to the virtual layer and all the way to application workflows, providing service desk, monitoring, security and other functions across multiple cloud deployments.
Clearly, an integrated management stack would be preferable to a series of discrete solutions, although companies like Juniper do make a good case for a dedicated system when preparing legacy environments for cloud-based services. The company’s Brendan Hayes points out that without network automation to link resources together, even the most virtualized data center will lack the agility and flexibility needed to support advanced cloud architectures. Conversion should be a two-step process: First, shore up core switching and routing infrastructure, and then construct a robust automation stack that provides the needed programmability for complex provisioning and dynamic resource allocation.
But how do we square all of this with the new Datacenter Infrastructure Management (DCIM) platforms hitting the channel? As Data Center Knowledge’s Jason Verge reported from the recent Data Center World conference in Las Vegas, top platform providers like ManageEngine and iTRACS are adding new customization and third-party integration capabilities to help the enterprise tie in-house automation capabilities to the broader cloud ecosystem. And some, such as Nlyte, are turning toward SaaS-layer approaches to speed up deployment and provide a more dynamic means of linking the data environment with in-house power and facilities requirements.
Regardless of whether you adopt an all-in-one or multiplatform approach to data center automation, some basic practices still apply, says Dell’s Matt Smith. For one thing, automation requires proper handling of the metadata used for processes and record-keeping; Post-It Notes or simply remembering the correct procedures won’t cut it in the cloud. As well, make a thorough assessment of the time admins spend doing certain tasks. Chances are, jobs that occupy the bulk of their attention are ripe for automation.
The cloud, then, will change the way data interacts with other data—hopefully toward a more efficient and productive work environment. Automation, however, changes the way in which humans interact with underlying infrastructure, particularly those who are responsible for the care and maintenance of that infrastructure.
In the end, it will be easier for users to provision their own resources, but it will be more of a challenge for IT to set and maintain the limits of that new-found freedom.