When evaluating the benefits of Infrastructure-as-a-Service (IaaS) and Platform-as-a-Service (PaaS), there are number of considerations to take into account. Some fairly obvious starting points may include the ease of scaling capacity up or down as your business requires, along with the associated costs, and the dependability and quality of service you’ll need. Right behind those considerations is the security of your applications and data, and along with the ability to keep your workloads portable across multiple environments to give you more flexibility and mitigate any vendor or service provider lock-in. Another criterion that will certainly come up is “automation”, however this is an area that typically deserves more due diligence than it gets, and often ends up being a key factor in the long term success of the overall initiative.
Automation is important for the efficient and effective delivery of elasticity, utility pricing models, security services, and most of the other core capabilities you require. Doing automation poorly (or not at all) creates some serious caveats to the success of your overall cloud initiative, like wasted capacity, unnecessary architectural complexity, inefficient manual processes, higher operating costs, and extended time to implement.
One of the key objectives of virtualization and cloud computing is to make highly efficient use of IT infrastructure in a dynamic, real-time way. Underlying that success is the ability to automate sophisticated systems and orchestrate complex processes and arcane configurations on the fly. These are challenging problems, and there is a tendency by many vendors to avoid confronting these problems end-to-end, and instead attempt to address them by partially automating the solution where it is most convenient, or by significantly limiting a particular automation use case. The real world is much messier however. The devil is in the details, and that is definitely true when attempting to integrate and automate new solutions with existing enterprise processes and systems. Unless the proper framework and platforms are in place, you’re going to have a lot of manual steps, disjointed processes, one-off customizations, and patchwork solutions.
What should you look for when evaluating automation for an IaaS and/or PaaS solution?
- Provision and de-provision capacity in minutes
- Use the infrastructure for one use case (like test and development) during the day, and then for grid computing or some other use case at night
- Handle demand spikes rapidly and reliably through “burst” capacity scaling
- Apply layers of security services automatically as needed, depending on the workload’s profile
- Ensure that enterprise policy is always in effect regardless of the cloud or physical solution being utilized
- Provision virtual and/or bare metal environments with equal ease
- Ensure that policy and governance are applied consistently and appropriately in all deployment environments
- Be able to spin up or down hundreds of VMs in an hour or so from a remote access point
- Move workloads from one location to another or from one cloud to another, without manual intervention or configuration
As you can start to see from this short list, “automation” is actually an important success criterion for any IaaS or PaaS offering you decide to build, buy or rent. The better you can automate, the more successful you will also be in moving your IT staff up the value chain. You’ll be able to grow and manage your infrastructure much more effectively with minimal staff which is an important metric in almost all cloud computing initiatives. You’ll find that the best “automation” solutions are able to shelter end users from unnecessary complexity with strong design, good usability, and the implementation of best practices that span broad, real world use cases. The payoff for end-to-end automation is an “Agile” infrastructure that addresses your full range of requirements, and yet can still respond at a moment’s notice to new business opportunities.
You must be logged in to post a comment.