In my conversations over 2015, I have found that one of the top of mind goals for many Directors and CIOs for this year is the goal of fully automating the orchestration of the environment. It is a common pain felt across the IT staff, the lack of agility and automation when it comes to provisioning new workloads for the environment.
Whether the plan is to expand the VMWare suite through vRealize Automation, pursue a third party technology like Chef, Puppet, CloudForms, or move into a full IaaS or PaaS environment through OpenStack or Cloud Foundry, the objective is to speed up the auto-provisioning capabilities of the data center to meet the rapidly growing needs for faster, more responsive applications at a quicker delivery time. However, the benefits of moving to automated cloud orchestration, also create new challenges.
Why Orchestrate the Cloud?
To answer this question, let me throw out a scenario that many can probably relate to today. An administrator logs into his Outlook first thing Friday morning, and at the top of his inbox is a request for a new VM from a coworker, who plans to begin testing a new application in the next couple of weeks per the CIO’s initiative. The specifications are there (though whether they are the correct size remains to be seen), and the email is marked Urgent because the coworker needs this application ASAP. The admin logs into vSphere, and begins looking around for a place to begin deploying a new VM for the coworker. Thirty minutes later, the admin thinks he has found the best place for the VM, and begins the deployment process from the provided OVF template. As he glances back over at his inbox, he notices five new Urgent emails from coworkers, all requesting new VMs ASAP, while meanwhile he has received a phone call from a DBA complaining about latency on one of his SQL VMs, requesting an immediate fix. Sound familiar? If only there was a way for those coworkers to have some sort of portal, where they could request their VMs, and have the software build it for them and place it in the datacenter, giving the admin much needed time back to his day to begin planning for that new blade replacement.
The above scenario is exactly why Orchestration is so coveted by anyone who touches the data center, from the admin trying to handle multiple jobs at the same time, to the Director and CIO looking to cut costs and guarantee fast delivery of applications, to the end users waiting in the admin’s inbox for VMs. Technologies like Chef, Puppet, vRA, Ansible, CloudForms, OpenStack and Cloud Foundry meet all of these needs, and more. By providing self-service to the end user, and automating the deployment process of new workloads, we can truly take advantage of the datacenter’s capacity, providing benefits to everyone who touches the IT stack. Faster application delivery means more work can be accomplished, and bridges the gap between the application owners and the virtualization team to a seemingly idyllic harmony.
What’s The Challenge?
Unfortunately, the idyllic harmony that orchestration seemingly provides can often come crashing back down, due to a problem which we have neglected: the issue of application performance. We have given our end users the capability to spin up new applications whenever they want, but we have not changed how we manage them. What the administrator is left with is a more complex datacenter, with more VMs and/or containers demanding more resources from our infrastructure supply, resulting in more alerts, more end user complaints and more outages. It only accelerates faster and faster as we continue to grow out the environment.
Orchestration and auto-provisioning are important components of managing the software-defined data center. But important questions remain: How do I size these new workloads? Where do I place them? When do I need to add more capacity to my infrastructure? Without being able to answer these questions in real-time, orchestration and auto-provisioning only magnify the complexity of cloud environment.
On the other hand, marrying the capability to provision new workloads automatically with the ability to intelligently place and size them in real-time, while helping to plan out new capacity, we can achieve a truly automated, healthy and agile data center for the next generation of IT.