I was originally going to write about how OpenStack can enable the enterprise, but after hearing a lot of OpenStack discussion at the OpenStack Summit and Cisco Live, it's clear to me that the industry has plenty of examples proving OpenStack can enable enterprise IT. Now, it's time to discuss how it makes it better.
Some may think I'm too optimistic about OpenStack, but I have the unique vantage point of joining the OpenStack community from the opposite direction of most of the people involved. My background is enterprise hardware and virtualization -- specifically large-scale, service-provider grade shared infrastructure. I'm not a code contributor, AWS developer, application developer, or DevOps practitioner. I've spent a career building the platforms that run enterprise-grade applications and workloads to support large companies.
Today, every enterprise has a significant number of legacy workloads that must be supported, but they have increasingly been moving -- especially on the customer-facing side -- to a lightweight, web-scale deployment model. These new applications have different challenges than SAP, Oracle, and Exchange workloads, and they are delivered by different teams employing different methodologies on to different infrastructure, often via public clouds.
Thus, the challenge has moved from "How do we deploy and manage these new kinds of applications?" to "How do we integrate that process into our operational model, so that enterprise IT as a whole improves?" Enter OpenStack and my top five ways it can make enterprise IT better:
1. Extend AWS investments
You probably have developers in your company using (and paying for) AWS right now, whether or not you know it (or admit it). The fundamental challenge is that their consumption model (programmatic, API-driven, use and dispose on demand) doesn't match up with a traditional IT procurement and provisioning model, causing issues at many levels.
OpenStack can extend the AWS consumption model and all the skills that your developers have learned by making internally hosted and managed infrastructure available in the same manner. Randy Bias, CEO of Cloudscaling, lists this as requirement No. 4 of enterprise-grade OpenStack in his excellent series of blog posts.
2. Leverage enterprise hardware
The enterprises I work with have varying tolerances for change, and in some part that tolerance is driven by how much investment needs to be protected. Hardware is expensive, and though change can happen quickly, amortization happens on a fixed schedule.
Rather than looking at new application models as an either/or proposition or as a completely new operational model to manage, OpenStack can take advantage of your hardware, staff that administers that hardware, and processes used to manage those assets. The list of hardware companies that are actively involved in OpenStack (even if it's just contributing drivers) is long and growing.
3. Help drive IT transformation
Don't underestimate the amount of money that large enterprises are willing to spend in order to drive more efficiency into their operational model. The only way to do more faster is to become more efficient, and to use a cloud-first model that embraces both on-premises and public cloud deployment models.
By standardizing on a cloud management and deployment model, enterprises can start winding down the silos that have been created around hardware and its associated vendors. It's amazing how misaligned those legacy silos are with the business process they are supposed to support, and the efficiency gains made here can be significant and lasting.
4. Vendor negotiation
Yes, I realize this is a cynical statement, but it's never been truer. I'm not a fan of the uncertainty involved in the "VMware tax" idea, and the maxim that nothing is ever really free has certainly been true in my experience. However, OpenStack has become popular enough to be a significant factor when enterprises are negotiating enterprise license agreements. With this many players in the private cloud space, customers are learning how far Microsoft and VMware are willing to discount in order to compete.
5. Prepare for tomorrow
Enterprise IT faces many challenges from many directions. In addition to the financial, business alignment, and operational efficiency struggles we've discussed earlier, challenges with staff retention, development methodologies, architecture adoption, learning curves, and even basic troubleshooting are all increasing as the enterprise pivots into a new cloud era. Yesterday's hardware vendors have become today's liabilities. Yesterday's applications have becomes today's boat anchors.
Meanwhile, today's development methodologies and workload patterns have become the pattern for how IT will operate in the future, and that's where the best and brightest in this industry want to be. For every server or disk hugger out there who is wedded to his Java-based GUI and annual trip to Vegas, there are dozens of people more interested in tackling the interesting questions around what the cloud becomes when it grows up and how to make every IT process align with the horizontal business policy it's designed to support. For every virtualization administrator who is content to sit and let the industry come to him, there are many who are trying to help decide where the industry is going.
OpenStack, even with its infighting, politics, vendor meddling, and open-source soul, is a huge part of that process; embracing it gives enterprises a front-row seat and a license to participate.
There's a long way to go, and there will be bumps ahead. Still, OpenStack is poised not just to be relevant in the enterprise, but to be extremely good for it. We are already seeing those at the front of the adoption curve jump in and find that the water isn't as cold as they feared. How far can it go? We're about to find out.