Make no mistake of it: vendor lock-in is a business goal. Sure, there are some solid benefits in going with one company for your hardware, software, cloud, and all other needs. But doing so can easily mean missing out on best-of-breed options, cost savings, or simple innovation.
It is important not to confuse ‘vendor’ with ‘framework’. For example, This article specifically discusses “ecosystem lock-in” and argues that adoption of an open-source Infrastructure as a Service (IaaS) platform such as OpenStack that is sponsored by a collaboration of industry giants may be just as confining as going with an enterprise solution. Still others hail open source platforms like OpenStack as the Linux of the Cloud, opening the platform to free innovation and adoption.
Though there is a lot of hype from all sides of the debate, one thing is clear: choosing a single vendor to provide for your needs in the cloud and on the ground can be incredibly confining. No one company at this time dominates all markets in all areas of scalability, performance, agility, and stability. If anything, you are more likely to find a vendor skilled at cornering the lock-in market than the perfect infrastructure market.
Now, let me pull my head out of the clouds a bit and address this issue on the ground level as well. Even in your data center it is easy to find yourself locked into a single vendor. Companies like Oracle and IBM are old pros at it (as evidenced by their sparring over who makes faster, cheaper servers), with combination services including hardware, OS, database, middleware, ERP, BI tools, etc. Set a meeting at the end of the fiscal year with a good amount of money in your budget and see how quickly you can turn your entire architecture into a single vendor’s playground. And I’m not knocking their solutions, mind you. Oracle’s new SPARC T5 processor and servers featuring the latest Intel Xeon E7 processors are nothing to sneeze at. Likewise, their various MPP (or pseudo-MPP) systems like IBMs PureData System (Netezza) and Oracle’s Exadata are absolute powerhouses (Engineered Powerhouses) that would be a welcome addition in any server room. These vendors and others are smart, capable, and have the components that could fill your entire server room and software stack if you so desire.
Beyond the server component are many others; storage, middleware, apps, ERP, CRM, CMS, MDM, the acronyms go on and on. Some companies may be interested in having a “single throat to choke”; that is, one vendor that provides and is responsible for every environment in the stack. However, keep in mind that just because all your components are owned under one roof, they may not be supported under one roof. A call to support will generally not create a meeting of the minds with a representative from each group; instead, phrases like “that’s not our area” or “we don’t support that” will abound, leading to confusion and post-purchase angst.
So what can the IT professional do to help their company either avoid vendor lock-in or to make the best of it should it become inevitable?
Be agnostic
All too often a company will choose to go with a certain technology simply because it’s something they’ve always used. In many cases they will continue doing so even when the technology is outdated and underperforming. The comfort level of old or familiar architecture is difficult to let go of.
Strive to be agnostic. Instead of preferring one platform or software over another, keep an equal playing field. Doing so will allow better flexibility to choose the right component for the right job.
Embrace open source
Enterprise PaaS or other full service solutions have one major flaw: they have to invest a lot of time and money into being the right solution. As I mentioned above, vendor lock-in is a business goal, and money and thought must be put toward locking you in. The Open Source doesn’t have to contend for your business or with the goal of locking people in. It is a platform or software that you can choose to use or not use as you please. Generally this comes with better integration support with a wide variety of hardware/software and can easily be more flexible as a whole than vendor products.
This may or may not apply to vendor versions of open sourced software. Companies such as Cloudera take open source software like Hadoop and build it into an Enterprise package. However, at the core of this package is still an open source software, and many if not all of the benefits will carry forward to the customer along with new benefits like ease of use and incredible performance enhancements (such as Cloudera Impala). This specific subject is still up for debate with many in the community, however.
Don’t let cost be your only decision point on hardware
If you absolutely positively only want the cheapest solution around, build it yourself. If not, you will want to shop around and do a comprehensive comparison including proof of concept (POC) with many vendors. And while you may not be able to afford the solution the folks with the nicest suits bring to the table, you also shouldn’t opt for the cheapest possible route.
Remember that cheap solutions come with cheap support, cheap enhancements, and cheap stability. I’m not talking about commodity hardware; there is nothing wrong with using cheap components in a massively redundant whole. But a cheap overall solution can be very costly in the end.
Deploy outside the box
This topic got a whole article from me a while back. There are software and hardware components out there you may have never heard of that could revolutionize the way you do business. Do your research and get a feel for what is out there. Don’t stop at the traditional vendors. Don’t think of what you have to deploy, but what you want to deploy.
Most importantly, don’t let your current vendors dictate the way things work in computing. Want instant QA deployments of your Oracle databases? Look into Delphix. Want your own cloud platform? Check out OpenStack. Wish you had the ability to quickly provision storage volumes and make use of snapshots and cloning? You may not need an entirely new storage appliance, see if ZFS can do it for you with your current architecture. Many of the companies I’ve worked with have the capability to do so much more with their current architectures, but out of fear, uncertainty, and doubt (FUD) they stick with what they have until the time comes that a huge sum of money must be spent to upgrade the whole kit and caboodle.
Conclusion – Your application needs should pick the vendor, not the other way around
Product development and deployment drives business. If you’re looking for a new MDM solution or building a new application the last thing you need is to be limited based on your vendor. In the end, the key thing to remember is that your business and product development should drive the decision on which platform to pursue instead of having your vendor dictate what you must pursue. There is a tool for nearly every need; some good, some bad. By avoiding a complete lock-in to a single platform, you will be free to choose the components that will help you build the perfect environment.