During the course of my life, I've seen more than my fair share of double standards (I'm not a young man). When I was a child, my parents frequently told me not to do things they themselves did with some degree of frequency. As a parent, I restored balance to the universe by telling my kids not to do things I did all the time. I've also experienced double standards in relationships, on the job, and in social situations. I've seen so many double standards it makes me...well, it makes me see double sometimes.
I assume most of you have experienced double standards in your lives as well. (Either that or my therapist has been lying to me all these years.) Therefore, it should not come as a surprise when I point out we have double standards in the IT industry. In particular, we have a serious double standard on the issue of server consolidation.
What's the nature of this double standard? Simply put, many people think server consolidation is good when we do it in our data centers, but bad when vendors do it in their product lines.
As is the case with many double standards, some aspects of it are completely valid, while others are utter nonsense. Let's take a look at the issue from both perspectives and see whether we can find some truths in the middle. (Or, if you want to take the parental approach, feel free to mumble, "Life is not fair get over it," and walk away.)
The practice of server consolidation is drummed into the heads of most IT professionals from an early age. The drumbeat of this lesson is simple: More servers, bad. Fewer servers, good. However, to make life more interesting, server consolidation takes many forms, each having its own set of challenges and rewards. I tend to break server consolidation into four forms:
Each form is described in "The Four Forms of Server Consolidation," below. To summarize, a good server consolidation strategy can reduce both capital expenses (CapEx) and operational expenses (OpEx). During these days of fiscal and staffing challenges, that can only be a good thing. No wonder we love server consolidation.
If you remember the halcyon times of the 1990s, the server market virtually was overrun with different processors, different operating systems, and even different vendors. Those good-old days didn't last too long the market couldn't sustain all the available choices and led to the consolidation of a number of vendors, including IBM's acquisition of Sequent and Compaq's acquisition of Digital Equipment and Tandem, which was followed by Hewlett-Packard's acquisition of Compaq.
As a result of these and other market forces, both Hewlett-Packard (HP) and IBM ended up with a range of different server platforms requiring different engineering investments, different supply chains, different manufacturing plants, and so forth. Needless to say, this was neither ideal nor efficient. As a result, both companies went on a quest to gain greater efficiencies in their server portfolios.
In the case of HP, it led to the retirement of popular operating systems, including MPE, RTE, and the OSF/1-derived Digital Unix and Hewlett-Packard Tru64 Unix. HP also converged its processor architecture, first moving to Alpha processors and then to Itanium processors for the surviving operating systems (i.e., NonStop, OpenVMS, and HP-UX).
IBM, on the other hand, took a much longer road to product line consolidation. One of IBM's initial steps in the 1990s converged its AIX and OS/400 operating systems on the RS64 processor; however, the company manufactured separate hardware platforms for them. This set the stage for the inevitable platform convergence of AIX and OS/400, which finally occurred in 2008 after both products had been renamed and rebranded too many times to recount here.
Implementing a common processor architecture across multiple operating systems let both HP and IBM maximize efficiency in their supply chains and manufacturing operations, optimize engineering investments, and reduce support complexity. Additionally, by converging multiple operating systems onto the same processor architecture and platform, HP and IBM can more easily afford to maintain multiple operating systems some of which may not be market leading.
Server consolidation is something we should all strive for within our organizations, and something we should be proud of when we accomplish it. However, and somewhat ironically, whenever a server vendor consolidates product lines and retires (or dramatically changes) one of the platforms we use, we often shriek like little girls on a school yard. (Okay, you personally may not shriek, but I've definitely heard someone shrieking.)
Back to our double standard: We think consolidation is good when we do it and bad when server vendors do it.
Why do we have such a double standard? Do we think only we should have the right to end or alter the lives of servers? Do we feel as if the vendors aren't giving us proper warning as if we're being ambushed? Or is it simply a control issue?
My belief is we are less than thrilled when vendors consolidate server product lines because it represents change for us, and change translates into work. From a hardware perspective, we have to determine the appropriate time (and funding) to move from our current platform to the new platform. From a software perspective, we have to make sure all our applications and utilities will work on the new platform. Oh, and we have to make sure the new platform is compatible with our network and storage investments.
I readily admit this is hard work; however, while we are grumbling about it, we might want to remember the alternatives. Vendors could charge more to justify the investments to maintain low-volume (or less popular) platforms, sticking us with a bigger bill. Or they could simply discontinue the platform and move on with their lives, leaving us in the lurch.
In short, this is one of the rare cases where I side with the vendors. I believe the overall market benefits of product-line consolidation outweigh the pain of change. Of course, vendors could make us feel better about the process if they would pass the savings they receive from increased efficiency. Hey! It could happen!
Sean Chandler is a computer and network consultant who has nearly 30 years of field experience. Astro, a border collie with more than 40 dog years of data processing experience, provides technical support to his master, Sean.
1. Reducing Servers
The most popular form of server consolidation is to reduce the number of physical servers in an organization. The benefits of this form of server consolidation are quite apparent. Reducing the number of servers leads to either reduced staff (not necessarily a good thing for the staff) or increased staff efficiencies. In either case, you are looking at OpEx improvements. Moreover, reducing the number of servers lowers CapEx costs because there are fewer servers to maintain, depreciate, and ultimately replace. A good consolidation strategy will even reduce future server purchase requirements.
2. Reducing Server Configurations
A less-recognized form of server consolidation is reducing the variety of hardware configurations purchased. This form concentrates buying on a fixed number of hardware configurations instead of buying a different configuration for each new project. For example, you purchase only two-socket and four-socket configurations with preset amounts of memory and internal storage instead of uniquely sizing each new server request.
The benefits of consolidating server configurations are twofold. First, reducing the number of supported hardware configurations reduces administration and operational complexity. Simply put, you have fewer unique moving parts to monitor and maintain. Second, reducing the number of hardware configurations lets you focus your purchasing with a vendor. Most vendors will offer better pricing if you are consistently purchasing the same model and the same configuration. It's easier for them to plan than if you purchase a unique configuration each time.
3. Reducing Operating Systems
The same benefits of consolidating hardware configurations also apply to consolidating operating systems. Most organizations have multiple operating systems I regard that as normal. Therefore, there is no shame if you are deploying Windows Server alongside Unix and perhaps another midrange operating system such as iOS (sorry, but I can't bring myself to call an operating system "i").
Unfortunately, many organizations have more complexity than that. They don't have one Unix; they have two or more variations of Unix (e.g., some mixture of Sun Solaris, Hewlett-Packard HP-UX, and IBM AIX) and often additional variations as well (e.g., IBM iOS or zOS and Hewlett-Packard NonStop or OpenVMS). And, of course, many organizations have a generous sprinkling of Linux on top.
Needless to say, maintaining multiple operating system variants (including variant release levels within the operating systems) increases complexity. This, in turn, increases administrative and operational costs. Oh, and it typically costs more money too, because you can't leverage spending on a small number of hardware configurations or even a small number of vendors (more about that in a minute).
Of all the forms of consolidation, operating system consolidation is the most difficult to achieve. The reason is that applications are written for and maintained on specific operating systems. In other words, your application choices often drive the operating system choices, not vice versa. As a result, consolidating operating systems often requires changes in applications, which requires business leader buy-in, which makes the whole process difficult, but not impossible.
4. Reducing Server Vendors
The final form of consolidation is reducing the number of vendors you purchase servers from. The degree of difficulty is directly related to the operating systems you have deployed. In the case of Windows Server and Linux, any of the mainstream server vendors can meet your needs. However, in the case of Unix and other midrange offerings (including the mainframe), your vendor choices are bound to your operating systems. As I just described, operating system consolidation is not always an easy process.
The benefits of vendor consolidation are similar to the benefits of operating system consolidation. You reduce the overall complexity of your environment, you're able to focus more purchasing power on a smaller set of vendors (which should get you better pricing), and you have fewer vendors to manage (your procurement people will thank you for that one).
One final word on vendor consolidation: It is dangerous to consolidate down to a single vendor for all your server needs. If a server vendor believes you are "locked in," you may very well see your discounts erode over time. As a result, I do not recommend you consolidate to a single server vendor.
S.C