In a previous blog entry I said that one of the surest roads to business success lies in understanding who customers are, what they want and how best to deliver that. But what happens when customers don't know what they want? This is a bit more awkward; now the organization has to help the customer figure that out. A pizzeria can make that happen with a menu... but most businesses don't have it quite so easy.
Netflix tackled this type of challenge via its famous $1 million Netflix Prize. In 2009, the prize was awarded < http://www.netflixprize.com/community/viewtopic.php?id=1537 > to a group who came up with an algorithm that could accurately predict what kinds of movies Netflix customers would enjoy most. It could do this, in fact, more accurately than Netflix's own algorithm, generating results that were more than 10 percent better. That's pretty impressive given the incredible diversity in taste from one Netflix customer to the next.
Modern IT vendors, whose customers' needs and goals vary just about as widely, have an even more difficult puzzle to solve. Typically, large IT infrastructures at established companies have evolved over time via a process that was more about Making Things Happen Now, and less about a long-term, governed plan of IT optimization.
The upshot is that today, IT workloads are often executed in a way the customer can easily see isn't very efficient or cost-efficient. What isn't quite as clear is how to move to a superior arrangement.
This, I think, explains the growing popularity of self-assessment tools in the IT world. Such tools, offered over the web, give organizations immediate insight into not just their needs, but also the available solutions -- often in a surprisingly accurate way, following a Q&A process.
These tools offer, in a limited sense, free consulting. And if implemented well, they can significantly shorten the path any given organization has to take toward creating a better, more optimized IT infrastructure.Platforms are just tools -- be sure you've got the right tool for the job
So given this context, it was a pleasure talking to Penny Hill, a marketing manager with IBM Software Group who recently helped develop two such tools
Hill reminded me that IBM's focus these days is less on the details of a given platform than on the business value it creates over time. She also suggested that this is an area of �low-hanging fruit,� where organizations can often make rapid headway because they've barely gotten started.
�It's crazy,� she told me, �that organizations continue to argue over the merits of a platform vs. looking at the workload characteristics that are best suited for the right platform.�
That strikes me as a really good point. In the time I spent in IT, platform choice was often taken for granted in advance for all workloads -- relatively low-end x86 boxes running Windows or Linux being by far the most common platform.
Then, based on that assumption, subsequent questions were asked: �How can we accomplish such-and-such on our platform?�
The concept that different workloads have different characteristics, require different resources and are better-suited or worse-suited to different platforms was really never taken into account. So the eventual business outcome was rarely as good as it might have been.
Distributed architectures aren't always the rule, either. At institutions like banks, mainframe computing has often held sway as the dominant platform largely because, well... it held sway in the past, going back half a century in some cases. But organizations should look at their current platform as well as others to make workload decisions
What Hill has recently worked on for IBM are two different tools that give organizations a new perspective on this whole area. If you consider distributed architectures and mainframe architectures as the two fundamental approaches, the next logical questions are: What kinds of workloads are best suited to each? And what kinds of variables should an organization consider to match platforms with workloads in every case?
Hill suggests that this switch in perspective -- from platform-prioritized to workload-prioritized -- has a natural analogy in a familiar area.
�Choosing the best-fit platform should be like buying a car,� she said. �You typically look at the qualities you're looking for, i.e., good gas mileage, safety, Sirius radio, and then search for the car that meets these needs. What you don't do is pick the car first, and then try to force-fit in these characteristics.�A tailored white paper of your very own
This is why both of the IBM assessment tools put the focus directly on workload characteristics -- albeit in very different ways.
The first tool, believe it or not, actually generates a customized white paper. Following a short series of questions on mainframe ownership, workload type, number of users and the relative importance of efficiency, reliability, scalability, security and utilization, this white paper
can be downloaded straight to your hard drive in Word format.
Additional questions might appear depending on your answers to the above. For instance, if your workload involves data warehousing, you'll also be asked the total volume of data in terabytes.
While the white paper is generated based on predefined content created by IBM partner IDC, the content is nevertheless chosen based on your answers, and combined in a way that will more closely reflect your particular IT context than any other white paper you are likely to find.
And as a result, it should provide unusually specific insight into the probable challenges that apply, and provide helpful recommendations concerning the pros and cons of different platforms and workload migration strategies.Interactive assessment: It's what all the cool companies are doing
The second assessment tool
offers an interactive experience based on your answers to three different sections.
The first section lets you define up to five different named workloads; for each, you'll need to provide both the workload's task (analytics, transaction processing, etc.) and current platform (whether distributed or mainframe).
The focus in the second section is on the characteristics of those workloads. For each, you'll need to specify eight different traits -- staff skill level, software license costs, capacity and so forth.
In the third section, you describe the characteristics of your current data center. Here, too, there are eight traits to consider, ranging from floor space to hardware maintenance costs to storage and energy costs.
Once you've finished your self-assessment, the tool then provides results for all your workloads. You can actually see whether a distributed model or a mainframe model is likely to yield optimal performance in each case, based on your specified criteria, via a color-coded model. And if you'd like to adjust your previous answers, to see if the results change, you can do that, too.
I found it interesting, entering different combinations to see what kind of results I'd get. Based on the sample sets I gave the tool, it appears my imaginary companies have invested too much in distributed architectures -- not too surprising, really, given the widespread canard that distributed computing is intrinsically less expensive. Quite often, due to hilariously low utilization levels and frighteningly high energy costs, it's the other way around.
Hill endorses both tools as a way not just to assess your current situation, but also plan for future scenarios. Since the tool lets you enter any values you please, you can test not just the values that apply right now, but those you expect to apply in the foreseeable future.
The results might surprise you -- in a good way.
�Looking at the right-fit platform strategy is often a major mind-shift in the IT world,� said Hill. �But once embraced, it opens the doors to major cost reductions and a smarter, more optimized data center architecture -- put simply, smarter computing.�Additional InformationTry out these workload assessment tools for yourselfLearn more about Enterprise Modernization
Find out how you can experience smarter computing todayAbout the authorGuest blogger Wes Simonds worked in IT for seven years before becoming a technology writer on topics including virtualization, cloud computing and service management. He lives in sunny Austin, Texas and believes Mexican food should always be served with queso
Likes before 03/04/2016 - 0
Views before 03/04/2016 - 6703