I thought it appropriate in my first blog posting to focus on the area of business technology optimization where we spend most of our time with Fortune-500 clients. A short exposé on the current state of affairs—or to be more precise—the current state of efficiency across Process Tables on Windows desktops/laptops is in order.
Some additional perspective would probably be helpful as I share our findings around a problem so pervasive yet well cloaked, and yet so ironically simple to recognize if sufficiently exposed, that really anyone in business, technology or procurement can instantly understand the enormous implications. Yet the problem subsists gnawing its way into corporate profit margins in an almost insidious repetitive way—costing companies millions upon millions of dollars each year. The story begins within your Windows Process Tables, so basic an issue, so complex and detrimental a problem, so globally endemic as to be simply accepted most unthoughtfully and much to the misfortune of business.
As we consistently tell our clients, yes procurement personnel are certainly doing great work to ensure companies are receiving favorable discounts on new desktops. But here’s the rub! You may indeed own those fast new desktops, but guess what? You really don’t have any control over how those desktops deliver or prioritize delivery of your business-critical applications to your users. More precisely, you don’t have control over how Windows delivers its system services—or the quality of its system services—in support of your business-critical applications. Is this gaping deficiency clear to you? Can you begin to fathom the implications to your business productivity? The very applications you developed or purchased to support your bottom-line on which you perhaps spent substantial dollars, you have little to no control over how Windows treats those applications. Does this lack of control make good business sense? Of course it doesn’t. And most ironically, we’re seeing end-user perception as it relates to desktop workspace and the user experience at an all-time low. Based on our field data, it is the number #1 issue facing the Global 1000. So how exactly is this situation possible when PCs are now so powerful?
Herein lies the problem:
Unlike the days of mainframe computing where finite system resources were allocated on a job-by-job basis—and scrutinized, the distributed desktop systems of today do not include this concept. Windows operating systems remain close to neutral insofar as how applications run on the system. How are you then to regulate system resource allocation amongst the many application processes vying for those finite resources—and all the while these systems are connected to the Internet? This lack of control is extremely problematic to say the least as it directly impacts business productivity. So when your procurement manager comes to you and says…we just purchased 1000 desktops. What you really just purchased are 1000 “application service delivery engines” with no control over how your business applications are delivered to your users. Are you comfortable with this business consideration? You shouldn’t be as it costs most Fortune-500 companies millions upon millions of dollars each year. And from the perspective of business technology optimization, it makes absolutely no business sense whatsoever.
So what exactly does this lack of system control and service quality mean to your business? Well, I’ll gladly tell you what it means: Your inability to dynamically control how your applications operate at a systems-level across a finite amount of chip, disk I/O, ram, network, and kernel, across each and every one of your desktops (local or virtual) is an extremely costly problem. I would argue that it’s the most costly IT problem impacting Global-1000 organizations today. From a BTO perspective, it is a far more important consideration than most IT infrastructure initiatives currently underway, such as the rush to test new VDI architectures.
Our field data indicates that as much as 28% of all running processes are extraneous (unnecessary) with 80 to 85% of these auto-starting, which means always booting up with the machine—always there.
Another 27% of running processes represents the operating system itself.
Another 25% represents infrastructure software (desktop security, patch management, systems management, updates, system backup).
Put another 4 % on top of the above considerations for your IT tools and support, and you have just 16% in system resources remaining for your business applications. So 84% of your desktop system resources have potentially been used up and you haven’t even started using your business applications.
Now take away 6% for your office software (Office, Lotus Notes, Email, Word, and Presentation Software) and that leaves you with just 10% system resource availability to run all your business-critical applications (9% or less if system is infected with malware):
So if you’re a Bloomberg, Standard and Poor’s or Moody’s…it might be financial modeling. If you’re a Ford, GE, or Exxon, it is perhaps unit operations at your production plants. Basically, when you look at the pie chart below, the only things that are making you money are the dark gray slice and the little blue slice. It makes no business sense to put all this unnecessary competition onto the two things that actually make your business money. We see these absurd ratios across most lines-of-business every day. Any technical or business person can look at this level of competition for system resources and realize it makes absolutely no sense.
Even if you are moving towards a VDI infrastructure, quality-of-service problems associated with Windows System Resource Allocation are exacerbated when Guest Process tables are aggregated on VDI servers. System-level inefficiencies that we find across distributed system environments are of course dispersed and therefore subtly more obfuscated and seemingly tolerable; however, when Process tables are aggregated on VDI server blades, resource allocation issues at the Guest Process table-level become more visible. With say fifty or more guests active on each VDI blade, what had been distributed inefficiencies, scattered out in the field, now become concentrated waste right inside your datacenter.
Net-Net:
BTO initiatives are starting in all the wrong places. While there are many areas in the IT Delivery Infrastructure that are in need of re-engineering, one-third of our IT expenditures relate to the desktop, and yet this aspect of our IT services is not even aligned with our business: The magnitude of losses relating to this lack of business alignment is staggering by any measure.
Ziften Technologies with its BLISS product suite (Behavioral Lightweight Intelligence for Stressed Systems) is returning about 25 to 40 man-hours (or more) per-user per-year in additional business productivity across the enterprise. If you multiply 25 man-hours @ 5,000 desktop-users, this equates to approximately 125,000 man-hours per year in additional productivity returned to the organization. What will this additional productivity do for your organization?
It’s also important to note that these gains in man-hour productivity are based on end-user efficiency implications. What these efficiency metrics do not reflect are the IT cost reductions achieved when you better align your distributed systems with business users, such as:
- Reduction in Boot-Times and Shut-Down Times
- Decreased help desk calls
- Reduction in PC reimaging
- Reduction in power consumption
- Longer PC refresh cycles (up to two additional years)
- Reduction in RAM Upgrades
Our goal as a company focused on Enterprise Systems Efficiency and Desktop BTO is to return one week per-user per-year in business productivity to all our customers. What would this additional productivity mean to your organization?
In conclusion, better management of distributed system resources and their allocation in support of business-critical applications is where organizations must first and foremost focus their BTO efforts. Dynamic System QoS (see pie chart below) is all about aligning your distributed system resources to your business-critical applications across every desktop: The myriad of cost and performance benefits demand it. As stated previously, the problem is pervasive yet well cloaked, and yet so ironically simple to recognize when exposed.
Also see Technical Blog: Enterprise Desktop Anarchy
http://desktopblisstech.com/2011/04/11/tech1/
- Dynamic System-Level Quality-of-Service (QoS)
Ziften Corporate Blog:
Tags: APM, Application Performance, Desktop Optimization, Desktop Performance, Endpoint QoS, Endpoint Quality-of-Service, System Optimization, System Performance, System QoS, System Quality-of-Service, Win 7 Optimization, Win 7 Performance, Window QoS, Windows Optimization, Windows QoS, Windows Quality-of-Service, XP Optimization, XP Performance
