Active Continuous Optimization (ACO) for Server Infrastructure Performance Improvement
When is the last time that you used a timing-light to tune your engine or used a screw driver to adjust the fuel and oxygen ratios? Unless you are a classic car aficionado it has likely been awhile, and for good reason. All new cars come equipped with an Engine Control Computer that tunes the car actively and continuously. This has lead to greater reliability (remember pinging and knocking), better gas mileage, and improved performance. In fact many economy cars today have more horsepower and better gas mileage than many of the most expensive sports cars built before engine control computers. Antilock brakes, regenerative braking, autofocus camera, video recorders, plane autopilot systems, etc. all rely on active continuous optimization using surprisingly similar methodologies.
Ironically one system that hasn’t yet been taken over by this revolution is the computer, until now. When your computer (physical/virtual, server/desktop) or application isn’t performing well a team is often assembled to address the problem. Each arrive with their own “timing-lights” and “screw drivers” and the system is tinkered with until the problem goes away or worse, the system is completely replaced. While the datacenter has enjoyed major improvements such as self-healing networks and live automatic migration, they still lag behind almost every other device that has benefited from active continuous optimization. For one they require significant configuration based on the goals of the environment. They also largely focus on a single resource constraint and find an available device with that resource.
Every component that is assembled to create a server, from device drivers to applications, are built to be optimized for their specific environment and operating criteria. However, when combined into a single system, the level of interdependency and number of variables makes the task of doing so too difficult for a mere mortal. Tuning the heart of the computer: the operating system, has become a lost art and virtually all computers are installed with the standard default settings. These settings, delivered by the manufacturer as part of the installation process, or created by large IT organizations as “gold” configurations are designed to ensure the servers perform in a wide variety of instances, but not necessary to perform well. When cars became too complex to tune by hand, a computer was developed to do so automatically. Veloxum has created that computer, but to tune servers, not cars.
When Veloxum setup to address the server “self-healing” issue they looked to the active continuous optimization models used by other solutions and realized that the same concept could be applied to server tuning which would enable all of the many settings within the operating system to remain tuned for the current work load. The result was far better than we even imagined it would be. The state of the art prior to ACO was for a person to spend several days adjusting the many settings and relying on past experience to come up with a “tuned” server. One problem (just like for the automobile without ACO) the system needs to be tuned up again once the work load or environment changes. Possibly an even a bigger hurdle is that with so many possible permutations to the many settings a computer is much better than a person at determining the best settings. The result is that even a recently “tuned” computer by an expert will not be quite as efficient as one tuned by the computer itself using ACO.
Application Delivery Challenges
According to a recent survey by CIO Magazine, virtualizing servers, centralizing data centers and adding high-bandwidth collaboration tools are IT’s top priorities during this economic downturn1. While these projects may make sense financially and tactically, they also have a negative effect on application operation. Consider these results from a recent survey2:
- 77% of corporations reported that virtualization had had a significant, negative impact on application performance.
- 69% of companies said data center consolidation was causing application degradation.
- 43% confirmed that high-bandwidth applications, such as video and collaboration, were making it difficult to maintain performance.
Operating System Settings: Lurking Inefficiency
A covert but significant performance obstacle exists within Enterprise server infrastructure: the native inefficiency and static nature of operating system settings.
For example, Microsoft Windows 2003 Server, Exchange 2003 and Windows XP – a common application set commonly deployed together – have over 350,000 permutations of configuration settings, not including the huge number of settings for browsers, or OS settings for CPU, I/O and RAM on servers, along with their associated desktop hardware.
Some of these settings may interact poorly with each other or even conflict. The majority are not well documented, especially in the significant way that they can cause synergistic effects with other settings. Although IT personal can manually tune all hardware and software systems, in larger data centers, it is impossible to do so in a realistic timeframe. Virtualized systems magnify this issue as local changes become irrelevant with shifting workloads. As a result, IT personnel encounter performance degradation and reduced capacity within their server infrastructure across the enterprise. According to a study commissioned by the U.S. Department of Energy, headed by Dr. Jonathan Koomey, the average server delivers less than 20% of its performance capacity when deployed in a commercial enterprise.
The Solution: Active Continuous Optimization
Enterprise server infrastructure optimization tools that utilize active continuous optimization (ACO) attack overt and covert causes of degradation. They unlock the potential of the existing server infrastructure thus increasing the efficiency of what you already own, improving application performance across the enterprise without costly hardware upgrades or increases in bandwidth.
Veloxum created its own comprehensive Enterprise server optimization solution specifically to tune server performance and to maximize server capacity. It targets the heart of the server, the operating system, working to maximize the efficient use of all server and client components that depend upon OS interaction including:
- Applications and Databases
Veloxum’s ACO approach
Veloxum’s features make it an exceptional tool by:
- Identifying and fixing performance problems via active remediation through a patent-pending process
- Rapidly optimizing both server and client, often within minutes
- Adjusting existing hardware and OS settings “out-of-band” while staying within the manufacturer’s supported ranges
- Periodically optimizing with changes deployed either automatically or through existing change-control procedures
- Supporting Linux, AIX, SUN and Windows OS with supplied OS modules
- Optimizing popular apps such as Exchange, SharePoint, SQLserver, Lotus Notes and Oracle with optional Application Optimization Modules (AOM)
- And deploying quickly and easily