Enterprise IT has embraced virtualized infrastructure, from servers to storage to networks. This romance comes with the promise of improved performance, increased resource utilization and reduced capital expenditures.
As with any technology, shortly after the first rush to implement them in a production environment, reality hits with the realization that the management of virtualized resources is more complex and even more necessary in this new operating environment. Some specialized tools are needed to supplement traditional management solutions to achieve the promised benefits.
Disconnecting and managing the allocation of physical resources from their physical reality represents only the very first hurdle to the promised benefit. Every virtualized service with its attendant application, server, network and storage resources carries with it the same management challenges as a physical machine. The problem is that virtualized resources require specialized management tools to assure their most effective and efficient allocation. All the various management functions -- problem, resource allocation/de-allocation, security, performance, etc. require some measure of adjustment to handle logical resources.
Performance management and maintenance operations come with more risk and an increased potential for serious damage if done wrong in the virtual data center. Imagine populating a change or making an adjustment that results in dramatically slow application response times or crashes all your virtual machines at the same time!
You simply cannot afford to operate without some help in the form of a tool providing automated, comprehensive and highly informative performance management and capacity planning functions. That's the bad news. The good news is that vendors have tools that address these problems.
Virtual server capacity planning and performance management tools
IBM divides the problem into three parts – monitoring, trending and capacity modeling. They provide logical and physical infrastructure monitoring using IBM Tivoli Monitoring. Tivoli Performance Analyzer can be used for trending to help with performance analysis; IBM Tivoli Composite Application Manager (and the OMEGAMON family for z/OS and subsystems) assesses service level impact and performance. Capacity modeling is addressed with a range of modeling and sizing applications tied to the platform and what exactly is being modeled. They offer a consolidated modeling tool for logical and virtual server modeling.
BMC's Performance Assurance for Servers and Virtual Servers delivers complete performance analysis and capacity planning functionality for both x86 and Unix virtualization environments supporting VMware ESX, MS Virtual Servers, IBM (LPARS, DLPARS, SPLPARS), SUN (Dynamic System Domains and Containers), and HP-UX (n/Pars and v/Pars). Of course, BPA integrates with all of BMC's BSM management tools, such as BMC Event Manager for event filtering and automatic alerting, and BMC Service Impact Manager for service management, in addition to offering full access to configuration data details stored in Atrium CMDB.
BMC Performance Assurance manages physical and logical resources from a single user interface and includes the ability to perform complex 'what-if' analysis using predictive queuing theory analytics to test the impact of infrastructure changes and workload fluctuations on application/service performance prior to their implementation. This helps operations staff avoid service disruptions and performance-related problems. Staff can test, properly size, consolidate and understand the current and future capacity and performance of business applications.
Big vendor solutions are complemented by products available from specialty vendors such as OPNET and Akorri. These products can integrate with and extend functionality described above. Let's focus attention on specialty vendor Akorri.
In early 2007, Akorri introduced BalancePoint, a product that provides visibility into the impact on performance of an application as it interacts and operates across multiple infrastructure domains. BalancePoint automatically discovers and builds a dynamic model of an application's performance and its interactions across server, network and storage infrastructures. The resultant mapping and active monitoring of these interactions and their impact on application performance provides operations staff with an end-to-end view of the level of service delivery.
Akorri separates the monitoring, analyzing, and reporting of physical from virtual (logical) resources. This allows operations staff to run separate 'what-if' scenarios to view the impact of changes to these resources. Therefore, operations staff not only have the view they need for real-time, fine-tuned control of operations, but they can also analyze where and when contention bottlenecks may occur, identify potential performance problems and evaluate alternative ways to resolve them.
We must say that these are not the only players with interesting solutions in the market. Also, emerging solution providers continue to introduce innovative products. There is little doubt that virtualized infrastructure is the wave of the future for the data center; therefore we will be writing more on this topic.
Seen or heard of an interesting product? Let me know what it is and why it struck your interest.
ABOUT THE AUTHOR: Richard Ptak is an analyst with Ptak, Noel & Associates. He has over 30 years experience in systems product management.