Understanding the network application environment is one of the foundations of application performance management, as I discussed in the tip entitled Application performance management: Developing a strategy. In this tip, I go into detail about how you can better understand the application environment for your network.
Three areas are essential to understanding the application environment:
- The network environment: This includes the network geography in terms of end users and application locations (data centers/servers) and the design of the network to support application delivery.
- The application mix: This includes understanding both the number and types of applications as well as the resources consumed.
- The business environment: This includes current and future initiatives that will have an impact on the number of applications delivered.
All of these are primarily discovery tasks. The objective is to gain a complete, enterprise-wide understanding of the application environment, then assess the technologies and tools required to control and monitor the application services.
It can be very hard to determine which applications are running on the network and even harder to maintain their performance proactively. The first order of business is to discover the applications. There are several techniques that can address this, including interviews with the network and server staff, as well as the business units, to ask what applications they are aware of. This is generally step 1 and can be a good way to determine important applications, as well as problematic applications. Try to develop a classification of applications prior to beginning the interviews so that you can ask questions relevant to the categorization. A general rule of thumb is to categorize as follows:
- Real-time, delay-sensitive applications: These require strict controls on the network to ensure performance. SAP, Citrix, VoIP and signaling protocols are common examples.
- Mission-critical applications: These can be both office applications and applications that drive business production (such as credit card transactions over the network).
- Best-effort applications: Commonly used but not necessarily critical to the operation of the business. Internet traffic is a good example of this.
Once you have categorized the applications, you can move on to discovery and allocation to a category.
Interviews take you only so far during application discovery. In order to fully understand the application environment, you have to gather actual application information from the network. Many products provide application discovery, and choosing among them can be daunting. NetScout, Visual, CompuWare, Network General, NetQos, Packeteer and a host of other vendors can provide products that sit on the wire and capture application statistics, as well as report on the performance of those applications. Some form of tool is critical, as you must collect application information, then monitor on an ongoing basis, to provide true application performance management.
In addition to application discovery and classification, the network plays a huge role in application performance management. Traffic is traversing different parts of the network in different quantities. Depending upon server locations and network geography, the traffic patterns can be significantly different for different areas of the network. This requires a wide view of the network in terms of capturing application flows.
Today, technologies for server virtualization and consolidation, along with application acceleration techniques, make data center centralization a key to application performance monitoring strategies. Centralizing applications can also provide a centralized capture point for application traffic flows. Understanding and controlling that traffic is critical, and centralizing computing facilities for servers can provide greater visibility without the need to decentralize application traffic capturing capabilities. Having probes or application data collection engines at data centers and key sites is much more cost-effective than distributing appliances to all sites.
Intelligence in the network via QoS allows the real-time and mission-critical applications to get bandwidth when they need it and pushes the less critical traffic down the priority stack for delivery over the network. Understanding the network capabilities and designing the network appropriately should be priorities. A discovery and assessment of QoS capabilities embedded in the network technology should be addressed immediately. QoS is well understood and is supported in all major vendor products, and it is a critical component for gaining control of the applications and application delivery.
Finally, the business drivers must be understood. This will provide the justification for enhancing the tools and the network, but the business units must also understand that new application deployment cannot be done just at the business level. There must be open lines of communication between IT and the business units.
These are the key tasks that can be initiated right now in any environment. The choice of tools and network design lies in the hands of IT, and the interface process between the business units is the responsibility of both parties. A good plan is to start with discovery of the applications, looking at the appropriate tools and looking at the network.
About the author:
Robbie Harrell (CCIE#3873) is the National Practice Lead for Advanced Infrastructure Solutions for SBC Communications. He has more than 10 years of experience providing strategic, business and technical consulting services. Robbie lives in Atlanta and is a graduate of Clemson University. His background includes positions as a principal architect at International Network Services, Lucent, Frontway and Callisma.
