Arthur Cole spoke with Kenny Van Zant, chief product strategist, SolarWinds.
Cole: It seems that the more enterprises come to rely on distributed architectures, the greater the challenge to maintain centralized management. What are some of the ways enterprises can keep control of their network resources as they continue to expand over wider and wider areas?
Van Zant: It's true that a distributed network presents management challenges to any IT organization. Keeping close tabs on the health of the network becomes more difficult. As an organization grows, the IT group needs to decide how they will be able to most effectively monitor the availability and performance of the network. Factors such as geographic location, WAN bandwidth and latency will affect the management architecture. Generally speaking, highly distributed network architectures with latency or bandwidth constraints are usually best served by a correspondingly distributed network management architecture. In this case, it is usually best to deploy network monitoring remotely at key locations and roll up views to a central management console. That way, a centralized IT staff can maintain a bird's-eye view of the network while allowing local IT staff to manage their part of the network as desired. If, however, there are no concerns about WAN usage, or if remote administration of servers is not practical, then centralized management can scale effectively.
Cole: Should enterprises be concerned about multi-vendor networks? Does it add to the complexity of the management system if there is a wide variety of systems to contend with?
Van Zant: Yes-multi-vendor networks are a reality, and will be forever. However, that doesn't mean management complexity should increase substantially. Multi-vendor networks that require multiple management applications and significantly different skill sets are a problem. But it is quite possible to roll the entire network infrastructure under one management umbrella without requiring a dreaded "framework." Network management interfaces are, after all, highly standardized-SNMP, ICMP, flow-based protocols (such as NetFlow) and WMI are just a few of the standard ways network management tools manage a wide variety of vendors' technology. Therefore, network management isn't a trade-off between single-vendor vs. complicated multi-vendor environments. An effective network management product can and should adhere to standards, and be easy to use, configure, and maintain without requiring a team of consultants to install. Also, it can and should scale to the size of any organization no matter how small or how big, without a forklift upgrade or professional services to implement.
Cole: The Orion system uses a central console to communicate with multiple Orion servers throughout the network. How does that help administrators keep tabs on overall network health?
Van Zant: One of the key features of the Orion Enterprise Operations Console (Orion EOC) is to present a top 10 (or top X) list of everything that warrants attention across the global network infrastructure. This includes top 10 interfaces by bandwidth utilization, top 10 CPUs by percentage utilization, top 10 server volumes by disk utilization, etc. This type of rolled-up network health status is invaluable to IT administrators. Orion EOC presents a top-down global view of network health, up/down node status, network performance, application performance, etc. and presents this to the user in a single console, while also providing the ability to drill down into specific node, application or interface details.