Blueprint for a Perfectly Healthy IT Network

In the world of make-believe, it’s assumed that networks work perfectly for all users all the time. In reality, it’s far from that—sometimes borderline nightmare where things glitch and users are constantly complaining.

What’s the difference between these two worlds? Visibility and understanding. You need to be able to know every impairment in a network, the moment it happened, so you can solve problems before users even noticed the problem.

Here are the top 4 things you need to have a perfectly healthy IT network:

1. Visibility into the entire network
2. Deep knowledge of your network’s operation
3. Automated network diagram
4. Root-cause troubleshooting into every link, switch, and router

You need to know what’s happening on the entire infrastructure, not just a few interfaces here and there.  If you only monitor a few trunk ports on devices, you don’t really know what’s happening.  If that trunk port changes and you don’t update your monitoring software, you’re blind until you realize what happened.

Make sure you have total network visibility into every device on your network, and every interface on every device. The tool you use should automatically configure your network within minutes. And allow for rapid reconfiguration when there are changes in your network. You don’t want to spend days manually configuring your network with chances of lost visibility due to misconfigurations or forgotten settings. You’re responsible for the entire infrastructure—you should have visibility to match.

You need to have deep knowledge of your network’s operation including error counters, packet loss, routing performance.  Many NMS systems don’t collect enough information to make it easy to troubleshoot and you end up with users that complain about network issues, yet your NMS still says “Everything’s fine”.

With total network visibility into your network, you’ll be able to automatically analyze for performance, errors, QoS, and configuration, with the ability to collect and analyze 19 error counters on every interface. That way you’ll know when, where, and why packets are lost, buffered, or mishandled anywhere in the environment.

Make sure to set up a daily automated health report so you’ll know what issues exist and what problems are developing so they can be worked on proactively each morning. And create customized reports that can be sent to the appropriate people to keep them aware of their environment: MOS Reports, Interface usage reports, Transmitters, Errors, and many others. With more awareness of your environment, you won't get caught off-guard by users having problems – you'll be finishing the fix before your users even notice a problem.

Use predictive analytics to continuously analyze your environment to produce forward-looking prediction reports about your network so you can remedy situations before they become problems. This should include predictions on cabling and bandwidth. With this intelligence, you'd work to prevent the problem from ever happening.

You need a good dynamic network diagram. Knowing how things are connected and where users connect is invaluable to understanding your environment.

Network diagrams should be automatically generated so you don't have to worry about updating it, and be flexible and interactive so you can move elements around and lock them in place so you can have a static or dynamic view of the environment. With a single picture, you should be able to see how your network is connected. The days of manually updating network diagrams are over.

You need to know how devices connect to services.  Easily knowing every link, switch, and router used to connect any two devices across the network means it’s easy to root-cause troubleshoot problems that occur because you know what happened along the path.

Your root-cause analysis (RCA) on network problems should be in plain English (NLT: Natural Language Troubleshooting) with the ability to enter a simple English question and have the system answer it in English. For example: Entering "What happened between 10.0.2.12 and 172.16.4.36 20 minutes ago?" should do an analysis of all involved links, switches, and routers used to connect those two IP addresses to tell your team a simple answer that the Finance2 switch interface #2 was dropping 12% of its packets at that time due to a cabling fault. You shouldn’t have to look up cryptic answers to find out what’s going on. Using plain-English answers allows for rapid remediation of network issues that lower level engineers can handle.

Use real-time intelligent alerts to assign parent-child relationships and to assign global alerts that cover the entire organization, or only a group, or device, or specific interface. Configuring alerts should be very quick and easy, and the alerting should automatically cover any newly added devices in the infrastructure or group. Getting tons of alerts is never fun, and doesn't lead to quick recognition of the problem.

The ideal network management tool will automatically discover devices, computers and apps, will automatically create a graphic depiction of the network components, and proactively monitor the health of every device or computer. It should be configurable to send alerts to different individuals or groups depending on the problem, can escalate notifications when problems persist, can perform root cause analysis in plain-English. Plus, it should produce useful, easy-to-understand and timely reports, is highly scalable, reliable and is easy to use.