Monitoring and managing network devices is becoming increasingly challenging. But it shouldn't require a computer science degree or weeks of training to use a network management system (NMS). Here are five tips to help you improve the quality of your monitoring efforts and reduce licensing costs using your NMS.
Tip 1: Only monitor devices you really care about.
Be sure to review the inventory list discovered by your NMS. Remove any devices that your team does not manage or that you wouldn't respond to if a failure occurred. You may be tempted to keep these "just in case," but they can clog your database, slow your system and make it harder to see more important issues. This can even save you money, since most vendors charge by the number of devices you monitor.
Tip 2: Reduce the number of SNMP community strings.
Networks that have a large number of SNMP community strings take more time in discovery, often much more time since discovery systems must test each community string and timeout before moving on. It's a common problem since strings frequently get added whenever a new Network Engineer joins the team. A better option is to create a uniform policy with no more than one SNMP community string for each type of device. For example, you could create a unique community string for routers & switches, security appliances and servers. If you need additional security, consider SNMP v3 or use the SNMP security feature on most devices to block SNMP requests from unknown destinations.
Tip 3: Use visual aids.
When appropriate, use a floor plan or geographical background image as part of your network map. Network layouts become more tangible and you'll recognize devices and where problems or bottlenecks are occurring more quickly by using a visual aid.
Tip 4: Implement a standard naming convention for your network.
Use a naming convention that is readable and can grow with your network. It may seem hard to go back and change all the devices now, but recognizing those elements will become much easier in the future. Try something like this:
Which in friendly terms, means: San Francisco California, Site 1, router, cisco4500 #1. Now you'll know right away where the problem is and on what device. In addition to being clearer, these types of conventions are scalable. Even though you may not support multiple sites or cities today, you'll be ready when your organization grows.
Tip 5: Set rules and alerts.
Red lights and green lights are fine for showing the hard up/down status of devices, but there are many states in between. Take some time to understand normal performance levels in your network then tailor the rules of your NMS accordingly. It will pay off if you can be alerted when conditions start to deteriorate and allow you to respond before customers experience a problem. For example, monitor the interface utilization of end users or the switch ports they are on. If a user switch port is at 70% utilization, you may want to be alerted of high bandwidth consumption. On the other hand, bursts above 70% may be normal for your users, so that type of rule would cause an excessive amount of false positives. Instead, some systems will let you set alerts only when the interface utilization stays above 70% for an extended period, say 30 minutes.
Take some time to consider these five tips and if you have any feedback let me know.