Every network is different, and produces its own unique set of challenges for an administrator. When choosing which solution to use, you need to look at all angles to determine whether it's right for you.
The most important part of a network monitoring solution is that it's clear, easy-to-understand, and can give you an overview of your entire network's health in one glance. You need to choose a solution that will give you such an overview of your network's health at startup, and show you your monitored systems and give you an idea of what systems and services require attention.
Multiplatform monitoring is a reality in nearly all of today's networks. You will often find a dizzying mixture of hardware and operating systems. Linux and Windows systems, NetApp storage, APC UPS's and Cisco equipment are just a few of the types of components that could make up your network, and an application that can monitor all of these systems from one place is invaluable.
A network monitoring solution that provides you with predefined monitoring rules (called monitoring packs or monitoring policy by some vendors) is indispensible. Whether one uses built-in rules that are provided or build their own, these types of predefined rules can let you rapidly change how different classes of equipment are monitored.
Every network monitoring product provides alerts and notifications, but it's important to look at how these can be customized. For example, there are resource owners who want to receive only messages from the devices and servers used in their daily business. With the aforementioned monitoring packs, a monitoring solution should also allow you to define the alarms you are looking for and then setup different notifications based on the alarm levels and control how these notifications are sent to users (via email, SMS, etc). For example, if you stand-up a new SQL instance, you want to be able to define it as a node and then assign an SQL monitoring pack so the resource owners can start getting their required notifications.
Customizing views for SNMP to fit your personal preferences and needs is vital for an efficient monitoring solution. You're going to want to be able to roll your own views for your SNMP data. This way you can have customized views for anything, your UPS's, NetApp storage appliances, Cisco ASA's, etc. You also want to have routing and switchport mapping, that can immediately show you which port is connected to which network segment or device.
A safe and effective backup and restore system is crucial. Nobody wants to face the occasion when they have to rebuild their database after a server crash. Having the ability to restore it from backups taken regularly will save you untold hours of work.
Monitoring dependencies is especially vital if your network is spread out, or you have sites in remote locations. By establishing the proper dependencies, you can reduce the number of alarms generated by one of these outages to just the site router.
Normally you need to monitor devices and the performance thresholds relevant to them. For servers, you want to be able to monitor their operating systems and services or applications running on them. If you can assign automatic corrective actions to alerts related to server or application performance issues (for example to restart a service or run a script) it can save you time and trips to a remote location. In this way, network self healing is slowly becoming a reality, and one day will be found in all monitoring solutions. Administrators should focus on making important network decisions, not on mundane repetetive tasks.
How should your daily operations look like with a network monitoring solution that fits your needs? For your daily task, you're going to want to open up your console and see your network summary. A quick scan of your network map should show you the health of your network, and any alarms that were generated overnight. You want to quickly be able to determine which require a simple acknowledgement, and which require hands-on attention. SMS alerts should let you know 24 hours a day if there is a critical failure in your network. and you should have a reports feature that can give you a daily and weekly overview of your network's health based on your needs, for example by device class.
The decision on which software to use can be daunting, there are many solutions to choose from. A modular approach can seem like a good idea, but you always run the risk of problems when patching together multiple applications. Your best bet is to go for an all-in-one solution like NetCrunch 7, which offers all of these features, and many more in a compact and efficient product, which deploys automatically and is easy-to-use and understand.