Don’t compromise on monitoring, visualization and reporting.
As a network professional, you’ve got a lot to keep up with. All the everyday challenges of capacity planning and maintaining uptime, increasing demands, virtualization, cloud computing, the Internet of Things… the list goes on.
You’re probably working with all kinds of devices installed at different times, with devices and connectors ranging from the “Where the heck did this come from?” to the relatively new. The chances that are your switches, routers, other devices and appliances will all be of different make, model and age. It all makes life interesting.
If you’re going to effectively manage your network in all its complexity, you want to have clear visibility, end to end. Even if you’re not spending a huge amount on a new solution, you shouldn’t have to compromise on the monitoring, visualization and reporting that your solution gives you.
Here are three key pointers for obtaining maximum performance and value from your budget.
All of your network, all of the time.
First up, you need to be using a scalable solution that can monitor your entire network (including software defined networks, SDN) in near real-time. It should be quick to configure, able to discover all network interfaces, and – for most networks – run using minimal hardware. Our model is based on 1 million interfaces on a single hardware server; or 500,000 for each virtual machine (VM) server (all positive news for OPEX and CAPEX).
Make sure your choice can monitor and troubleshoot all port behaviors (per port, per device, per site and per region) right from startup. As well as delivering rapid discovery out of the box, your monitor will need to self-maintain, with automatic rewalks of the network.
Being able to monitor every port over the entire network means that you can collect and store real-time data from all NetFlow, sFlow, syslog and SNMP trap-compatible devices. Look for a solution that uses minimal bandwidth so that there’s no noticeable impact on your network’s speed and cost.
Meaningful reports that don’t take forever.
With some systems you’ve got time to go and make coffee while the reports run appear. Look for something that can do the job in seconds – you want information on demand, not having to hang around waiting for it to finally appear.
In terms of output, you’re looking for reporting on data usage, latency, errors, discards, CPU, memory, disk and temperature values. If you can have user-configurable dashboards and viewers that still allow you to drill down to the raw metrics and data, so much the better – a single pane of glass means information is easier to share with others, and saves you fishing around for what you need.
Fast, full and frequent polling data.
Most of all, to cope with innovations and trends (increasing numbers of IoT connections, say, or the joy that is BYOD – bring your own device), on top of the day to day demands on your network, a state of the art network monitoring system has to be fast.
You don’t want to be waiting five minutes each time to see what’s going on out there, so don’t settle for anything less than 60-second polling. This means that you get data on every single port in near real-time, with reports generated on the fly, and alerts that are instantaneous and actionable.
And if you’ve taken the trouble to obtain minute by minute data, there’s little point in working with averages further down the line – make sure that your history is stored in full granularity for detailed comparison and accurate forecasting and trendline analysis at any time in the future.