Getting ready to deploy a site-wide network management system? There are a lot of things to consider, so it's always best to create a plan to implementation and maintenance. Here are the cliff-notes for installation and continued service.
The first thing you'll want to do is take stock of the devices on your network. What do we actually have running out there? What devices are on our network that we'd like to keep tabs on? Create an inventory of all your devices and be a thorough as possible. A solid network inventory with detailed descriptions of the devices' hardware, software, and network configuration can be invaluable as you progress.
You may need to purchase a license for your management software that will limit the number of devices you can work with - it's better to have a good idea of how many devices you're going to monitor before you settle on an application so as to avoid costly upgrades for increasing device limits.
Once you've created a good network inventory, you should now start thinking about what you want to actually achieve with your management software. Review your devices and start prioritizing which ones are most important to your business. Which devices are most likely to cause critical failures across your network? Which devices have a history of problematic behavior that we want to watch out for? Which devices require the most amount of attention to keep functioning properly? Consider these types of things when determining your monitoring needs.
Now that you've spent some time thinking about device prioritization, start making a wishlist for what parameters and events you want to you monitor for on your most critical devices. Keep in mind that some types of monitors may not be available depending on the software you choose, and some monitors may create more network traffic than they're worth. These factors largely depend on the software you choose to monitor your network and which protocols/methods are required to accomplish the monitoring you need. Consult the software developer directly and let them know what you plan to do - they'll let you know if the software supports the monitoring you need and what type of network overhead to expect from the program.
Armed with a full network inventory and wishlist for management needs, you're ready to start shopping for a software package. There are many different types of monitoring and management solutions out there, some more expensive than others. You may chose to purchase a full-featured solution that you install on-site, rent a cloud-solution for a monthly or yearly fee, or even build your own solution from scratch using open-source solutions researched from the internet. The decision will largely depend on how much time you have to invest and how big of a budget you've secured for the project.
See our full article on choosing a monitoring solution, here.
Now that we've selected a management application, we'll need to make sure our chosen server is ready to act as the central destination for all of the management and monitoring tasks we intend to carry out. This usually means setting a static IP for the server and choosing a meaningful host name to point agents and other alerts at.
Make sure your server has the horsepower to back up the management software. Does it meet or exceed the minimum required specifications for the application? Do we have enough disk space to accommodate the level of history we plan to collect? We have a reliable network connection to our target devices? Are there any firewalls in the way that might hinder our commands from reaching the devices? These are all questions we'll need to consider, and we need to be prepared to address any problems we may have with the installation.
Most likely, you'll need to install some kind of agent software on your target devices in order to monitor and manage them. If you're using Simple Network Management Protocol (SNMP) as your primary method of monitoring and configuring devices, you'll need to make sure the SNMP agent is installed on the target device, and that it's configured with secure authentication (community strings for v1 and v2, user name/password for v3). We'll also need to make sure the devices are sending traps to our monitoring server and only accepting commands from our server's IP/host name.
If you're going to manage/monitor Windows machines and servers, we'll need to make sure they're either set up to use SNMP (by installing the Windows feature for SNMP and configuring the SNMP service) or by enabling Windows Managment Instrumentation. WMI is a common protocol for monitoring and managing Windows computers and has many useful features, but keep in mind that the network traffic created by WMI monitors can be significantly more than SNMP monitors.
It can be tempting to set up as many monitoring and management tasks as possible, but realistically we only have so much time, so much bandwidth, and so much disk space to work with, so we'll often need to choose which tasks are most important to us. Before you waste time setting up low priority tasks that will be deactivated later to make room higher priority ones, it's best to create a simple priority list for your devices. Figure out which devices are most important to you and your organization.
Maybe certain devices are critical to production for the company and need to maintain certain levels of up-time reliability. Perhaps you have core backbone network devices that would cause mass connectivity loss if they were to fail. Often times offices have troublesome devices that end up consuming a lot of hours in maintenance and service. Consider which devices you're responsible for, and which ones would be most helpful to monitor or manage, then work up from there.
In order to gather metrics or make configuration changes, you'll need to be able to authenticate into a device. Using your device inventory and the list of management/monitoring tasks you've devised, determine what types of credentials you'll need to gather. SNMP commands will require SNMP credentials in the form of community strings for v1 and v2 devices, and user name/passwords for SNMP v3 devices.
If you're going to be doing any Windows monitoring, you should make sure you have a valid domain administrator account to use in order to access the target devices. It's recommended to create a central domain admin account on all target Windows devices to make password changes easy and simplify your credentials library.
At this point, you should be all set to deploy your management or monitoring solution. Each application installation is different, but they generally involve installing software and tying that software to a database, either on the server itself or via a remote database connection. You may need to consider things like IIS for web console support, SQL for database management, RAID configurations for disk failover and mirroring, as well as things like port usage and access control for the software.
It is always recommended to host your management or monitoring software on a stand-alone server, and not try to install it on a server that already has any major software installed, particular similar packages such as other monitoring or management software. Reason being: applications like these often use specific ports in order to communicate with other devices on the network, and will most likely create conflicts within the system that can lead to constant or intermittent issues
Now that the software is installed and ready to run, we can go ahead and start creating our monitors or management tasks within the application. Start with the highest priority tasks and go down the list. Always test your monitors if possible to ensure that the results you're getting are accurate and reliable. There's nothing worse than assuming a monitor is working only to find out later that it was never reporting status properly and was basically ineffectual from the beginning.
As you create monitors and tasks, we recommend doing so slowly and methodically. Don't jump around from task to task, as this can quickly lead to forgotten work. Check tasks off of your priority list as you proceed, and keep an eye on bandwidth usage, memory usage, disk usage, and the volume of historical data. You may find that some monitors create too much activity and need to be scaled back.
With monitors and tasks in place, it's time to set up notifications for when one of those monitors indicates a failure. Always try to minimize the volume of your alerts - less alerts is actually better! We've learned from experience that constant inundation from alerts is a quick way for them to become more annoying than useful. Make sure that when you receive an alert, it's something you really needed to know If you start receiving alerts that you shrug at, it's time to immediately access the validity of that alert and adjust the sensitivity if possible.
You can get alerts many different ways - you'll need to check with your monitoring or management application to determine what types of alerts are available to you. Almost all systems support email. Many support SMS messages, and if not, you can always set up your cellular phones to accept emails in the form of text messages. Otherwise, you'll often find things like dashboard pop-ups, sound notifications, Windows log files, and color-coded visual cues that show up on device maps and lists.
Getting alerts is great, but to really understand trends and historical data, we'll need to generate reports Most monitoring and management packages offer some type of reporting - the quantity and quality largely depends on how big of a check your company wrote for the licensing, but not always. If reports are important to you, inquire with the developers beforehand about the scope and depth of reports available in the application.
Reports can usually be viewed from within the application, or perhaps within a separate reporting dashboard. Some applications offer convenient ways view reports by having them emailed to you on a schedule or interval. Consider making reports available on a need-to-know basis. Without the full picture of the network, reports can often be misunderstood or over-emphesized.
In addition to notifications and reporting, some applications allow you to schedule recurring actions, or to trigger actions if a certain event or monitor is activated. You may be able to automate some of your troubleshooting and maintenance tasks via scripts or batch files being kicked off.
For example, you may need to restart services on a regular basis or delete files that build up too much and bog down a system. You can kick off configuration changes via scripts as well, switching the mode on a component or activating a feature that was off. With scripting at your disposal, the sky is the limit.
Depending on the scale of your installation, database maintenance may be trivial, or it may be a major weekly task. The type of database you'll be working with will depend on your selection of application, and the size of the database will directly correlate to how much polling and data collection you're doing.
Consider setting up database fail-over solutions and definitely back up your database as frequently as possible. The database will contain not only historical data, but also probably house the platform configuration settings for the management or monitoring software. Some applications include tools to perform database clean-up and back-up tasks.
Last but not least, you'll need to stay on top of new versions of your application, because these often contain crucial bug-fixes and security updates to keep your server from being vulnerable. Some developers offer these updates for free, while others provide access to updates for a fee - this depends completely on the developer and varies from program to program. You'll most certainly find that open-source applications are updated far less frequently, albeit at a lower price (read: free). Enterprise applications may require service agreements and updates may be performed as part of this service.
Should you wish to increase your device count limits or gain additional functionality, you may be able to purchase upgrades for the application from the vendor. Many applications have niche plugins that will allow you to monitor or manage specific device types that have unique parameters and uses, such as NetFlow monitoring, VM monitoring, or VOIP equipment management.
As you can probably see, deploying a network management or monitoring solution can be quite a project. We at Network Pipeline have performed hundreds of implementations and deployments over the past years and have leveraged that experience to design and build a simple, all-in-one network monitoring solution we called the Complete IT Box.
Learn more about the Complete IT Box, here.