The State of Network Management

A Network Management System (NMS) is broadly defined by Techopedia® ™ as;

… a broad range of functions including activities, methods, procedures and the use of tools to administrate, operate, and reliably maintain computer network systems. Strictly speaking, network Management does not include terminal equipment (PCs, workstations, printers, etc.). Rather, it concerns the reliability, efficiency and capacity/capabilities of data transfer channels.

I am going to blur the lines between traditional network management (the data channels and switches mentioned above) and servers, storage, mobility …even databases and cloud. These are all elements of critical infrastructure. They all matter. All of these elements are connected. Through these devices flow all of the critical data and applications of the modern enterprise. Savvy NMS vendors recognize that they have the opportunity to provide unified tools to benefit their customers.

The value of an NMS is to give administrators a single pane of glass to observe the condition of their networks. Initially, the NMS was monolithic and lacked ease of use features and the ability to scale up as needs grew. With more recent offerings, the NMS vendor tends to take a more modular approach; building modules for a specific purposes (wireless management) under the main NMS interface. 

This discussion will remain agnostic in terms of vendors, although I will call out a few products when they embody features that clarify the discussion.

The NMS was created to fill a need. And it evolved as the challenge of managing large networks grew more difficult for administrators.  Let me take a step back to the 1980s and describe the necessity for, and evolution of, network management.

The production of software to manage complex networks of computing systems emerged during the period when each network device or server was a discrete entity. You had to log on to that system to make changes or to monitor processes. Any changes made during that configuration session only affected that single device. Clearly, this was not a very efficient process. Eventually, server administrators developed their own scripts and software so that monitoring and settings could be observed and changed at a system-wide level (for instance, encompassing all of the servers in the local network).

Some vendors were innovating, most notably Apollo’s Domain/OS ® ™.  Apollo brought out a line of systems with broad network capabilities and an ease of use and administration. The shared resources model and the management features of the OS for users and groups was innovative at the time.  The Apollos favored Token Ring, but could also run on Ethernet networks.  The Apollo line was brought under Hewlett Packard’s umbrella.

It was the Hewlett Packard Corporation which launched a notable product, Network Node Manager ® ™ (NNM). This product was network monitoring software based on the Simple Network Management Protocol (SNMP). Eventually, this set of tools became HP OpenView ® ™ which has since adapted into an integrated product line with many variant modules. In my view, this platform drove some of the important standardization which had to be built before large scale systems could be managed efficiently.  For some time, the large UNIX vendors were the leaders in the development of network management platforms. They had the talent pool and resources to create the software necessary to manage larger and larger networks. We also needed tools beyond telnet and the ubiquitous ICMP and the ping command.  

If you want to manage systems, you need compatible protocols between the systems of different vendors. Our networks, by now, had become heterogeneous. We had a diversity of systems under our control. Each vendor needed to produce an interface that would allow communications between the network management platforms and the devices under management. 

The result of this activity was standardization.  Listed in the table below, are the protocols that allowed monitoring, polling, performance measurement, data counters, and alerting as well as configuration and change management.     

Ideally, a modern NMS is a software system that can easily be expanded as the network grows, with the capability of discovering new nodes and networks. Modules provide new capabilities as business needs change. For instance, latency and performance monitoring may be a business requirement between two geographically separated data centers.  A module devoted to data storage, replication monitoring and trending could become an essential component under the umbrella of an NMS.       

Vendors are still providing tools that are built to operate within their product sphere.  A small sampling of these products gives us an idea of the diversity of tools. In the industry, they are classified by vendor, and taxonomy. For instance,

• Microsoft - System Center Configuration Manager ® ™
• Citrix -  Network Manager ® ™

NMS Systems which are targeted at the multi-vendor enterprise and feature modules include;

• SolarWinds  - Network Performance Manager ® ™
• InfoBlox  -  Network Insight ® ™

In the second installment of this post, we’ll delve deeper into NMS protocols and how they are used.