The cycle of life within nature occurs before us every day: conception, gestation, birth, maturation, decline, and eventually death. As humans, we've taken this inherently natural process and abstracted it to include many of the things we create. For instance, buildings have an effective use period, and then we replace them with new ones, conceived with the view of better services, access, technology, or capacity.
Software developers are deeply committed to a life cycle, and more than a few trees have been sacrificed to describe various methodologies. Networks, too, can be described as having a useful life cycle. To further understand them, this article will examine a network's migration from one life to the next.
Most data networks will follow the cycle as described above. The cycle of a network's life is similar: idea or concept, design and implementation, operation, legacy integration, and eventually transformation.
There are four basic steps to the transformation process. They are:
- Assessment (conception)
- Design and definition (gestation)
- Implementation (birth)
- Maintenance and support (maturation, decline)
Conception occurs when an organization finally realizes that the existing situation is no longer viable. After obtaining the necessary approvals and funding, it is time to begin the transformation process. A new life germinates. But what color will you paint the baby's room? Do you even have a room?
The process of transformation requires as much understanding of what the network is, and how and why it got there, as it does of where it's going. This task is not as simple as it would seem, for the counting and classification of switches and routers is merely the first step. A comprehensive survey and documentation of all active and passive components is required to determine if anything is re-usable, if right of ways or building access points currently in place will meet your needs, and to determine if environmental controls, fire suppression, and backup power are sufficient to support any new equipment. This is by no means a comprehensive list, but is a good indicator of what is generally required. While it may appear that more emphasis is being put on assessment than on the other areas, that is because it is the most critical element; the other three areas will not function well over time if this step is hurried or overlooked.
Gather all existing documentation and compare it to the infrastructure as you perform your assessment. Update it as you go. The Assessment Worksheet, which you can download here, though not exhaustive, will help you isolate issues and prevent problems in the future.
The seed is planted and you're "pregnant" with the idea of a new environment to operate in. But before birth occurs, the embryonic network must develop its personality and life's direction. That requires architecture and design.
Network design is sometimes looked upon as a 'tinker toy' exercise. The impression is that as long as the dots are connected in some manner, a network appears. This type of design usually results in the oft heard complaint "the network is down!"
Good network design takes into account all of the servers, applications, storage modules, projected usage patterns, and a well educated estimate as to future expansion of the network (users and sites). Robust designs allow the dynamic addition of users and sites, along with the associated traffic, with little or no impact to the original user set.
Begin with an architecture that defines the overall functions of the network but doesn't specify equipment. This is similar to a building architect defining the use of a structure that must withstand the elements. Once accomplished, you can move to a design phase where technologies are chosen, vendors are selected, and details of location, power, fire support, security, access, and other areas are finalized. Include the network management function, spares and maintenance plans, and network elements for help desk as part of the architecture and design functions.
Both the architecture and designs should have written documentation explaining details along with drawings at various levels of complexity. It is critical that the high-level perspective of this documentation be understandable by non-technical personnel, as these will be presented for review to executive teams for final approval of funding.
Like the stages of childhood, a network goes through different stages before reaching adulthood. Segments may come online at various points in time as you roll out, but, as with any building project, implementation begins from the ground up. In the network world this denotes the elements at the bottom of the open system interconnection (OSI) stack or the cabling. Cable plant is often the most difficult piece to implement, especially if you're required to dig trenches to lay cable between buildings or across town. These are also the most expensive elements and the longest to deploy as they usually require environmental impact studies or special building permits and depend on weather conditions in some areas of the country. Short-haul wireless solutions can sometimes overcome the need to trench cable.
Building new networks from scratch is often much easier than transforming existing infrastructures, because there is no user community on the system. In the case of transforming existing systems, construction must occur in parallel with the functioning network. After physical buildout is complete, the logical elements of routing, security, numbering and naming are installed and tested. Cutover testing during outage periods must be arranged with each site's user community to minimize impact to the business. This can be especially tricky when the site also holds applications or storage for mission-critical systems. If possible these sites should be tested either first or last in sequence, or they may be transitioned to new facilities to minimize impact to the enterprise.
This is the adulthood of a network -- the prime of its life. The network functions day to day, as does any adult human. And, as any adult human, it occasionally needs medical attention. Circuits, routers, switches, cables, connectors and the like may suffer failure or degradation at some point. Routers need logical tune ups, users and applications come and go, or new server equipment and storage is added. And like an annual physical, all of this equipment and its associated configurations needs to be reviewed on a regular schedule to ensure optimum performance. As the network grows, changes, and ages, it is critical that the documentation of all aspects of the network be maintained and available, just as a medical record for a human must be current and accurate to ensure proper treatments are applied.
The transition from an aging, suboptimal networking environment to a modern high-performance network is often a painful process. We are frequently required to relinquish processes, personnel, and equipment that has in the past seen us through difficult times. As in the cycle of life, this passing of the familiar and safe, this death of the old system, can be unsettling. And, as in the cycle of life, the birth of a new environment can be welcomed with much celebration. Remember that new systems, like babies, will occasionally have an accident or get a "boo-boo," and they need a fair amount of attention until they reach their prime.
Following a life cycle model to develop new networks or facilitate change to an existing network will help ensure that a network has the smoothest and longest life possible. As with any life decision, to ensure the success of the networking project, spend the initial effort in assessment and design before jumping into implementation. The process will not be without its challenges, but the results should speak for themselves.
**To simplify your network assessment, be sure to download the accompanying worksheet here.
About the author:
David Lease is chief architect of Netco Government Services, Inc. (Netco). Netco is a leader in network design, management and engineering for government agencies. Contact David at firstname.lastname@example.org.