kentoh - Fotolia
The tradeoffs and cost of network redundancy
Network redundancy costs vary depending on enterprise use cases, but the determining tradeoff usually depends on how long a company can sustain network downtime.
If cost wasn't an issue, enterprise network teams would make every component of the network fully redundant. But, if cost is an issue, and it nearly always is, teams need to analyze and determine which elements should be redundant.
Each redundant component adds both the cost of the component and the incremental cost to maintain that additional component. The decision to add more redundancy depends on a tradeoff between the cost to a firm if the network fails versus the cost of an additional component. There is no one answer for all firms.
Network redundancy costs for e-commerce
Clearly, a firm that earns its revenue through online sales will have different needs from a company that doesn't sell online or one with nearly all its employees located in the same office. For an online retailer, all network components that carry customer traffic must be made redundant. In this case, the cost of redundant components is nearly irrelevant compared to the cost of lost sales due to a network failure.
Large e-commerce companies maintain entire redundant sales facilities located around the world. In some cases, the volume and location of sales may require these measures, but the redundancy also provides the ability to redirect traffic if a disaster, such as a large-scale power failure, puts one facility off the air.
Any online retailer needs at least two links to network providers. Networks fail for any number of reasons, so it's necessary to be able to quickly switch over to another provider. The number and capacity of links must be weighed against the cost and expected traffic level.
Protection against malware is essential, so the firewalls that receive and filter incoming traffic must be made redundant. Failure of a unit must not be permitted to block or slow the flow of incoming transactions. The web servers and devices located behind the firewall, such as web load balancers, must also be redundant, as a failure of these components would interfere with the flow of data.
A second set of firewalls protects the servers that process transactions and the storage units that record them. All these components must also be made redundant, as a failure here can also block or slow transactions.
Servers and storage can be located in-house but are increasingly located in a public cloud. Moving to a cloud environment provides the needed redundancy because clouds provide a large number of processors and storage units and are geographically distributed. A cloud-based storage offering may be less expensive than buying and maintaining equipment, and it also offers the possibility of continuous backup.
Network redundancy costs for offices
Other types of companies have less need for redundancy. Some enterprises may be able to deal with loss of the network or part of the network for minutes or hours. The costs that can be justified for this type of company are much lower than for those that depend on the network for their revenue.
Enterprises where all or nearly all the employees work in a single office offer an extreme case. These firms may maintain a website, although it would probably be hosted by one of the external vendors that advertise this service and it would likely also handle email.
The office network should have redundant elements. Redundant switches and routers prevent parts of the network from being cut off due to a single failure. The cost of redundant units in this case is small, as these devices are relatively inexpensive. Enterprises would still need to employ someone or hire a service to keep the devices up to date when software patches or updates are released. This type of company would likely find it advantageous to move processing and storage to the cloud, just as with e-commerce firms.
Reevaluate redundancy tradeoffs
Many enterprises have redundancy needs somewhere between these two extremes. When circumstances change, they must reevaluate the tradeoffs. For example, needs change when employees must leave their offices and begin working from home. An organization that didn't need multiple external links when employees were in the office may now need redundant links. Any other changes in how the network is used will cause a similar reevaluation.