In the ongoing debate between supporters of pure-Internet Protocol (PIP) telephony systems and supporters of hybrid-IP/TDM (HIP) systems, each group claims superior reliability or availability. Both are correct, but each of them tells only part of the story. Is there an overall story?
of all, let’s clarify the terminology. The word “”reliability”” captures only the downtime due to faults. On the other hand, the word “”availability”” captures the broadest measure of down-time, due to faults, maintenance or upgrades. A highly resilient system is one that can be shown to have high availability. “”Redundancy”” means the provision of spare capacity, the effect of which is to improve availability. Let’s now return to the debate.
High availability is not cheap
HIP system supporters base their claims on the availability of traditional private branch exchanges (PBXs). This is challenged by PIP system supporters, who claim that traditional PBXs were single points of failure. The counterclaim is that the availability of PBXs is so high to begin with, that the challenge is a red herring. PIP system supporters base their claims to superiority on inherent redundancy in IP networks. This, in turn, is based on the assumptions that there are multiple paths over which traffic may be routed, and redundant servers, switches and routers. Note that, being a hybrid, the HIP network has some of the same attributes as the PIP network.
Now, all of this is fine in an abstract way but, like other criteria in selecting IP telephony systems, useful answers depend on the user. It is important to realize that either type of system could be configured to have higher availability than the other — at a cost. So, what should IT or telecom managers do? The key objective is to identify a range of affordable options that meet requirements reasonably well. Before scoping the options, some analysis is required.
The first task is to find out from vendors the various elements (and related costs) that contribute to availability. Vendors should explain their strengths — such as operating system hardness, redundancy, hot swappability, resource pooling, failover/failback, distributed intelligence and network recovery time. This will disclose the probability and impact of outages at different levels: power supply, network, system, shelf and card. Some impacts will be more tolerable than others. It might be assumed that more of everything is better, but the trade-offs are cost and complexity. It does not take long to reach a point at which the cost outweighs the benefits.
The second task is to review requirements and assess the tradeoffs. Consider operating with reduced capability, as a compromise under certain conditions. Identify functions that really need high availability and those that do not (for example customer calls versus internal calls). Additional expenditures could be limited to supporting critical functions. It is also useful to consider reduced capacity under certain conditions. For example, two network elements could operate normally in a load-sharing mode. If one element fails, the other could provide service, but with some reduction in quality.
Regardless of the level of availability required, it should be determined in advance of system selection. Much has been written about the need to assess the impact on LANs and WANs when selecting new IP telephony systems. Most of the emphasis has been on quality of service (QOS), which is an important issue. However, availability is just as important.