I can never remember who first said it, but I've never forgotten the pronouncement, "We must all hang together, or we will all hang separately."
While the consequences in the situations I've applied this concept to are never as dire a life or death struggle as the original (although it may seem like it at times), it can apply in a lot of networking cases where there is a common success or failure that will occur based on the decisions of all the pertinent players.
Often, for example, when ring WANs are developed to serve a number of organizations (be they institutions or businesses), the choice of the right common transport, the proper routing through each organization and the support that each entity gives to common equipment can have a significant effect on the viability of the network for all of the users.
Institutional Networks (I-Nets) that have been successfully developed for interorganizational communication have had to embrace the "hang together" concept on many fronts in order to achieve successful operation. A number of recent I-Nets have been developed along the classic ring-hub-star configuration, where intra-organization communication occurs on the star portions of the network; each user's data center serves as the hub for their access to the ring; and the ring provides transport between each organization's sites and to common destinations (such as Internet Service Providers, Application Service Providers, etc.). While each organization may make its own decisions on its particular star connections, the use of a common transport around the ring will typically be necessary to make efficient use of fiber infrastructure; provide for a common interface for inter-organizational traffic; provide for efficient aggregation of traffic; enable redundant, fault tolerant backbone operation; and provide for ease of expansion of inter-organizational applications. While these all sound like prudent requirements for WAN development, getting a number of different organizations (with different IT staffs, different communications networking philosophies and, most likely, an emphasis on different primary inter-organizational applications) to choose and/or support a common transport system can be a little like trying to herd cats.
Often, as you try to move toward a workable solution that, at a minimum, will satisfy each organization's basic inter-organizational communication goals, you can run into three camps. Specific-ally, they are:
- The Native Etherneters: Theirs is an IP world where VoIP reigns, and all video can be streamed. Gigabit Ethernet (GbE), perhaps employing GigaEther-channel between hubs, is typically their WAN transport of choice. They argue that this will allow seamless LAN/WAN transmission of data from all of their sites to the sites of other organizations, and is eminently scalable from 10 Mbps to 100 Mbps to Gigabit, with 10x Gigabit on the way.
Their detractors argue that QoS (Quality of Service) issues are still not resolved, which are critical to the operation of a common transport. Further, in order to ensure successful operation of the WAN, they will still need a standalone GbE edge device (as opposed to an uplink module integrated with their data hub switch) in order to ensure that the WAN can be isolated from their internal network as necessary.
- The ATMers: To them, ATM is the standards-based, rock-solid, well-proven technology which enables exemplary QoS that guarantees bandwidth for critical voice and data services. It also can satisfactorily facilitate video transport. They also argue that it currently provides for the most cost-effective implementation. Their detractors argue that it is "old" technology which is not efficient for high-speed data transport (often citing the significant overhead needed to transport ethernet traffic in a cell format). They also argue that its scalability (from 10 Mbps to 100 Mbps to OC-3 to OC-12, OC-48, etc.) is not cost-effective once site-to-site transfer rates at gigabit speeds need to be achieved.
- The Sonetans: If fiber transport is the basis for the network, and digital is the format, then this group says, why not choose Sonet, since it was developed for just these purposes. They argue that complex inter-organizational networks that must facilitate a host of different applications need a sophisticated transport like Sonet that can efficiently multiplex digital streams from a variety of sources. The opposition here largely complains that Sonet's cost can well exceed the other two, and especially for school districts and other institutional users with tight budgets, will not provide a reasonable cost benefit.
So, what to do? Run three different systems? Well, some large networks serving well-heeled institutional or business users run more than one. But for many, it is important to choose one underlying WAN technology. For instance, one choice allows for aggregate purchasing where all users can benefit from larger quantity purchasing power. Organizations can band together and issue a joint RFP for network equipment, training and equipment maintenance. Spares can also be purchased and shared between organizations.
One choice also enables easier administration of the ring portion of the WAN. Often, common transport on an institutional WAN will be administered by a User's Group, an internal or outsourced administrator hired by the group, or a lead user that may have more expertise than the other participants.
There are a number of application-oriented issues to look at in order to ultimately arrive at a common transport system, all driven by the common applications designed to traverse the ring and also potentially interconnect through a gateway to other regional WANs. But in summer parlance, it's time to hang 10 for now; see you in September with more.