Demand for frame relay and even ATM switched virtual circuits never fully took off. Despite our enthusiastic writings about the scalability and automated user self-provisioning potential of SVCs back in the early days of this newsletter, SVCs got relegated to a niche.MCI was the one major carrier that offered frame relay SVCs.\u00a0 AT&T offered ATM SVCs, explaining that multimedia applications like voice and video tend to have a stronger peer-connectivity requirement than data. And back then - when this newsletter began life in the spring of 1998 - ATM was the multimedia protocol of choice.But a few trends persisted to bump SVCs off the list of must-have services:* Once frame relay was positioned as a less expensive leased-line replacement, frame relay topologies largely mimicked those of private networks, in hierarchical hub-and-spoke configurations. Enterprises could easily compare leased-line and frame relay costs for connectivity configured in exactly the same way. For data applications, many remote sites needed access to a hub location only. And if one site needed connectivity to a peer site, a small performance delay in traversing the hub site en route didn't impact the user experience much if only data was involved.* Voice over frame relay, while certainly do-able and popular in very highly distributed companies such as the Burlington Coat Factory, never really went gangbusters. Voice should have been a strong driver for SVCs, because users need to talk directly to other internal users. But in the U.S., the cost per minute for public phone network calls got so cheap for very large companies that they probably felt wasn't worth investing in voice FRADs and configuring them.* ATM proved itself to be quite complex for enterprises, which were rescued by the advent of 100M and, eventually, 1G and 10G bit\/sec Ethernet. As a result, ATM got relegated more to a role as a high-speed carrier aggregation backbone than as a subscriber service, at least in this country.* Last but certainly not least, IP VPNs emerged to steal the thunder from SVCs by enabling simple peer connectivity using connectionless IP addresses. Particularly if you are using an Internet-based IP VPN service, your reach becomes the size of the entire Internet.We're still waiting for user self-provisioning, whereby you will be able to turn up a circuit, resize your bandwidth, and so forth. Some of the efforts of the new Frame Relay\/MPLS Alliance might eventually result in these automated capabilities.