In a recent newsletter, we mentioned that users do not seem to want WAN optimization tools that are too complex. This attitude was further confirmed by a user who was kind enough to write, "I have found the same effect at my government agency. They are afraid to implement WAN bandwidth optimization tools such as [the tool from a specific manufacturer] because they are afraid it will be a new point of failure and that it will make troubleshooting difficult."At the same time, there is a most reasonable argument that having independent devices for troubleshooting purposes can provide a great service. For instance, a strong historical argument for a separate, intelligent CSU\/DSU as opposed to integrated CSU\/DSU has been that it provides an unbiased point for troubleshooting.So we've become intrigued by the question of how users feel about "one box" solutions as opposed to having multiple devices. In a recently released study by Sage Research (conducted on behalf of Cisco), the finding was that users found a significant advantage to implementing multifunction edge devices as opposed to multiple point solutions.\u00a0Interviews were conducted with a roughly equivalent number of customers who had and had not implemented multifunction devices. And while there's not space in this format to share all the results, 56% of those with integrated devices reported that it was easier and\/or faster to implement new services; and half reported improved network uptime.\u00a0 For the full results, a presentation (including an MP3 podcast) is available at Webtorials.But we're also interested in hearing your opinions. What do you see as the optimal number of edge devices? Independent devices allow for "best of breed" mixing and matching, and the failure of one device, depending on the configuration, may or may not bring down the entire network. To rephrase the question, how many eggs should you put into a single basket, and what happens when the basket has a hole in it?