Computer Science has recently begun a significant upgrade of the Sitterson network. This project will involve all aspects of the network--the physical infrastructure, logical network, and protocol management. This page is intended to be the primary source of information on this project for affiliates of the Computer Science departments, and will be updated regularly. At this time, all information presented here is preliminary and subject to change, and we want feedback from all interested parties. We therefore ask that anyone interested in the project review this information and contact us with comments or suggestions. We are also maintaining a timeline of sorts to keep folks up to date on what's coming.
For better or worse, this project has been significantly delayed. Although we have been aggressive in providing ATN and Lockwood Green with information regarding our needs, and they have been responsive, the plan we've developed is quite complicated and unique among the 22 buildings in this major phase of rewiring UNC CH campus buildings. So Sitterson has been moved to the bottom of the list, so that other more standard buildings could be completed. The section in use by CS affiliates in Phillips is part of the penultimate phase, however, and work began on that building in fall of 2000 and is due to be completed in May of 2001.
In a way, this has worked to our advantage, in that we have had the chance to use Phillips as a "dry run" for the Sitterson upgrade, and the experience we gain in working to complete that project will aid us in making sure that the Sitterson wiring plant is everything we need it to be once completed. Given that we will likely not have another chance to rewire the building completely for as long as 10-15 years, it is better that we get it right, than to get it in a hurry.
We have established a small Network Upgrade Planning Group as a subcommittee of the DFC composed of individuals to facilitate this process and to act as liasons to different groups with Computer Science. Team members will develop strategies and review each stage of the process to insure that all needs are reasonably met. They will also collect and disperse information to insure that all relevant parties are kept well advised regarding the goals and status of the project. Members include Bil Hays, Don Smith, Leandra Vicci, Russell Taylor, David Musick, John Sopko, and David Harrison. If you have any questions or suggestions, please feel free to contact these folks.
The list of groups involved include:
The will be a lot of meetings:
ATN Networking and Communications has received a significant level of funding to upgrade networks in the College of Arts and Sciences in order to provide departments with new state of the art switching equipment and implementing building rewiring where needed (for a list of wiring projects see http://www.unc.edu/~gogan/wiring.html). In addition to the project to upgrade the production network, we will also be pursuing the establishment of an experimental high speed network to link selected campus buildings for use in experimental projects, such as that planned by the Nanomanipulator group and the Colab. The Sitterson networking upgrade is one of twenty-two buildings slated for upgrades with these monies. ATN will be providing expertise, equipment, and will be coordinating the physical upgrades to the building per our specifications in conjunction with Lockwood-Green (the firm that has been awarded the contract for designg these upgrades and writing the RFPs for contractors). We will provide ATN with our operating requirements, locations where and of what type connections and equipment should be installed, review plans at each step of the process, and finally test all installations to make sure that our specifications are met or exceeded. This opportunity will allow us to make significant improvements in our production network without significantly affecting our budget, and will free up money to be used for other projects (including additional networking upgrades).
As the project proceeds, there will be three main phases, an immediate equipment upgrade to upgrade existing hardware (badly needed at this point, since our existing switches are overloaded by the new machines from the Intel grant), a planning phase to detail how the building's wiring should be improved, and a wiring phase as those improvements are actually made. Each phase will involve a number of review points, allowing all parties to review progress and reevaluate upcoming events. We will thus be working closely with ATN and Lockwood-Greene throughout the project to insure that the unique needs of faculty, staff and students working in Sitterson can be met, and we believe that through a cooperative effort, Sitterson can become the model of high capacity networking on the UNC-CH campus. To guarantee that this occurs, we will ask for a written document of understanding detailing service level agreements and control issues signed by both groups, with the Chancellor's office specified as the final authority in case of disputes.
Sitterson's current network is composed of older category 3 twisted pair and AUI thickwire linked via switches and transceivers. Uplinks are handled by three older FDDI rings and a newer horizontal/vertical ethernet network. Individual computers are linked over this wiring topology via a mix of AUI repeaters, switches, and transceivers. Access to the internet is over a single 100 mbps Ethernet line to Phillips. Although originally state of the art, this scheme no longer consistant with the minimum set in the Uniform Wiring Plan (the established campus standard), is difficult to maintain, and ultimately does not support the heavy demands for high speed networking found in a major Computer Science facility.
Although the project to completely rewire Sitterson has been delayed, we've made significant improvements in the network by upgrading to new switches and installing new wiring in the core of the building. We've
These changes have had enormous benefits:
In the long term we intend to augment or replace existing wiring (Coax, AUI, and Category 3 twisted pair) in favor of:
These plans are work in progess and represent the best estimate of the networking upgrade group as what additional wiring should be installed during the project (wiring already in place is not shown in these diagrams). They will change, and often, as the project proceeds. If you find an omission, or do not think the plans will meet a particular need, please let us know as soon as possible. The gif file is large, and may take a while to load.
All wall apprearances not below the floor or above the ceiling will be flush mounted, and the standard height will be just above desk/terminal tables and just below white and cork boards.
To facilitate feedback, we will begin placing pictures of outlets in the actual location of placement beginning in October 2001. PLEASE DO NOT MOVE THESE PICTURES!--it will only delay construction since we have to get approval for almost all changes. If you feel an outlet is incorrectly placed, contact the network upgrade planning group.
Many people have brought up the question of why use copper cabling instead of fiber for the majority of the building's ports. We want to emphasize that this is not a case of "either/or" but rather of which will be the predominate scheme in Sitterson for the typical workstation.
Currently a typical workstation shares a 10 mbps connection to the fiber ring with 4-10 other machines. Both fiber and category 5 copper offer throughout approaching 100 mpbs per station, so either represents a dramatic improvement. Cost becomes the critical issue in that fiber is significantly more expensive to deploy widely--network interface cards for fiber run between $200.00-$1500.00 each and transceivers capable of reaching 100 mbps run between $200.00 and $500.00. 10/100 mbps NICs, on the other hand are typically included with new PCs, and cost between $50.00 and $150.00. Fiber cabling is more expensive to install and terminate, and more difficult to handle than copper. Given the cost differential and that the difference in throughout is small, deployment of category 5 copper on a wide scale basis for general use makes sense, as does augmentation of existing fiber installations for use with equipment that requires it or comes equipped for it. Fiber does, however, offer a number of advantages over copper. It is not prone to electro-magnetic interference, and fiber runs can be 2 km long (as opposed to 100 meters). Currently it is used for faster connections than can be achieved over copper. We already have significant quantities of fiber installed of various types and grades, and we intend to augment and upgrade the existing fiber to allow for it's use with newer technologies such as ATM and Gigabit ethernet.
Long story short, we need both fiber and cat 6 copper in significant quantities.
