Print 

Caimis Launched as CAIDA Commercial Spin Off

Goal Is to Give Internet Industry its First Production Quality Network Management, Traffic Analysis and Location Sensitive Measurement Tools

pp. 1 - 8

We interview Tracie Monk, Kim Claffy (kc), David Moore, and Daniel McRobb on the details of CAIDA's creation of Caimis. Caimis www.caimis.com is a profit making spin off from CAIDA that will take many of CAIDA's research oriented, network traffic management tools and upgrade them to production quality commercial software for use by major ISPs. Kc explains the CAIDA Internet Teaching Laboratory program, the san Diego Nap and plans for OC48 mon - as well as CAIDA's work on defining and understanding the Internet core.

The Caimis web site went live on July 12. This interview is the first detailed description of Caimis to be published. Caimis will focus on monitoring systems, network management systems and geographic location systems. In the monitoring area it will offer a variety of solutions that are for passive and active monitoring through active polling and passive analysis of router and link data. Its network management systems will be used for acquiring, storing, correlating, and analyzing data. The third area, geographic localization services, involves a set of solutions aimed at mapping different types of network data to geographic locations. It is the latter area that has direct relevance for e-commerce and financial transactions.

In June Caimis formed an affiliate company, Caimis Geo, Inc., to develop the geographic location services business. These will be tools aimed more towards content providers or e-commerce websites, where they want to know where their customers are actually located - perhaps because they have restrictions on distribution of software to people in different regions, restrictions that are geographically based or by country. Or, you might have web sites that might want to change the language presented on their homepage depending on which country the user is coming from or return targeted content based on where the user lives, e.g., content appropriate to Southern California versus Boston.

Caimis finds that ISP service level agreements generally still do not measure up to the standards that business to business customers expect to see. Hewlett Packard, Cabletron and Seagate produce network management systems but do so without an adequate understanding of the Internet problem domain. While current solutions can tell network engineers what's going on with a particular router, when that ISP needs to move up to something like defining an end-to-end service for a particular customer, there is a big void. Providing such solutions are challenges that happen both within and between providers.

When you start talking about inter-providers, the situation gets worse. Information can't be exchanged. Providers don't want to provide any sort of service level guarantees for infrastructure that they don't control and to which they have no access. Among the things that need to be done are the creation of some sort of common interfaces that providers can use within their own network, but between different systems. For example provisioning and network management itself, and element management. Today, basically most of the network management systems sit in the element level, meaning they tell you what's going on with particular devices in the network. They don't interface well with any of the other business processes within a provider. These are the customer care processes, billing, provisioning, ordering, trouble-ticketing, which is a big problem in providing service to the business-to-business environment. Ironically in this respect the ILECs have better systems than the Internet industry. Caimis intends to solve that shortcoming.

Caimis has two offices. The San Diego headquarters is focusing on the development of the geographic-locations services under Caimis Geo, Inc. The San Diego headquarters will also be doing all of the productization, sales and distribution of all of Caimis monitors and network management systems. The Ann Arbor office will be the leading edge site for senior software and network engineers who are doing customized network management solutions. They will be the ones staying on the cutting edge technology-wise and requirement-wise for the large providers. San Diego will take some of these solutions and develop standardized products that can then be distributed and made available for the smaller ISPs, enterprises, and others.

Cogent Communications Guarantees 100 Mbs Internet for $1000 per Month

Buys Dark Fiber and Builds National Network Optimized Entirely for Data

Business Model Emphasizes 100 BaseT LAN Interconnection in Buildings it Serves

pp. 9 - 14

We interview Dave Schaeffer, CEO of Cogent which n July 17th announced $90 million dollars in second round venture capital fnding.. Cogent is the first national ISP to be formed with the business model made possible by the new combination of SONET free fiber, gigabit Ethernet and dense wave division multiplexing. This technology permits IP data networks to be built for only one one - hundredth of the cost of their voice equivalents less than 10 years ago. It will use a two hundred million dollar 30 year IRU on a pair of fibers from the Williams Communications Network and fiber from Metromedia to connect office buildings in twenty cities nationwide. It will offer tenants of these buildings 100 megabit per second internet connectivity for $1000 per month. Cogent has signed commitments to spend more than $600 million on its build out. Without SONET equipment Cogent is protecting its network at layer 3. It describes itself as a facilities-based nationwide switched LAN.

Gigabit Ethernet is used in their local loop. SONET framing in long haul. The SONET framing is a transport standard that is built into the Cisco routers. Consequently there is no additional cost. In the event of a fiber cut instead of 50 milli-second SONET based recovery, they get one second recovery at the router level which, for data, is adequate. Any physical disruption in service would cause the core routers to start rerouting packets. This process takes about one second. Cogent promises to give its customers a full 100 megabits of band width pledging not to put more than 96 on a OC192 - or 9.6 gigabit - loop of 1500 miles in length.

Schaeffer explains how Cogent selected its fiber providers. He notes that: "There are some vendors whose equipment will only work with certain physical shelter spacing. As we evaluated our potential dark fiber vendor, we looked for that network vendor whose shelter spacing would support the maximum number of equipment vendors. Note that the shelter spacings are not inter connect points. They are there only to amplify the signal. All that we need to be able to do is to rent standard size rack space in the shelters." He points out that "fiber can be optimized for the number of wave lengths, for distance or for the power applied to it. There are a number of different design criteria and each vendor takes a slightly different approach to which of those criteria they emphasize with their fiber."

Schaeffer developed a matrix for the evaluation and selection of both his long distance and metropolitan area software. He discusses in detail how Cogent arrived at the decisions for selecting and integrating both sets of equipment. Cogent's Network will open in New York, Philadelphia, Washington and Chicago in October.

GAO Report Shatters ICANN Founding Myths: Neither Privatizaton Nor Power Sharing

Policy Control Over Operation and Content of the "Authoritative" Root - ICANN's Single Biggest Prize - Remains with US Government

pp. 15 - 21

We describe how ICANN marches onward to the chant of its own version of reality. The GAO report released in July 7 found ICANN's creation to have been legal but it could not find any support for the idea that the Department of Commerce had the right to manage the root. It concluded that it was also very doubtful that the DoC had the legal right to give control of the root to ICANN saying that it did not look exhaustively at this point because the DoC told it that it had no intention of doing so anyway. In Yokohama only a few days later Beckwith Burr proceeded to muddy the waters by stating that she was prepared to give ICANN operational but not policy control of the root. An ominous development considering ICANN's statement that it was ready to demand that the root sever operators sign contracts with it.

After having caved in rather than face discovery on its first lawsuit, at Yokohama the ICANN Board proceeded quite skillfully to disarm its next, potentially more serious, opponent IO Design. Having given the impression that it might rule on who would get in the root, it collected a number of TLD applicants on its web site, lied that it had consensus for only a very small number of new top level domains, (between 1 and 3) set up a $50,000 application fee and then announced that it would receive applications August 1 and announce the winners on October 1. In one swift move it deprived IO design of both due process and anti-trust grounds for suing it. It will choose on October first .eu and a gTLD that will have no negative baggage for its trademark rulers. Dot web will not be in the running.

The Board also showed its continuing fear of letting in any outside influence that might expose its operations to public view. For the ninth time in less than two years ICANN changed its by-laws. To ensure continued control by the original Dyson, Roberts, Touton cabal, it extended once again the terms of four of the original directors through the annual meeting in November 2002. It decided to study the question of whether it should even have an at large membership. In its efforts to make certain that no one outside the original cabal would ever have a voice it entered the following paragraph into its by laws: "The Corporation shall not have members as defined in the California Nonprofit Public Benefit Corporation Law ("CNPBCL"), notwithstanding the use of the term "Member" in these bylaws, in a selection plan adopted by Board resolution, or in any other action of the Board. Instead, the Corporation shall allow individuals (described in these bylaws as "Members") to participate in the activities of the Corporation as described in this Article II and in a selection plan adopted by Board resolution, and only to the extent set forth in this Article II and in a selection plan adopted Board resolution."

Optical BGP Proposed as Means of Wide Area Interconnection of Bandwidth-Rich Edges Without Burdening Largest Backbones With Additional Traffic Created by Gigabit Ethernet and Dark Fiber Driven Local Bandwidth Explosion

pp. 22-24

"OBGP is a proposed extension to BGP for the manipulation of optical cross connects to permit them to be automatically setup and configured as BGP speaking devices to support multiple direct optical lightpaths between many different autonomous domains. OBGP may also allow customers at the edge to control a subset of lightpaths within another network's wavelength cloud so that they can manage their own light path routing within that cloud." . . . . [We suggest treating] 'each optical cross connect as an independent virtual BGP router with only one input port and one output port. A virtual BGP router can then be set up for each optical cross connect and separate BGP sessions initiated with its peers." . . . . "The physical characteristics of a lightpath give it an intrinsic capability of being a "poor man's" logical switched path with a predefined Quality of Service. " . . . . "The exchange of lightpaths may also allow for a simpler mechanism to allow for settlement in peering and transit between ISPs." . . . . "In future there may even be wavelength commodity markets where ISPs can trade wavelengths and adjacencies on the open market."