A brief history of MV

Here is a little bit about the history of MV Communications-- how we got started; how we happened to become the first ISP in the state, several years ahead of most everyone else, and one of the first in the country. Parts of this are drawn from our July 1996 newsletter celebrating 5 years of incorporation; parts from a speech written for the first annual NHISPA conference in October 1999; and parts written for this web page over time.

Incorporated on June 21, 1991, MV Communications is one of the oldest and most experienced ISPs in existence. We:

• were one of the first companies to offer affordable Internet access to the public at large;

• were one of the first to introduce high speed dialup access technology (first with Telebit 23000 baud, then the new 28.8Kb and 33.6Kb modems, and later "56k" and ISDN;

• were one of the first to offer IP access on a casual dialup basis;

• were one of the first to provide Internet domain registration and unique email mapping within those domains;

• were one of the first independent ISPs to offer Frame Relay access in New Hampshire;

• were a charter member of the national ISP/C organization;

• were a founding member of the NH ISP Association and helped to provide the impetus for peering amongst local ISPs;

MV's Beginnings

To increase the speed and reliability of local mail, Mark registered the domain "mv.com" in late 1987 (no small feat at the time), and established a UUCP connection to Harvard University as a prime mail gateway. Computer professionals and others in the area were allowed to connect into the "mv.com" domain, and Bruce Dawson volunteered his system in Nashua as another gateway into the domain. By the beginning of 1991, there were some 40-50 members of the "mv.com" community. The two gateway systems were becoming overloaded, as were the budgets of their owners who were paying fairly significant out-of-pocket expenses to support the domain and the many long-distance data calls. Additionally, the number of requests for connections and domain membership was rising dramatically. Bruce Dawson moved away from Nashua, leaving one mv.com hub system to handle an ever-increasing volume of email and Usenet news.

Mark was faced with the prospect of refusing new connections and possibly trimming back some existing ones, a course that he believed would make the mv.com domain experiment a failure. An alternative was to try to charge to recoup the cost, but he felt that in order to charge money for connectivity, something more had to be offered. He then began to investigate the idea of acquiring a direct connection to the Internet, something that would vastly improve the reliability and performance of electronic mail to the mv.com subscribers. At that time, this was a pretty far-fetched idea. It was uncertain that it would be possible to convince enough people that Internet access was worth paying for. Nevertheless, Mark hornswoggled some friends and associates into trying it, and in June of 1991, MV was incorporated and moved into its Litchfield office with a single 386/33 UNIX server and 4 Telebit Trailblazer modems attached to dialin lines. In September, after several months of delays, a leased line was installed into NEARnet, and MV began its commercial operation offering UUCP accounts.

The Participants

Mark Mallett

Mark was involved in computer communications since the mid-70s, sometimes being paid for this involvement but often doing it on his own. The details can be long and boring, but if you want to subject yourself to them you can check out personal history links on his personal home page at MV. In summary, Mark wrote about a half dozen email systems from the mid-70s through the mid-80s; wrote BBS software and ran a BBS out of his house; was involved in projects writing low-level and high-level communications software using a number of platforms and protocols. Somewhere along the way he met David Dick and Kenn Goutal while on a contract with Hendrix in Manchester to build a publishing system for the Chicago Tribune newspaper. Through this association he was introduced to the Usenet in 1981/1982 via Software Innovation's facilities, and around the same time was exposed to the ARPAnet through another contract at Sanders Associates. These things eventually led to his creation of the Usenet hub "zinn" referred to earlier, and the registration of the mv.com domain and so forth.

Marta Greenberg, David Dick, and Sii

Marta and David were principals of Software Innovations, Inc. (Sii), a company whose focus was the migration of large scale software systems from various platforms over to UNIX. Through a contact made at Hendrix in Manchester, it happened that Sii was one of the first companies hooked up to the Usenet -- in fact the earliest published Usenet map, one that fits on an 8x10 sheet of paper, lists Sii as one of the Usenet nodes. (See it here.) Sii operated their Usenet and UUCP mail hub in Nashua, NH for a number of years, and provided guest accounts to some of their compatriots including Mark Mallett and Kenn Goutal.

Kenn Goutal

Kenn also contracted at Hendrix in Manchester in the late 70s and early 80s. Hendrix was one of the hot spots for contractors with experience in DEC systems and varied backgrounds, and the project with its distributed computer architecture naturally fostered lively discussions about computer communications. Kenn was actively interested in these topics; pursued, digested, and shared research material on them; and for both professional and personal interest made use of many communications systems such as Delphi, Usenet, BIX, ARPAnet, and others.

Other influential people:

David Betz

Another contractor at Hendrix, David was an active participant in our conversations and a prolific implementor of ideas. David wrote a dialup network system that shared some characteristics with the UUCP network, but could be run on lowly personal systems such as those that ran CP/M. (David is also the author of XLISP, and later in the 80s was a chief software engineer at BIX, the large online service started by Byte magazine.)

Bruce Dawson

Bruce is a well-known figure in the UNIX and software consulting community in NH. As mentioned above, Bruce provided one of the two main hubs in the operation of the early mv.com domain park until he moved away from Nashua. After MV was incorporated, Bruce provided significant help and resources in the formation of our Dover node.

Don Fairchild

Don and Mark Mallett worked together as contractors at Calcomp, and it was over lunches in the Calcomp cafeteria that the idea of the mv.com domain park was discussed and, indeed, the name for the domain was chosen.

Our Formative Years

A backbone connection

Our initial Internet "backbone" connection was a 9600 baud digital leased line from NEARnet. The installation of this line was more expensive than T1 service would be just a few years later, and likewise the monthly service fee was comparable to that later-day T1 service. At this time, the Internet backbone was mainly run by the National Science Foundation (NSF), a part of the US Government. Regional backbone services such as NEARnet provided connections, mainly to educational and research institutions, on a membership basis. Membership in NEARnet had to be carefully justified and accepted by the NEARnet board on the merits of that justification. After a lot of work and a lot of negotiations, our membership was finally approved -- albeit with a number of restrictions and conditions.

In the spring of 1992 we upgraded our backbone link to the heady speed of 56Kbps.

During the early years, a lot of Internet traffic had to be specifically non-commercial in nature. The ANS commercial backbone was built in parallel to the NSF backbone; commercial traffic *could* be carried on this backbone between parties who subscribed to the ANS backbone service in one way or another, such as subscribing to a "commercial routing option." (Since ANS backbone was often the same physical fabric as the NSF backbone, the distinction of traffic on the NSF backbone vs the ANS backbone was sometimes rather fuzzy.) The commercial routing agreements also allowed some traffic to be handed off to other commercial backbones that were in early development at that time, such as UUnet and PSI.

Evangelizing the Internet

One of the difficult tasks we faced was convincing people that Internet access was useful, and that indeed it was useful enough to pay for. There were several obstacles to this. For one thing, most of the old-time technical people who already knew the value of having an electronic connection were offended at the very thought that it should be paid for -- regardless of the fact that it cost real money to provide (a lot of money at that) and that it brought them real value. (For what appeared to be very related reasons, we also encountered this attitude from some institutions, notably the Boston Computer Society.) For another, there was no web; what people mostly got out of their connection was email and Usenet news. And very significantly, there were no convenient software platforms that were easy to use on any system other than UNIX. (There were some UUCP packages that facilitated the automatic exchange of mail and news, but use and installation required a good deal of technical skill.) Nevertheless, whenever we could convince people to sign up, the reaction was almost universally "how did we get along without this!" We found ourselves doing a lot of speaking around the state, talking about the Internet and occasionally being able to do live demonstrations using gopher (an early menu-based hypertext system that preceded the spread of the web). This went on for a few years; once the web came along and better software was available, live demos were much more reasonable and acceptance was more rapid.

NHNet

In early 1992 a project was commissioned by then-Governor Gregg to demonstrate networking in New Hampshire. Several hundred students, teachers, and selected other participants would have dialup access to the "NHNet" experiment. They would be able to access certain information sources and communicate with each other. At almost the last minute, the project leaders realized that they ought to be including Internet access; and since MV was the only company offering access to the public in NH, we were asked to participate. We had about a month to get ready and we spent almost every minute of every day (including the entire Christmas and New Year season) working on it. We leased some computer equipment from Shannon Computer in Alton, worked to make it Internet-compatible (a very large project), and developed menus and special software to allow these hundreds of people to use the Internet.

In the end, MV was the only participant who had to spend real money to participate in this project, and Internet access was by far and away the most successful part of the entire 6-month program. (There was very little mention of MV in the press reports, and MV was kept at arms length during almost all of the project by the bureaucrats in charge of it, yet the people who used the system and got their first taste of the Internet were very much aware of our contribution.)

IP and Shell access

Through extensive negotiations with NEARnet, we were able to modify the terms of our contract so that we were allowed to provide dialup IP access as of mid-1992. This access was provided at that time using the SLIP protocol. For various reasons we kept this access fairly closed until mid-1993. However, during the spring of 1993 we made an agreement with a group of consultants led by Marc Evans (who later went on to form The Destek Group ) to provide dialup SLIP and PPP to that group, and this led to us opening up SLIP and PPP access to other customers in general. Initially this service was provided on Xylogics Annex terminal servers (acquired as part of the above-mentioned deal), but in the fall of 1993 we began purchasing and using Livingston Portmaster equipment for all dialin access, including shell, IP, and UUCP.

In March of 1993 we began offering UNIX shell accounts (which we rather foolishly referred to as "online accounts"). Shell users had access to a UNIX command prompt, or to a menu system that we developed using gopher. We also provided ClariNet news to all shell users.

Hello Sprintlink, Goodbye NEARnet

Towards 1994 we needed more bandwidth and we also needed relief from the restrictions of our NEARnet contract. We placed an order for a full T1 from Sprintlink, part of which we had agreed to resell for a time to Destek in Nashua. The Sprintlink circuit came up in April 1994; the NEARnet 56Kbps service was discontinued at the end of June.

Sprintlink was a commercial carrier. Use of commercial traffic on the Internet was still evolving; at this point we needed to be a member of the CIX (Commercial Internet eXchange) in order to ensure that our commercial packets were sanctified and able to reach destinations on the NSF/ANS backbones.

POPs

In order to offer wider local dialin coverage, we created POPs (points of presence) in various local calling regions in New Hampshire. Each POP was connected to our main facility via a high speed line, and had its own modem pool with local numbers that you could call. When we started creating POPs, we installed computer systems for shell and news servers at each POP location. Changing usage profiles led us, after a year or two, to shut down those remote computer systems installed at the POPs and move their services back to our main location. POPS were established at:

• Concord: November 1994;
  
• Dover: October 1994;
  
• Peterborough: December 1994;
  
• Salem: January 1995;
  
• Keene: February 1996;
  
• Nashua: March 1996;
  
• Manchester: April 1997.

More backbone

By the second half of 1996 we began to look for more bandwidth and for redundancy in our Internet backbone connection. (When you've only got a single connection to the net, an outage on that single connection is potentially severe.) In January 1997 we contracted with MCI for a T1 connection. This ended up being a fiasco, with MCI forgetting our contract, a salesperson leaving and being replaced by one who didn't normally sell Internet and didn't really care to sell to us; nobody to complete our order and nobody outside of the local office to take responsibility. We never did get an installation from MCI, although they began billing us some months later and continued billing us for over a year, severely hurting our credit rating during this period. We eventually had to resort to legal means to make it stop.

Because of the fiasco with MCI, in May of 1997 we purchased some bandwidth from our friends at Destek as a temporary measure. This temporary solution worked out well enough that we kept it.

The Move to Manchester

You can see our moving notice, as well as links to some photos of the empty office space before it was an office space, if you follow this link.

Our big move came on June 16, 1997. We had prepared well: we had established Internet connectivity to our new office area when that space was just a hollow hole in an office building. Our servers were moved in the wee hours of the morning, with only a couple of hours of outage, and the rest of the stuff moved during the following day.

After the Move

Brooks

In early 1998 we contracted with Brooks Fiber, a competitive local exchange carrier (CLEC), to provide us with local statewide dialup coverage. This contract would accomplish several things: it would provide us with pure digital dialup lines so that we could support "56k" (V.90) protocols; it would provide us with statewide local numbers without our having to deploy POPs in local areas; it would let us offer ISDN dialin access on all of our Brooks numbers; and it would allow us to take down the POPs that we had already deployed.

However, since this was a fairly new telephone service, and because the "56k" protocols were still evolving, we felt it necessary to maintain our old analog dialups and POPs in parallel with the new Brooks service. We ended up maintaining two complete sets of dialup facilities at great expense for quite a long time -- the end of this came when our last our last Litchfield modems were removed some 20 months later.

The Brooks service was delivered to us via fiber direct from their facility and was brought up in May of 1998.

End of POPs

As mentioned, because of the new digital statewide dialin service, we eventually decomissioned all of our POPs:

• Keene: December 1998;
  
• Concord: April 1999;
  
• Salem: May 1999;
  
• Peterborough: May 1999;
  
• Dover: January 2000;
  
• Litchfield: January 2000.

(As of this writing in July 2000 our Nashua and Manchester POPs are still in service.)

More bandwidth: BBNPlanet

In November of 1998 we installed another T1, this one to BBNPlanet. BBNPlanet more or less grew out of NEARnet, so here we were back with our original backbone connection. However, BBNPlanet found that they had overbooked their facilities in New Hampshire and asked us to shut this new link down for a short time while they upgraded. The short time ended up being 2 months, and we were back up with them at the end of February 1999.

BA fiber

In the fall of 1999 we had Bell Atlantic fiber installed. This let us put some of our customer high-speed circuits onto a high-speed T3 mux. Using this fiber we also installed a T3 ATM service for higher-capacity peering to Destek in Nashua. This fit in with a recent arrangement that we had made with Destek to provide us with access to a backbone connection to Frontier (now Globalcenter); we opened up this access to 3Mbps (two T1s worth).

More bandwidth changes

In October 1999 we dropped our Sprintlink connection, partly because of their resistance to moving their circuit from Litchfield to Manchester, and partly because of their unwillingness to offer us their current terms. We had placed an order with UUNet in June; this finally became ready in November and essentially replaced the Sprintlink backbone connections. Still feeding the demand, we placed an order for yet another T1, this one with Cable and Wireless (who had bought the old MCI backbone and were much more pleasant to deal with than MCI had been.) This circuit was installed in January, 2000.

Finally gone from Litchfield

In April 2000 we finally removed all lingering circuits from our old Litchfield office. (We had kept some high-speed circuits there while we still had a modem presence; after the modems were removed we were ready to be gone from there, and we pulled those remaining circuits and left.)

DSL

In a culmination of nearly a year and a half of effort in finding what we felt was an appropriate partner to provide us with underlying DSL loops, in the spring of 2000 we signed on with two different companies, as documented in our DSL service pages. After some delays, our DSL ordering began in the fall of 2000.