A NEW INDUSTRY LOOKS AT THE BELL
MARCH 1, 1971
The joining of computers to communication networks is like the marriage of a less of what each fully understands reality, however, young campus rebel to a rich widow. Regardless of what each is looking for in the marriage, neither the attitude and needs of the other. In many of the problems that arise are psychological rather than technical.
Fortunately, marriages can often be saved by the use of group therapy and I suspect that is what we have tonight. Such therapy is only useful if the participants are very frank and I hope for that reason you will accept my comments in the proper spirit.
First, I believe it is important that we define the new computer services industry. Leaving aside that segment of this new industry having to do with application package development, consulting and other areas that are primarily manpower industries, let us consider where most of the interface between the common carriers and the industry takes place. This revolves around the remote access to computers.
Some of this access is through low-speed key-driven terminals such as the teletype or the IBM 2741 typewriter terminal. This is an interesting area that has problems of its own. Among these are the requirement to use a full voice grade line when in fact a 160 baud sub-channel would be quite adequate. This means there is a demand for the multiplexing of such service to gain better use of a voice grade line.
Another problem in the low-speed area involves the use of acoustic couplers. Carrier resistance to these devices is understandable from the viewpoint of potential network pollution, but is an example of customer demand running ahead of the apparent ability of the network to adjust to a desired product.
A third problem often discussed in the low-speed area is that of the method of tariffing data transmission. As has often been noted, the key-driven terminal user occupies a line for a long period of time but, in fact, puts very little data over the line. The time/distance rate structure tends to make this type of transmission very uneconomic to the customer.
I only mention these problems in the low-speed area for the sake of completeness. The rest of this evening’s talk will be directed toward the medium and high-speed service as this is the one on which our company depends for most of its revenue.
THE SDL CONFIGURATION
Systems Dimensions Limited operates one of the world’s largest and fastest computers, an IBN System/360 Model 85. The data input/output bandwidth of our computer, as presently configured, exceeds 36 million bits per second. Through the addition of presently on-order channels and input/output devices, this figure will, by mid-1972, exceed 60 million bits per second.
Common carrier data communications are extensively used to bring our customers’ work to our computer for processing and to return to them their results.
Our presently installed data communications facility utilizes a data bandwidth of about 120,000 bits per second. This amounts to about one-third of one percent of our installed data input/output capability.
Nevertheless, we currently operate at a level such that this 0.3% flow generates and relates to work absorbing approximately 20% of our currently installed processing capacity.
REMOTE ACCESS CUSTOMER PROFILE
SDL’s remotely located customers represent practically every major branch of industry, business, science and technology. There are many common factors that reflect their data transmission needs.
While the majority of our customers are familiar with the operation of large computers and their access methods and economies, relatively few customers are conversant with common-carrier data transmission techniques and economies.
Most of our customers are familiar with programming, do their own programming and/or make use of pre-programmed generalized application support packages. Our remote access customers are normally ready to begin transmission to us within a few days of deciding to use our services. They are usually quite familiar with the operation of the installed terminal business equipment and require only minimal instruction in the art of organizing their processing work for maximum effectiveness.
SDL presently receives and transmits customer data through 62 data entry ports on the Model 85. These are distributed as follows:
22 165 bit per second (BPS) dial-up ports used for access by IBM 2741′s and other similar key-driven terminals
27 2000/2400 bit per second (bps) ports used to communicate with intermediate speed remote card reader/punch printers or small computers such as the IBM 1130 and the UNIVAC 9200
13 4800 bit per second (bps) ports used to communicate with specific customer terminal equipment in the intermediate speed range such as the IBM Model 20 or the UNIVAC 9300
At present, our Model 85 can have a total of about 150 ports having an aggregate data transmission capability of about 400,000 bits per second. Additional multiplexing equipment on order for fourth quarter 1971 delivery will more than double this figure to give 350 ports and an aggregate capability of nearly 2,000,000 bits per second.
For yet additional port capacity, small processors such as the IBM System/360 Model 25 can provide many more additional ports and be linked to the Model 85 via high-speed channels at rates of about 1,000,000 bits per second each.
A significant number of our customer/prospects are organizations now spending approximately $5,000 per month on their own computing hardware such as IBM Model 20′s or UNIVAC 9300′s. Our service prices, when swollen by terminal modem and toll charges are often only marginally advantageous to the small customer despite the normally enormous price/performance gains of our Model 85 over his own computer. A typical breakdown of our monthly charges to a customer in this workload range is:
Common Carrier Toll Charges or Lease $ 500
Two Modems (normally owned by 500 – 20%
the common carrier)
Terminal Equipment (reader/printer) 1,000 – 20%
SDL Model 85 Computing Services 3,000 – 60%
The productive use the customer gets from this type of system is what the Model 85 produces for him. The terminal is essentially non-productive and the expenditure of 20% on modems and common carrier toll charges is equally non-productive. Our desire is to be as efficient and inexpensive as possible in these two areas. Needless to say, it is the 20% for modems and common carriers charges that we are talking about tonight.
Once one of our customers is sold on using our service, he is ready to make use of this service immediately. One of our biggest problems has been in obtaining adequate delivery of lines and modems. We can stock-pile terminals if necessary. We could even purchase non-Bell modems although we frankly believe this is not our business. SDL has made a large investment in computer capacity and feel that having to use further capital to buy what is essentially communications equipment, should not be our responsibility.
If we cannot start service to a customer very shortly after he has signed an order, we lose considerable revenue. Further, the customer who has now decided to convert to our system has people who are used unproductively while waiting to start the conversion. The best we can do is provide an in-city delivery service to some location where we have our own terminal. Also, to be honest, the longer the time lag between the close of the sale and the start of the service, the more likely the customer is to alter his decision.
Yet, the typical delay is six to eight weeks to obtain a leased line.
Now, consider a typical situation. Knowing that there will be this considerable delay in getting a line installed, SDL eventually resorts to ordering a line in anticipation of getting an order. Now, one of two things may happen. The Bell delivers earlier than expected and arrives in the customer’s office without notification to SDL to install a dataset which the customer does not expect. This leads to very interesting situations. Or, the customer decides not to use our service and we have to cancel the line.
Now the Bell is critical of SDL because we cancelled a line on short notice.
In a word, by trying to correct the Bell’s problem of slow delivery, we have worsened the situation rather than made it better.
A recent embarrassing incident arose when, in anticipation of a requirement from Northern Electric in Montreal, we discovered that the only way we could adequately get good delivery was to install CN/CP service. This solution seemed rather unpalatable to Northern.
The expectations of the computer service industry may be too high but the delivery the Bell is able to provide on leased 2400/4800 baud- lines is not responsive to the industry’s needs.
Here again, those in the computer industry are intolerant of the error rates on the common carrier network. As an example of the difference in attitude I referred to earlier, we should bear in mind that those who have been working with computers expect virtually faultless operation. On the other hand, in voice communications, transmission is satisfactory if the person at the other end can reasonably understand what is said.
Let me give you an example of why noisy lines are unsatisfactory. Mr. W.V. Wolontis, Executive Director of the Data Systems Engineering Division at Bell Labs noted in the Bell Laboratories Record, August 1970 that “typically for the great majority of switching offices, Data-Phone service gives you less than one error in 100,000 bits on about 90% of the connections dialed”. If 100K bits equals approximately 12,500 bytes and the average print line is 120 characters this would mean that one could expect an error for about 100 lines printed. Naturally, this does not happen as the one error in 100,000 bits is an average over many dial-up calls and clearly a bad connection can generate hundreds of wrong bits thus biasing the average.
Also, the use of data compression and other standard techniques would really allow you to print many more than 100 lines before expecting an error. However, it is obvious that this rate is far too high for satisfactory data transmission.
This problem now ties into the delivery problem. You might have expected that we could use a dial-up facility prior to obtaining a leased line but dial-up tends to be highly unreliable and leased lines are on a long delivery schedule. Hence, the computer service industry is caught in the middle.
3. Type of Service
Last year Bell Canada announced Multicom. One executive of the Bell pointed to this with pride noting that this is a service available in Canada that was not available in the United States, and hence, Bell Canada was endeavouring to help the computer serv
ices industry by anticipating its needs.
The computer services industry may, however, seem somewhat ungrateful. I do not know how many Multicom customers you have in Canada at the moment but I suspect the number is very low. This is not entirely the fault of the service as we certainly expect to be using 50KB lines in the near future. However, the cost of the service is such that it rapidly becomes uneconomic relative to a Telpak line. My concern is not to criticize the Multicom service but rather to comment on the way it arrived on the scene. I suspect that it was the result of what someone in the Bell thought the computer services industry needed. I know that we in SDL were not asked or we could quickly have pointed out that the immediate need could likely have been solved much less expensively by something such as a Telpak- equivalent service for the high speed transmission of magnetic tapes during off-hours, when I suspect that the microwave network is lightly used, e.g. after prime viewing time.
My point is that the Bell has tended to work in an unnecessary state of isolation and only recently has started to enter into a real dialogue with the industry. I will come back to this point later.
4. Service Personnel
This is a touchy area as one can never fault the Bell for polite responsiveness. However, it has been our experience over many months that the sales and service representatives, however well intentioned, are not necessarily adequately trained for the rather demanding jobs in the computer services field. Again, you should bear in mind that those in the computer industry are used to dealing with highly trained professionals often with one or more university degrees.
Our experience in trying to get a price quotation on a simple service can drag over many months and often indicates either a lack of understanding of the problem or worse, an attitude that implies “if it isn’t in the book, it doesn’t exist”.
Once again, the demands of the computer industry may be unreasonable. For example, we may ask for 9600 baud service that can be split at our bidding into a single 4800 baud line and two 2400 baud lines. The answer from the Bell is almost certainly that this is totally impossible. This is often somewhat frustrating as we may know of cases in the United States where similar service has been provided and it is rare that Bell Canada knows that AT&T has a ready-made solution. (I will come back to this point in a moment.) Our request may have been unreasonable but so was the Bell’s answer. What we were really doing is posing a problem to the Bell and would hope for an answer such as ‘we cannot provide that exact service but we could provide a 4800 baud line and two 2400 baud lines during the day and provide a 15KB line after 6:00 p.m. Would this be acceptable?’
In effect, what we are looking for are people to work with us to solve problems in a frontier area and once the “it’s not in the book” attitude appears, then the computer industry concludes that the Bell again is not being responsive.
5. Foreign Service
The Trans-Canada Telephone System regularly places full- page ads noting that they alone have international expertise. In fact, they quote a number (toll free) that one should call to find out about service between Canada and other countries.
We do have extensive business in the United States and the service across-the-border resulted in this type of situation:
A 4800 baud line was ordered between New York and Ottawa on August 12, 1970. After a considerable number of phone calls, a reply was received on October 15 indicating that the leased line would be installed on December 28. It turned out that what AT&T Long Lines had done was to present the order to Bell Canada. Bell Canada then requested an RPQ for the engineering of the leased line and apparently received a reply on November 20. At this point, it is interesting to note that a simple 4800 baud line is still considered special engineering.
The earliest installation date was then nine weeks after receipt of the approved RPQ. This made the earliest installation date January 28, 1971 or close to six months after the original order.
This situation is typical rather than exceptional. Our next experience was with the installation of FX lines between New York, Boston and Ottawa.
Without detailing the experience, it is sufficient to note that minor problems involving AT&T and the Bell took an inordinate length of time to get solved, e.g. the lines would drop out without notice and what should have been a simple correction took several months to correct.
In a word, this is one area where the Bell should have a great advantage over CN/CP but the evidence is you are not taking advantage of this.
6. Cost of Service
Another aspect of Bell’s service that is particularly irritating to the computer services industry is the apparent reluctance of Bell to consider types of tariffs designed for the computer services industry.
In a Request for Price Quotation submitted to the Bell in September 1970, SDL took a first step in defining what the needs of the computer services industry really were. We made numerous suggestions as to how rates could be altered to make them more feasible for data transmission. We p4inted out that the almost linear charge distance relationship for dedicated inter-city data service was geographically restrictive. We understand, of course, that some front-end loading may be necessary because it presumably costs a certain amount to make a connection. However, once that connection is established, we doubt that it really costs Bell Canada much more to send data from Ottawa
to Vancouver than it would from Ottawa to Toronto. We recommended, therefore, a smoothly decreasing non- linear charge to distance relationship for dedicated inter-city data service.
We realize that we are not in the position to fully understand the cost structures behind the present charging arrangement. However, once again, this was an example of where we were trying to define our needs and putting forward suggestions to the Bell. The answer that came back was that such an arrangement was not presently possible. Here again, we were looking for the solution to a problem and got an out-of-the-book answer.
Another factor that irks the computer services industry is that we know that the cost of modems in the United States is about half the cost in Canada and the cost of long distance transmission service is similarly far more attractive. It is also irritating that we in Canada are unable to make use of this transcontinental American service as we are forced to use the more highly priced Canadian service.
In effect, it would appear to the computer services industry that Bell Canada is seriously standing in the way of any amount of east-west transmission and is a major factor in forcing Canadian computer services companies to seek business in the United States.
7. Advanced Design in Canada
The question of manufacturing modems in Canada brings up another interesting point. We recently suggested to Northern that they might consider manufacturing some types of modems in Canada and at the same time vastly improve the flexibility of the present modems. This is a subject in itself but if this could be undertaken in Canada, I am sure that the BNR could vastly improve present designs.
For example, one cannot readily put two datasets of supposedly the same type at either end of a line and expect them to work. Minor incompatibilities and on dial-up calls lack of simplified balancing and diagnostic procedures make this situation one which forces the computer services to use non-Bell modems.
This in itself often proves to be a problem because of the lingering reluctance of the Bell to interface responsively with non-Bell equipment.
I could go on with many other examples where it appears that the industry and the Bell are far apart in understanding each other’s needs. Let me now look at some fundamental problems as I see them.
One reason for the lack of communication between the computer services industry and the Bell can be laid on the doorstep of the Government. Without doubt the possibility that the common carriers might enter the computer services industry has made us view the Bell with suspicion. No doubt, we were reluctant to describe the needs of our industry to a potential competitor.
The Bell, on the other hand, appears to have taken the attitude that they know what is required and until recently did not seek out the opinion of the industry. When they did so, the result was a further cause for irritation. I am referring to the now famous Bell Questionnaire. This extensive questionnaire appeared to be a broad survey of the data communications needs of the computer industry. If it had stopped there, the questionnaire would have been most useful. Regrettably, the questionnaire went on to ask any number of questions that would clearly only have been asked by an organization about to enter the computer services industry. These questions included such items as:
– Are you happy with your present computer supplier and how do you rank the following suppliers?
– How much do you pay for maintenance on your computer system?
– What applications do you now have installed on computer service companies?
There is a more fundamental problem. Mr. Scrivener, the other day, mentioned to me that one of his major concerns was to ensure that the Bell always provided equal service to all users. This worthwhile aim is fine for supplying Contempora phones to a mass market but cannot provide a responsiveness to a developing field that is clearly anything but a mass market. The Bell must be prepared to work with individual companies to try new services. From time to time, this will lead to one company having a service that is not at that point, available to another company, but this is the only way in which the field can develop.
We are a very small industry and are not yet in the position to provide Bell with much revenue, but if we can believe AT&T’s own figures, this industry will supply a vast market for Bell Canada in the decades to come. If we are to realize this potential, I believe it is essential that the computer services industry be more realistic and more tolerant of the position of the Bell while at the same time, the Bell must be prepared to listen and react in a more flexible fashion.
I hope that SDL played some small part in breaking the ice by defining the needs of the company over the next few years in our RPQ of September. I am now very pleased to see the Bell reciprocating through a meeting such as that held tonight. I hope that this interaction does not end here but rather that this is just the start of a dialogue that will enable our two industries to work closely together to realize what I am sure will be a most rewarding experience in the years to come.