Transcript Tim O'Neil, Network Systems Engineer, Partners Telemedicine Center, March 20th, 1998

Audio Tape 1 of 1

C: Yesterday I was at BWH, I came in to see the CDC conference yesterday. Which was fascinating because it seemed like a real labor of coordination.

T: Did you see my little roadshow?

C: it was a beautiful.

T: It's timing. We're so new to this market we haven't had time to integrate all the spaces, so we kinda of bring it all with us. We have a consult before, couple days in advance , proof of concept, how it s all gonna work, kinda get a feel for what media they're gonna show. Its essentially a production screening, then we just show up with all the gear.

C: Well it went quite well, I missed the, something happened at the very beginning, when I walked in you were taliking to someone, and...

T: Oh, yeah the North Shore. See that's the interesting thing about this technology, things on occasion, things go wrong [laughter] The camera source that they were compressing their video source at, transmitting it over the network was black, so the cable. It turned out to be something as simple as the cable between the control and the camera itself had gone haywire and wasnt transmitting video, so we could hear them and the way a bridge works is the bridge is the central place that we all call into so that we can mix our video and audio, so if I speak for four seconds, my video pops up as your receive video, so when they spoke for four seconds, black. So everyone was like: "Did we lose 'em?" So what I wanted was at least a document camera, so that when they spoke it would at least show a little thing that said North Shore Medical Center so that we knew it was them. Its the nature of this technology.

C: How long have you been doing this videoconferencing stuff, were you trained in it?

T: Yeah, I'll give you the brief history. I've been in engineering for about 12 years, started out in the U.S. Navy, electricity, electronics, power plants, dish frequency regulators, complicated high technology electronics. I got out of the military, started a government job. Essentially automation of huge plants, electricity and pneumatics controls. Did that for about a year until I found a really cool electronics job, straight up electronics systems work, so I worked for about 5 years doing that, I have licenses in the state of Massachusetts, two electrical licenses: one as a systems technician one as a systems contactor. I'm licensed to put in fire security accesss controls, CCTV, all those systems in massachusetts. Only about 2000 of us right now, in the state. While working as an electronic technician, I went to school nights to supplement my intense Naval training and I got an honors degree in electronic technology, associates degree, core curricullum microprocessor based applications. After five years of a degree, all the technology licences and training, I kept bumping into a technology barrier in the systems I was using, no industry systems, everything was going to the LAN, and nobody could do it right. SO I kept seeing what I was doing and I saw the correlation to information systems. And about three years ago I jumped ship to the information systems department of the government, federal, the VA. And at first, one of the first projects I got to work on was pretty neat, I saw a videoconferencing unit. And I said: I understand how all this works you know this is kinda the same as what I used do but now its a little different and they had been workin this roject for a year and a half and hadn't gotten it up. The stuff was in boxes. They spent a quarter of a mllion dollars and everything was in boxes. Becaue the telecommunications industry was having some much trouble at the time trying to supply fractional T-1 and the bandwidth so ...........So they were wallowing for a year and a half, and as you know information technology is a huge whole right now and if I need 15 people I can usually find one. SO they were wallowing and I saw it and started tinkering with it and said this is cool. So about a month and a half later I got the first one up and running and within 3 months the whole system was up and running, that was a wide-area network based on Vtel technology . We had our own bridge to connect with Worcester; we had a clinic in Worcester, brockton and west Roxbury VA medical centers and hooked them all together. So that's how I got my start running the video portion, I mean I'd run cable TV networks before, same kind of technology, no compression, broadband network. And I moved my way over to the bits and the bytes and the digital compression, went to scholl at night for about 7 months t hours a night two nights a week, eight hours every other week to get Windows NT administrator certified. So thats the Industry standard right now within Wide Area Networks enterprise, "Are ya an MCSE, or have you at least been trainied as a Microsoft Certified Systems Engineer? Do you have a Microsoft certified professional number?" And I do [laughter]. So that's the whole thing, I spent about eight, nine grand and all my time and effort to get that kind of training, and it kinda... Right now, we're still not there. Window NT can't support video conferencing like we can now, um. So while I was still at the VA, another project came up, part of a 26 node video conferencing system for the Veterens Integrated Services Network One (VISN1) and thats 26 medical centers stretching from Maine to Maryland. There are a number of, If you read more about telemedicine you'll find that the VISN's and the VA are dong some really kickin' telemedicine applications. We were a little behind, more administrative, grand round, remote ed. at the time. High-end administrative, that 26 node ssystemn was put in so that the guy that was in charge of the 26 hospitals could speak to the guys that were in charge of the 26 hospitals. SO that's why that was up. So after about two years working there, I came here. Four days it will be one year.

C: So you've been here just since they started the video-conferencing here?

T: Yeah, same deal here, I got here and they had put all this network in and hadn't been able to get it up and running [laughter]. SO a week before I started to be officially paid here, the vendor that helped me get me the job here, same vendor that I worked with at my first job, who saw that I take a screwed-up project and make it work, so this was, the ISDN line ordering had been improperly provisioned and all the technology that was supposed to go in wasn't working, essentially the modems that we had then, they're inverse multi-plexers actuallym they're the means to get out there, to connect points- they weren't the right ones. The lines we had bought weren't provisioned properly, so they wouldn't call, I mean we did the first telephone call from the Prudential Center to this side of town on ISDN, and the reason we found that out was that the switch over there and the switch over here hadn't had their translation tables built, so that when you tried to dial a 248 exchange from a 262 exchange, it didn't know what to do about that. It really didnt' know... It couldn't get there from here! So we did the first ISDN call out of Beijing China, was another one of our firsts here. So as you can see the technology's kind of new, its been around in this incarnation for about 5 years, and ISDN's really making it prolific now.

C: Now the one that I saw, that was done with Tel Aviv, that was done over ISDN?

T: ISDN Lines, All of them, pretty much worldwide communications are done via ISDN lines. And that's a cost versus quality issue, our compression ratio is about....

C: Now just a technical question, since I'm not so clear on this, ISDN can be run over any kind of cable or fiber, or does it have to be fiber optic?

T: Right now, ISDN is twisted pair, its telephone wire. Its a turbo telephone wire with a cleaner run from here to the central office. But, SO lemme give you kind of a primer on what ISDN technology is. Um. Standard telephone line is capable of transmitting digital bits, you convert the analog into digital bits and start streaming bits at about 33.6 kilobits, 33thousand bits per second, pretty good, it can handle that. They'res higher than than 56K modems andstuff like that. They're trying to get the 56K over the same twisted pair that was feeding the 33.6 K line, now depending on loop-length, the distance between your house and the central office, or closest switch node, if they've gone through and removed the impeding resistors that they put in there during the building out of telephone systems, -there's impeding resistors that stop crosstalk, ok, kept things at a voltage that was constant pretty much- those impede speed. So they've had to go through and clean them up and gone to time-division multiplex circuits and central offices and now you can maybe to 56 on a plain old telephone or POTS line. Lets say, 33.6 to 56K thats what you got there. ISDN came along, and that was more of a dedicated line, these aren't retrofitted old lines, these are "Let's run some new lines." There could be fiber trunking somewhere in their network, could be ATM trunking, terrabit networking, but primarily its essentially a twisted pair running from a switch to your house, think of it virtually that way. And the ability to transmit on one ISDN line 128K, so you go from 33.6 to 128. Then there's a process called nverse multiplexing, what inverse multiplexing allows you to do is bond ISDN lines together, so what we do is we take one 128Kilobit ISDN line, you look inside that 128K line and you find two 64K bearer channel and one 16K data channel (mainenance). So when you provision these lines and you get into the backdoor and you can see what's happening between the communications you're gonna see your D channel between you your inverse multiplexer and a switch and you know that you've got a good line of communication, this is hardware... So I wanna establish a call: "Hey howya doin, I'm an imux, I have this spid (spid is a service providor identification number, its my phone number, I get one per b channel so actually every ISDN line has two phone numbers)." Um so that's an ISDN line its two 64K channels, I wanna use inverse multiplexing in bonding mode one (bonding mode one is a standard in the industry that allows me to glue 'em together) So now my software- I don't know if you understand the OSI- the seven layer from application down to physical layer. That application looks out at the physical layer and what it sees is not 6 64 K channels that it has to get all stuff goin to, but one big 384kilobit channel. Now when you talk about 384 Kilobit channels, here's the price versus quality issue that you have to address. When your doing 384 kilobits worth of videoconferencing, you're taking at a compression ratio of 240:1. Hmm. You got one representing 240, that's not so good. But it's pretty good compared to 700:1, for one ISDN line, so 700:1, what do I spend? $33 bucks a month for a phone line for an ISDN line, for $111 bucks a month I can get 240:1, and near, near 30 frames a second at full screen 352 by 288. The bulk of installations on this planet are 384K because for the price versus the compression its good enough for talking heads, some collaboration, still transfers of images, maybe not real-time intesive medical images, like endoscopic images or echocardio images. Echocardiograph that's a 10 meg signal, 10 meg and 384K is a substantial exponential difference. Um but when you go to higher end, you can go to a T-1 videoconferencing. Videconferencing is quite capable of going to T-1 speeds and that compression ratio is 60:1. But now your not talke 100 dollars a month, if you sign a multi-year, your talking $500 dollars a month, for a five year contract, but you can do just about any image transfer, excluding echocardiographs.

C: Now the Tel Aviv one that we saw what was that?

T: 384. And they actually played an echo, and it was visible enough for them to use it. I was surprised they liked the echo at 384, and these aren't theys these are people who should know, that's why they were calling from Tel Aviv. [laughter] Dr. Cohen, I mean the guy's world famous. And Yesterday we did a demo for the endoscopy people and they didn't like theirs at 384, much smaller signal, much lower bandwidth, but they didn't like the pixellation. So its kind of funny, one that you think would work better at a lower band width, but there they want to show the world their endoscopy and they don't want pixellation in their image, and so they want the higher bandwidth. That's pretty much a technology {garble}

C: And that's constrained by where the lines are laid?

T: Well, you can get T-1 in most metropolitan areas and You can get ISDN pretty much throughout the whole bell atlantic operating company from florida to Maine.

C: What about ATM

T: Now ATM's a different story, ATM essentially the backbone that most carriers use to carry their phone calls. So if I'm gonna do an ATM call I dial into a central switch and then I wanna do a hop, they call it a POP really, a poin of presence, to go long distance, well when I get on to that backbone to go long distance I'm probably getting onto a terrabit ATM networking backbone. Um I wanna go to Russia, I'm taking the terrabit ATM backbone to California, and then their gonna go terrebit ATM uplink and then from one bird to another bird (as you see on my chart here) that's ATM goin from. So from here to here is an ATM based network, from here to here and here to here and here to here to here to here [pointing to wall map of satallite network] is ATM, but I'm only really using ISDN, all that ATM stuff belongs to the carriers. Cause their doing everybody's ATM, they have huge backbones with fiber running ATM throughout the terrestrial network, that's where ATM comes in. We're looking at ATM because, well, bandwidth primarily. Right now I have, my network looks like this [draws on whiteboard]: You've got a Local area that has a link to, this cloud represents the wide area, so you have a bunch of PC's, on your network and they'd all go, if they wanna make a long distance call, they go out to the cloud. Well what's kind of funny about the way we have to do video conferencing right now, is everyone, rather than having a network internal, we've got a node that goes to the wide area, call that the H.320 cloud, cause thats the standard that do it by. So if I wanna call here, here's the MGH, and here's BWH, This is ISDN right here. and so is this. This here, who knows? Could be ATM

C: Its up to the service providers what they do?

T: Right, its their network, What the cloud represents is my ability to dial anyone in the world, so I can be connected to this cloud and this could be the Brigham and Mass General or this could be The Brigham and Tel Aviv, Okay. Whatever this is is the variable I can't control. Now When we talk about ATM we're talkin' about doing something a little different. Where, we're trying to economize our bandwidth, right now I'm paying for this link, this iSDN link is the problem: I have to by 384 kilobytes of bandwidth for every Unit I own, 60 of them, okay? I'm not using them all the time, but I have to have them, in case I should want to use one. SO if 60 times.. just generically, that's like $6000 per month 60 times a hundred, plus usage. Well for $6000 a month I could buy a heck of a lot of network, okay. I mean think of it, $800 a month for a one year contract for T-1, and that gives me not 384K but 1.5megs. So what we're lookin at doing is something like this: Here at the MGH we want to build a private network based on, oh well, I don't know if you'll catch this part. This here would be an ATM switch. and from the ATM switch, this wouldn't be ISDN, this would be T-1. So know within my network- right now I can talk from one unit to another unit within this local network- I can call them at 384 Kilobit. Talk about ATM, I have the ability to put a 25 megabit line to the desktop. Now my codec can't really support that [laughter] but my codec can support 1.5 megabit. So now I can do 1.5 megabit anywhere internally, so if I'm a radiologist and I've got somethin I need a second opinion on, and I want to talk to my second opinion over at CNY, well if he's within this private network, I can get to him. Odds are there'll be a wide-area link to H.320 and then we'll have another star over at CNY, and that'll go out and that'll be T-1, and then at BWH you'll have another hub, and then some other guys so what will actually happen andwatch here's the economy: I wanna do a video conference call across town at 384. Well I've got T-1 here and I've got t-1 here so this guy here can dial here and now I don't pay toll charges on that at $6.20 an hour is what it costs me now, I have lines already in place, so an individual call isn't my concern, my total consumed bandwidth for the network is my concern now. So for like $2000 a month I can do 384 calls all day long, it doesn't bother me, rather than $6000 a month. Now what gets really coll is I could have multiple T-1's this could be three T-1's okay and this over here could be 3 T-1's so I've got 4.5megabits worth of bandwidth. So I can do: Brigham, has got this guy in endoscopy and the guys on the table and you know where they're at, okay, he doesn't wanna come back, something weird happens and maybe they need a second opinion. Or say they have someone on line, so brigham calls, and so here's the unit, network now looks like this. This guy here has a patient on the table, here's Mr. Smith, okay, I've got a video camera shooting Mr. Smith, plugged into a unit. SO now this guy here is going through here and here's Dr. Jones, watching Mr. Smith at 1.5 megabits. The economy is there because I'm I already own it and I'm not paying any more for this one, and when I'm done I'm gonna do 6 or 7 more videoconference calls okay, over that same T-1 pipe that Im using for this one endoscopic consult.

C: And could you do it faster at 1.5 if you had different equipment?

T: Yes, actually I could, actually I could. There are three standards out there: there's h.310 which is fractional ATM, or MPEG2 compression, we're not there yet. It's close, it works, its in product, its delivered, people are using it, we're not using it yet, cause we really haven't had... we've been talking heads remote-ed. We're doing a research project now for MPEG2 over ATM its gonna go from North Shore to...

C: The Echo project?

T: The echo project, that's gonna be a 45 megabit connection, its hugely expensive, see these links here: from here to here, from here to here, that's $5000 and that's $5000, a month, $10,000 dollars a month, for an ATM connection.

C: So the one from North shore would add another $5000 and would work the same way s these two?

T: Right whereas T-1 is about $800 a month for a T-1 line, at 1.5, but for a T-3, which is 23 T-1 lines, your talking $5000 dollars a month, so its gonna cost us $10000 a month just to get one echo from the north shore to the Mass General.

C: T-3 is that the same as ATM?

T: T-3 is a pipe that they use in the industry that's 45 megabits worth of bandwidth, so we can run 45 megabits worth of bandwidth. SO we choose to use a protocal stack of H.32.. a protocol stack of ATM, rather than one of TCP/IP. So you got TCP/IP, you've got ATM, you've got ethernet , okay. Cause ATM is a protocol, but its also a topology.

C: Right. [yeah, whatever, Chris]

T: T-3 refer's to the size, or the quantity of DS0 channels. ISDN has two DS0 channels, DS0 is a 64 K pipe. SO one ISDN line has two DS0 channels, one T-1 line has twentty-three DS0 channels, and one T-3 has twenty-three T-1's. Or twenty four really, depends how they do it for parity check, different service providers do different things. But that's where ATM starts to impact us, and if you think of it in a telemedical environment, ATM, and this is my opinion, .... but you know what, you like to read don't you? Here's another interesting guy, Joesph Gatto, he's a molecular biologist, who also knows a bit about ATM. [hands me document]. He discusses telemedicine. I haven't read in full the whole document , but I've read half the document and as far as I've gotten, the guy has a firm grip on it. And he, his push here is ATM, and if you think of the politics in the medical environment, you have IS, they're incharge of all the bandwidth, but radiology wants to use it all, and we want to use some more. If we went to an ATM environment, the way its built it would allow for better economy, so we could have one pipe we all share, based on the quality of service we institute in our applications. SO if I want to do an endoscopic call from one side of town to the other, I just need 1.5 megabits, I might have 45 megabit out therem but I don't need it. And when I'm done with my 1.5 megabit, it collapses, so if someone needs 45 for a full burst, they get it, so the hospital saves in the end. Right now we've got little factions with their own bandwidth, so what would happen if we said everyone give me your bandwidth, how much bandwidth do you have? They're gonna do some math and thy're gonna say: "My, God we're going broke!" [laughter] They really would probably be frightened at the amount of bandwidth they're spending money on, bt we could economize it with the ATM network. That document gives you a real good primer on ATM technology in a telemedical environment including PACS and medical records and looking at a medical record and clicking and linking to an endoscopic image, pulling it in, or retrieving an echocardiogram off of a PACS server.

C: Now is it easy to switch between ATM and TCP/IP, because I mean just thinking of the politics of having everything be web-based or potentially web-based

T: Well with in ATM they have something called LANE1, Local Area Network emulation, and as a said earlier, (I like to draw)... Right now the OSI seven layer, there's seven layers, and the ATM which is one through three its a simpler stack. I'm gonna give you a walk through of what we're doing with ATM. Right now I have an OSI stack. Seven deep. Windows 95, it's IP, the problem is you can't really port video over IP, it's a quality of service issue, routers have latency , skip, jitter, you know, you've seen streaming media off the internet, "Hi gobbadoood ddididididid," it's not interactive, it's one stream coming at you and it can't get there, never mind two way interactivity. So the way we've resolved that is that, remember I said there are three standards, these are the two transports or protocol stacks, and then we're doing H.320 videoconferencing, so it chews all this stuff up and sends it out ISDN. So this goes to ISDN right now, its called a cicuit switch. So I establish my ISDN call and all the switchs go dat-dat-dat-dat-dat: now thy're all looking at you and we have a guaranteed circuit goes from my application where I say "Hi how are you Chris?" down through the protocol stack out over the ISDN, over to you and back up your stack and out the application you're gonna hear "Hi how are you Chris?" that's how it works. Well we wanna go to ATM, but we're still doin H.320, tell me if I lose you? We've got H.310, we've got H.320, we've got H.323 okay this ATM, this is essentially WAN based, T-1 or ISDN, this here is IP based. Okay? Now I wanna use ATM, I have H.320 legacy systems, so how do I get from hither to yon? Well, manufacturer out there right now makes a thing called a MOS [draws]. Multimedia Operating systems, and what it does, it's is essentially a stack converter, so I can use my legacy equipment and get the benefits of ATM's Transport without routers. Okay? I don't know if you understand TCP/IP, but I have a huge address and I wanna go from here to there. What you'll read in here, he does something interesting, he talkes about the way it used to be is that eighty percent of the traffic stayed in your subnet, so "hi how you doing Chris?" "hey here's some email, here want some email?" "Hey what about Joe, oh Joe's in California this week" so I send it to ten people in my group and one guy Joe, so 89% or 85% of my traffic stays in the subnet and 5% goes off. Why would I need to videoconference with a guy in the next office? I don't. Why would I need to multimedia collaborate with a guy whos from here to there. I don't. So what do I need to do if I'm multimedia collaborating? 95% of my traffic is going out now, and 5% is staying in. Okay so its a flip. so This stack and this standardard and even IP were built under that premise. Problem is is we're multimedia wide ares. And for today, I wanna get from my legacy OSI seven-layer stack to my new three layer ATM stack. I wanna get the bandwidth, the no-latency I don't want routers in my way, I wanna economize by buying one fat pipe and letting everybody use it at different altering times, its like time-division multiplexing where its not "Hold on a second and let Mr. Smith go by its like "Hey Mr. smith, Hey Mr. Jones hey Mr. dt-dt-dt-dt-dt" multiplexing they all stack 'em. With an ATM, you come up, you want service, it says "I know you," because when I built the network, I told'em who he was and what quality of service he needed, okay, and this is the endoscopic unit, he's gonna need T-1 speeds, so its gonna start, he's gonna get a guaranteed path, anybody else comng on later will be accomodated cause you can scale an ATM. You buy it in a constant bit rate, like, so here we go [draws again]: Legacy, we wanna get to here, The way we have to do it now is a MOS. The reaseon is because we're using this, this is legacy. Now, what are we doing with the echo project? Oohh, we're doing this. Remove this, we go straight ATM. Now the compression that we're using H.320 is, um, yeah, we're gonna go from. oh this H320, I don't have the Mpeg compression. So just think of 261 at T1 is 60:1. MPEG compression can't be done at T-1 speed, there's not enough bandwidth. But its almost a two for one, Just 50% of the image. Not 1/700th of the image, okay, at 128. But so 5-% of the image, It's so clear, You're not even missing anything to the naked eye. So, how far away from your original question did I get.

C: I forgot what my original question was.

T: ATM

C: Oh and TCP/IP, so in the future the potential to move to a TCP/IP if needed?

T: The only way you're gonna get there, I'll tell ya, this standard here h. 323 is based on, okay. ATM, 310 amd 320 is called circuit switched, thats the dt-dt-dt-dt-dt-dt I switch my circuits to you. IP is shared resource, I send it in packets, so one packet goes this way, one packet goes that way, Albequerque, Idaho, New Mexico, California, they all end up in Washington. Okay. Is there a time, a guaranteed time slice when they all get there? Well they'd like to think they could get there in a one second time slice, but what guarantees that? Nothing. Within the protocol for TCP/IP and ethernet there's no standard for quality of service.

C: Kind of completely democratic in a way, right?

T: Right, who got there first. Or who built an optimal system that it didn't care if there were collisions it could work without that information in the network. DO you have a basic idea of how ethernet networks work?

C: Yeah

T: With carrier sets multiplied with collision detection: A packet sticks its head out in the network and hopefully doesn't get his head knocked off. If he does, he pulls back and he waits a second and then he looks ou it looks good and he starts sending. If he collides down on the other end he's just vaporized. And within the packed is a header, where he's going, where he's coming from, what packet string he is in the sequence. So when he gets to the other end, within this seven layers, the transport layer three up, four up from seven, he counts 'em, I've got packet string one, two, three... seven thousand, seven thusand one's missing, so he sends the information back down bumpin into everyone, and says "Hey I missed that other one!" so itsays, okay, I'll resend. So from here to albequerque to idaho back upto Washington he sends that new one. Hopefully the packet that collided, was asked for a resend, makes it there in time, what are the odds? Not very good. And according to the gentlemen I've spoken to most recently, the people that write this H.323 standard, is that hopefully the new IPv6 and the new H.323 version 2 which is coming out, may better accomodate quality of service issues. There are vendors in the industry, gigabit ethernet router vendors, there's Cisco with RSVP, which supposedly on your header will say: "I'm a video conferencing Packet, oh! we'll reserver you some bandwidth" Somehow, on the internet, can you imagine this? My router talks to your router and negoriates a path for this. C'mon, you're gonna run into someone who's router's from 1950.

C: But for anything that's asynchronous, you don't really need...

T: Real time. Its for real time is the issue. You need Quality of Service. So if you look at IP and you look at ATM, that's it: right there. Quality of service is non-existent in IP, security holes are there. ATM, you know, is guaranteed quality of service, and the security's there because its not going through everybody's stuff, it has a guaranteed path, it knows where its going, it doesn't have to negotiate a path through routers, its a switch.

C: But you sacrifice being able to reach anyone

T: No, you pay for bandwidth, You pay for the bandwidth. Remember, as I said, when you looked at the terrestrial network up there, its ATM up there too. His paper will be interesting, because he mentions "Why aren't hospitals going to ATM, because the fabric that's out there is going to ATM"...

C: When you were doing the Tel Aviv call, you mentioned that depending on who it went through, you got charged different things. Which service provider or whatever routed the call?

T: Right, dependnig on... I have ATT as my long distance carrier in the united states, I know that. ATTT has contracts around the world as well. So when I call from here to California my ISDN line is built to pick, they call it pick, to ATT, the minute that I reach the boundaries of Bell Atlantic and I have to go to a POP to get the backbone of a Wide Area, I've already negotiated a contract with ATT, so a local call will cost me about $6.12/hr, a call from here to the boundaries of Bell Atlantic, you know, outer Massachusetts will cost about $22/hr, this is for a 384K call, and to reach the nation, it costs about $62 an hour, to reach California. These are agreed upon, we have a contract with them that saus "We're Partners healthcare, we're the number nine account for Bell Alantic" or we were for Nynex, I don't know about Bell Atlantic. So we have power there in our size, and so we negotiated both with Bell Atlantic for a low cost ISDN and with ATT for Netone services, they have a corporate pricing strategy. Now here's the problem, I wanna call Israel, and Israel doesn't have Bell Atalntic, they have Israel Telecom and they might not have ATT, they might have hypha [garble. I'm guessing.. It could be that, and what's their pricing structure? I don't know, could change tomorrow. That's the variaility in long distance calling, you might not get the best price. You get to russia, and here's a funny one, we tried to call Russia and we can't. Turns out they have a private telecommunications network, that some westerner probably went over there and said, "Hey you got no phones, I'll throw some phones in, Bell Atlantic did a good job, or MA Bell did a good job, what's in a network?" Turns out he doesn't have a contract that will allow him a gateway to international calling. So everybody in the country can call from his network. He built his own network and he wired to people's houses, or businesses, but when they wanna call us, he has a contract where he can route his customers out, but nobody coming in has a contract to get into his network. So we couldn't call!

C: That's so weird.

T: That's so the state of the art, right now. Its that variable in cost, you might not even be able to get the connection because there may not be a contract between the carrier we're using to get out and the guy he's using to get in.

C: SO does Vtel keep tabs on that kind of thing?

T: No its a telecommunications issue. Vtel is a manufacturer of videoconferencing hardware.

C: But do you think it would be to their benefit to say " Oh we shouldn't sell you a VC unit because you wouldn't be able to use it?" Wouldn't that be a service nightmare.

T: Oh no no no. Vtel's in the business of selling components and getting them up and running. I mean they service their equipment, but divided right now, here, telecommunications is a complete different animal.

C: Right, but if they sold something to someone within that network wouldn't they be in...

T: All they're liable for is we can do a call on your network. And if somebody else is on the network, and Vtel doesn't specify the network, they specify the way that you need to use whatever network. So when you show up here, you're supposed to have your network already, Vtel isn't a network provider or a network integrator, what they do is they try to integrate their equipment with your network. So they have an inverse multiplexer that is generic for world wide use, the Ascend VSX and they plug it into the ISDN lines, and they make sure it taks to a switch, which could be... switches are all over the map, you know DMS100s Norther Telecom, meridian switches, 5ESSs, there could be any number of switches out there. SO Vtel's not in the market for... I'll give you a small snippet of an experience I had, My first vtel experience. Told you it was a year and a half up and running. Because that project we bought vtel equipment and we asked Vtel to install it. Vtel couldn;t tell Nynex, ATT paradyne, they became Nynex during the project and they were using lucent technology and equipment at the time, and they had to go through the government's backbone for long distance. SO THAT was the year and a half, trying to get the network up and running and trying to get the nodes up and running, you had to wrestle with that then wrestle with the technology to get it up and running. My next project was pictureTel equipment, but the vendor was lucent technology, They brought the network before they even brought the nodes. Guy shows up and says "I'm here to do the install" I'm like: "On what?" " On your inverse multiplexer." I was shocked, that's how you're supposed to do it. Guy brings the network, he puts in the inverse multiplexer, he does a test to here, there and everywhere. The network works, the same day! They'd ordered it in advance. Then the picturetel equipment shows up, it works, and they plugged it into the network and it worked!! Okay, they did 26 in half the time it took us to do three. So that's the difference, the difference is who brought the network, well the network provider brough the network and then the vendor plugged into it. Where, same deal here, but without the influence on the vendor, I have had to strongarm bell atlantic since the day I got here.

C: That's why I'm wonderning about this, it shifts the risk to

T: It's my responsibility, I'm the customer of Bell Atlantic, not Vtel.

C: So Vtel kind of gets out of the loop..

T: They try to, "It's not us." And we always say, well you sold us the network and its not working with the network, and so they have to prove that their equipment is not the problem. And the problem with this state of technology is the odds of you getting two qualified ISDN on a phone call at the same time, very slim. Really. I mean you call their help desk and their like "Whatdya mean ISDN?" They don't even know they vend ISDN. You get a guy show up to do an install, you see he's got his tool belt on, he's got a head set, and he says "I'm here to do the phone line." I look at him and say "Its ISDN line, and they go like this "Hey that's somebody else, I do phone lines, I don't know what your talkin about , I gotta call my supervisor." He walks away. Two weeks later I get a guy to do my ISDN install, and he doesn't do it correctly. And I spend a week and a half trying to prove to Vtel that the networks good and their units bad. This is the difference, the small point that I was trying to make and ran on about is that if you network comes first or the vendor is a network provider the odds of the network working are a lot better, and if they have the agreement wiith the vendor whos supplying the videoconferencing, your project will probably go a lot smoother. So the bulk of my work here is wrestling between those two vendors, or it was to begin, we couldn't get calls from one side of town to the other and it was all thier fault, and they didn't know it and I had to slowly educate them about how it works and you know get some guys on the phone who I recognized their name, and they knew I was right: "oh it's that guy who always right, whatever he says just send the guy out there!" Once they figured out that I knew what was wrong more than they did, then we started to get a lot more cooperation out of them. It wasn't me callin and screamin at them you idiots can' get it done. It was me saying, "Look, your provisioning on this line at this switch is out of translation, could you go correct that issue." "Oh, he's got the line number, he's got the spids he knows what the problem is, we can go right to it and fix it."

C: Well their lucky to have someone who knows all that.

T: We're lucky. Because otherwise we wouldn't be doing video conferencing. It helps to know what your doing with this stuff.

C: So what do you see in the future, Partners Telelemedicine, or if you leave to go on to something else?

T: well were hopefully gonna hire some more people who can manage ISDN, um our future is the ATM network, hopefully, its just economy. The ROI, I can spend $250,000 this year on ISDN line phone calls or I can spend $250,000 on ATM equipment and for the next three years pay 20% of what I paid in ISDN on t-1 connectivity, so the economy is in the ROI. Utility costs versus return on investment, I can get a Return on my investment based on savings in utility costs. So we're gonna go to that topology, whic is a more stable topology, so that in-house people will manage the network, except for the wide area links, that's still a T-1 its ike ISDN, but people are a lot better at T-1 serveice than they are at ISDN service, so and here's the thing, if I lose an ISDN line right now, I don't know about and neither do they, if I lose a T-1...

Last Modified 11-Sep-99 9:31 PM ckelty@mit.edu

Go Back to the Start