The Fix is In

Last Christmas, thousands of California students got new computers and modem connections to the Internet. Great news, but Pacific Bell officials were worried. They were afraid of what would happen when all those kids plugged in their new toys and logged on to the Internet.

Network engineers held their collective breath Christmas Day, and then-nothing happened. The voice network, through which analog modem calls travel on the way to the Internet, held up.

On Jan. 6, the students went back to school and the engineers breathed easier. Then, at about 3 p.m., the new Web surfers got home. A vast number turned on their computers, logged in and hit the return key-at the same time.

A central office in the East Bay area of San Francisco immediately browned out. Businesses in Contra Costa County lost their dial tone, customers trying to dial in heard fast busy signals and Pacific Bell customers resorted to using their cellular phones to lodge their complaints.

The Bell regional holding companies have been dealing with problems like the California one for months as the Internet boom continues and more people work at home. A similar brown down happened during the Blizzard of '96 in and around Washington, D.C. Government workers couldn't travel to work, so they fired up their laptops and logged on to corporate networks or the Internet. Bell Atlantic started to have scattered overload problems with switches.

It's the mass flushing during Superbowl halftime syndrome, with one variable-timing. But network engineers don't know exactly when the flushing-logging on in this case-is going to happen (Figure 1).

During the last year, the RHCs have complained, cajoled and criticized Internet service providers (ISPs) and especially the government, which regulates access to the local loop. ISPs have a special enhanced service provider status and are exempt from paying the access fees interexchange carriers pay.

ISPs garner the heavyweight support of AT&T, MCI and Sprint because these IXCs are tired of paying the couple of cents per minute charges. The Federal Communications Commission is expected to decide soon what to do with access charges for both ISPs and IXCs, although the commissioners seem to be leaning toward keeping the exemption for enhanced service providers, a category under which ISPs fall.

So the problem is often misstated. People worry that the voice network may collapse under the weight of Internet traffic, or that the Internet itself will bog down and crash.

That's not it at all. The RHCs can and are fixing the problems by augmenting switches or re-routing data traffic off the voice network. RHC engineers freely admit that they are more than capable and have the tools to do it.

The problem, as RHC chief executive officers and business executives see it, is that they aren't getting compensated for those network upgrades caused by Internet traffic, especially at ISP-connected COs.

The real problem is not the ability to fix it-it is the question of who is going to pay for it.

All the answers With every problem, emergency or not, manufacturers will come up with a host of gadgets, bandages and products to sell. The Internet traffic problem is no different. A healthy handful of vendors has rolled out re-routing solutions. Some of the RHCs are considering them and others are already using them.

SBC Communications has implemented a jointly developed solution with Northern Telecom in 12 cities and for about 20 ISPs. Ameritech is using Lucent Technologies' Access Interface Units, although it re-provisions switch load as opposed to re-routing traffic from the voice to the data network like the SBC/Nortel solution.

Following are some of the vendors and their solutions: * Lucent Technologies' Access Interface Unit and Access Gateway uses software to identify long-hold time users, often at the ISP-connected COs. The software enables carriers to move traffic to a more cost-effective part of the switch or to the AIU. The AIU is a high-capacity line unit that provides a virtual non-blocking path through the switch.

The AIU can only be used on Lucent digital switches. The Access Gateways 1000 and 2000 are the re-routing boxes that convert data traffic on the voice network so it can be sent over a data network, and are usually used at the originating switch.

* Premisys' Interlude can be used in two environments-either for direct dial and ISDN traffic terminating on the local switch or for a remote hub environment (Figure 2). On the direct dial side, the Interlude siphons off data or ISDN traffic from the CO switch, routing it to an appropriate data network.

For a remote hub application, the Interlude concentrates and mediates traffic in front of the switch, then sends appropriate traffic to either the voice switch or the data network.

* DSC Communications' Intelligent Internet Solution is a combination of software and hardware solutions already on the market (Figure 3). The solution uses the SS7 network through the MegaHub STP to redirect traffic intended for on-line services and ISPs.

DSC's Advanced Intelligent Network detects data calls and then signals the Litespan or DEXC cross-connect to direct the traffic to direct data links. A major local exchange carrier is testing the system, according to DSC.

* Nortel's Internet Thruway uses several pieces of Nortel equipment, beginning with the AccessNode, a digital loop carrier, that looks at the number dialed and intelligently routes the call to the voice or data network.

The voice calls go from the DLC over a TR-303 link to the CO. But a data-defined call is sent over a digital trunk line to a Rapport dial-up switch and then onto the packet-switched data network. Nortel plans to upgrade the solution this year by adding support for higher access speeds, always-on connections and data over voice service.

* Telco Systems' Intelligent Call Routing features an access server that takes incoming calls from subscribers and identifies them to send on to the voice switched CO or bypass them to a packet network (Figure 4). That maintains direct packet access for the data.

* Siemens Stromberg-Carlson, the third-largest U.S. core voice switch vendor, after Lucent and Nortel, does not have a proposed solution yet, but hopes to soon. "We have the capability to deploy an overlay network to take traffic off the [public network] off the trunk side, not the line side, using data tandem switches," says a spokesman.

What's an RHC to do? SBC was not only the first RHC to try an Internet bypass solution, but is now the first champion of the benefits. Officials pepper conversations about their Internet/Intranet Transport Service with phrases like "win-win" and "good for everyone."

SBC is using IITS to beef up its own network and to sell the service to ISPs. So far, SBC has put the boxes in its network in Dallas, Fort Worth, Houston and San Antonio, Texas; St. Louis and Kansas City, Mo.; Topeka and Wichita, Kan.; Oklahoma City and Tulsa, Okla.; and Little Rock, Ark.

The carrier also has signed up about 20 ISPs and is negotiating with 60 more for IITS.

SBC can't quantify the improvement, but the feedback from both customers and ISPs has been overwhelmingly positive, says Ed Reisner, regional manager for technical and product development.

"We're trying to talk to as many people as we can, but I've gotten calls from ISPs saying, 'When are you going to call me? I'm really interested,'" he says.

For the ISPs, quality and better throughput are concerns, but another significant plus is being able to pass off modem and network management to SBC. ISPs can then grow more quickly by adding ports from SBC and can spend more time worrying about content instead of infrastructure, Reisner says.

"I think [RHCs] need to sit down and do their own cost comparisons, but to me, it becomes very obvious what the positive economics are," he says. "I don't know of any RHCs that will argue with the fact that data traffic does not belong on the voice network. If you look at our model, you'll see it's good for subscribers, good for customers and good for us.

Pacific Telesis is looking at many models and has tentatively called its future service FasTrack Local Access Gateway. Pacific Telesis, and especially its Pacific Bell subsidiary, is the RHC with the most exacerbated problem-some 30% to 40% of all Internet traffic originates or terminates in California, and 20% of the world's Web sites are located in California.

The company recently released a study showing that of the 772 COs in its territory, about 230 are hooked up to ISPs. About 62, or one-third, of those were experiencing congestion problems in January. That was up from 23 offices with congestion just three months earlier.

The problem may be small at this point, but planning for the future is necessary, says Jim Diestel, director of advanced services for Pacific Bell.

"Clearly, we have a congestion problem in this business," he says. "We can spend hundreds of millions of dollars to upgrade as a temporary solution, and we are doing that. But we don't recover the costs.

Other RHCs follow that line of reasoning-they can fix it, but they need some help paying for it. Basically, they're saying it's not fair that they have to pay for something the ISPs are causing.

Part of the problem may be the wide disparity in access charges, says Vinton Cerf, senior vice president of Internet architecture and engineering at MCI. Pacific Bell charges 1.4›, but Nynex charges close to 5› per minute for access to the network.

"There's a big disparity in margins," says Cerf. "All of us are interested in seeing the price be a lot closer to the actual cost.

Bell Atlantic is one RHC that is not looking at any of the vendor-proposed intelligence or re-routing solutions. Engineers can handle the problem with tools they already own, says Rick Hofman, director of switching and transport architecture planning. The company plans to look at the vendor solutions more closely in the next generation when those solutions are more efficient.

For now, Bell Atlantic is using several engineering methods to fix its problem. First, it is augmenting and rebalancing switches by moving high-use customers into low usage groups.

Second, Bell Atlantic is adding more fiber capacity between switches to handle the overload. And engineers are changing the way traffic flows at the originating switch by identifying and moving data users around and by offering incentives to ISPs to switch to packet networks.

The last option is the most difficult because ISPs, like most businesses, are usually reluctant to pay more for something they can get free. However, special packages and deals can help convince them of the cost benefits for better quality and throughput. "Those things will do it for many years to come," Hofman says.

Bill McSorley, director of network architecture planning for Ameritech, agrees.

Ameritech, which recently implemented Lucent's AIU switching fabric solution, has also spread ISPs across its switches, changed concentration ratios and added trunk capacity. It also watches the hot spots-five to 10 areas where problems crop up occasionally.

Ameritech is looking to the future, though, trying to decide how to evolve its network. "How quickly will it move to this longer term vision? That's the question," McSorley says. "Our perspective is that the market will drive it.

And so it comes back to the users. The market is already driving up Internet traffic on the voice network at a phenomenal rate. When will users tire of the clogging, waiting to get on-line or putting up with fast busy signals and slow dial tone?

When that time comes, they will investigate alternate carrier networks or, in the case of business users, perhaps look to invest in a system like asynchronous transfer mode that will take them straight to the data network and carry voice and video. For telecommuters, that could mean buying digital subscriber line systems that travel only on packet networks.