Networking's greatest debates in LANS + WANS

Including IPv4 vs. IPv6, routing vs. switching, Packet switching vs. circuit switching, Ethernet vs. Token Ring, and Cisco vs. the rest,

Routing vs. switching

It was the debate that separated Cisco from the pack and forever changed the enterprise network industry.

The routing vs. switching wars of the mid-1990s pitted routing kingpin Cisco against virtually every other competitor in the enterprise space: Bay Networks, 3Com, and Cabletron, which all promoted a flat, Layer 2 switched infrastructure.

At its root, the debate centered around whether businesses should construct hierarchical router networks in order to better segment and administer traffic and workgroups; or if flat Layer 2 infrastructures with virtual LANs establishing broadcast domains did the same thing, only less expensively.

The stand-off also ushered in technologies such as the routing hub -- or "Rub" -- and the Layer 3 switch. It also produced such innovations as Ipsilon's IP Switching, and Cisco's Tag Switching for service providers, one of the early incarnations of what is now known as MPLS.

The debate also made for memorable marketing slogans, such as Bay Networks' "Switch when you can, route when you must."

But crafty slogans weren't enough to keep the market from moving to Cisco. Because of its ability to meld de facto standard routing with LAN switching, deft marketing and a bit of a price compromise, Cisco was able to lock up the Layer 3 switching market to complement its dominance in routers.

Bay, 3Com and Cabletron put up a valiant fight but in the ensuing years, these companies vanished or were marginalized as bit players in enterprise networking.
Jim Duffy

Fat WLAN access points vs. thin WLAN access points

This is one argument that was decisively ended. And then started up again.

Historically, wireless LANs (WLAN) relied on "fat" access points, which handled a wide array of tasks in software, each a separate IP address wired directly into Ethernet switches. All that changed around 2001 with the introduction of the WLAN switch (usually now called a controller) from start-ups such as Airespace, Aruba Wireless Networks and Trapeze Networks. Most of the access point's functions were shifted to the controller, which incorporated the Ethernet switch. The argument: centralize management, security administration, client handoffs and more. That argument seemed over when Cisco paid almost half a billion dollars to acquire Airespace.

But during the past 12 months, WLAN vendors such as Trapeze have been offloading jobs like data forwarding from the controller back to the access points. The new argument: less load on the controller, no single point of network failure, and reduced network latency and jitter.

In May 2007 a brand new start-up, Aerohive bet the farm on a more radical version of this idea. Areohive distributes all of the controller functions through a mesh of intelligent access points, each with its own IP address. They work cooperatively to do the task formerly done by a separate controller. Aerohive won't dethrone Cisco in the WLAN space anytime soon, but it suggests renewed efforts to distribute specifically wireless intelligence more pervasively through the network.
John Cox

Page Break

Packet switching vs. circuit switching

Circuits as WAN connections battled successfully against packets for years because they guaranteed bandwidth, no matter what.

If you bought a T-1 circuit, the service provider nailed up 1.5Mbps from point A to point B, and that was your bandwidth come hell or high water.

Packets -- first frame relay then IP -- didn't do that. They shared the network capacity, and service providers typically oversold access lines, knowing that if all customers tried to use all the bandwidth they bought at the same time the network would be swamped. Everybody's traffic would suffer delays.

But frame relay service was relatively inexpensive compared with T-1s and it came in connections smaller than 1.5Mbps, usually in increments of 56Kbps. That meant sizing circuits to just handle demand, making business WANs more cost effective.

Plus frame relay services supported bursting -- the instant delivery of more bandwidth than customers contracted for to handle moments of spikes in traffic. Because the networks were oversubscribed, sometimes the bursts were available, sometimes not. But over time, they were available enough of the time to make the service attractive.

Frame relay also offered virtual circuits -- the logical carving up of access lines into lower bandwidth logical links tied to different sites. If a site needed connectivity to six others, it needed a single physical line subdivided into six virtual circuits. In the world of circuits, nothing was virtual. If one site in a corporate network needed to connect to six others, it needed six physical access lines.

As frame relay and later IP services matured, QoS was built into their standards so bandwidth could be guaranteed, just as it was in a circuit. That advance put packets over the top.
Tim Greene

QoS vs. more bandwidth

When a drain chronically runs slow even though it isn't plugged, it's time to get a bigger pipe.

That is pretty much the conventional wisdom about network connectivity -- if links slow down, get bigger links.

That has proved to be a useful rule if you judge by the progress of Ethernet, for example. It started out slow at 10Mbps, then grew faster at 100Mbps, then 1Gbps and now 10Gbps with higher speeds coming into sight. As network traffic increased, networks slowed because of congestion, so businesses built faster networks.

The limiting factor has always been cost. If important enough traffic slows down enough, businesses invest in bigger bandwidth. Inevitably, as the latest, fastest technology matures, the price-per-port drops so it becomes affordable even to those who don't need it.

QoS prioritizes traffic by type so the really important stuff gets the bandwidth it needs to cross the network unimpeded by the unimportant stuff. This can be done by queuing, dedicating bandwidth and rate-limiting low priority traffic.

Even if bandwidth is plentiful and congestion is unlikely, businesses may implement QoS anyway to make doubly sure that real-time traffic such as VoIP and IP video don't get bogged down. But the hassle of implementing QoS on routers for no noticeable gain has many businesses ignoring it.

The story is different over the WAN where more bandwidth means bigger bills month after month. There's no way to wait for a faster technology, buy it for lump sum and reap its benefits as in the LAN.

QoS to mete out bandwidth to the applications that require it becomes necessary if money is factored in. So this debate has a split decision: more bandwidth in the LAN and QoS in the WAN.
Tim Greene

Page Break

Frame relay vs. VPNs

VPNs burst on the scene in the late 1990s as a way to use the Internet securely and cheaply as the backbone for corporate WANs.

The downside was that the Internet was unreliable and insecure, but the Internet proved reliable enough for most data purposes, and security in the form of IPSec locked down connections between sites.

Frame relay didn't have either problem. As frame relay provider networks matured, they were adequately provisioned so bandwidth paid for was bandwidth delivered. As a Layer 2 switched technology, most customers felt that it was inherently secure and if businesses were worried, they could tack on their own encryption.

With the rise of MPLS, a new form of VPN came into being. Rather than installing VPN gateways at each corporate site to create encrypted tunnels to other sites, MPLS required just a router connection to the provider access line. The service provider would route traffic to any other corporate sites similarly connected to the network.

In the world of frame relay, such meshing of connections so all sites could connect to all other sites required setting up virtual circuits, the division of access lines into logical links to other sites. The result is similar to having MPLS route traffic among all sites, but MPLS doesn't require virtual circuits - a significant cost savings that customers couldn't ignore.

"Having any-to-any connectivity was a very big driver for us," Catherine Watterson, vice president of enterprise engineering services at Policy Studies told Network World in 2005, when her company was in the midst of moving from frame relay. "We wanted to simplify our network in comparison to what we had with frame relay, where everything came in to our headquarters location. We also were interested in lowering our costs."

With IPSec and MPLS teaming up to provide functionally similar and less costly service, frame relay lost its edge and is gradually being replaced by one or the other.
Tim Greene

Ethernet vs. Token Ring

It was the classical IBM vs. Digital Equipment battle taken to the network theatre.

Two computing behemoths each lining up behind competing LAN architectures: IBM advocating Token-Ring and DEC endorsing Ethernet.

Pioneered by IBM scientist Olof Soderblom in the 1960s, Token-Ring was initially successful for IBM -- as most IBM-endorsed technologies are. But beyond IBM, it never garnered much support among top-tier vendors, which gravitated toward Ethernet, the preferred LAN technology of research institutions and manufacturing companies.

Token-Ring began its downward spiral soon after the emergence of 10Base-T Ethernet, an inexpensive, 10Mbps transmission technology than ran across telephone grade unshielded twisted-pair copper media. Matters were not helped when Soderblom began demanding royalties from the few Token-Ring vendors and chip makers, driving prices higher.

And then, Cisco, citing lack of market demand, dropped out of an effort to develop the 100Mbps High-Speed Token-Ring (HSTR) in 1998, a move that essentially scuttled that initiative.

"Cisco is attempting to torpedo multivendor, industry-standard HSTR efforts and promote, instead, the Cisco proprietary technology," said Kevin Tolly, president of The Tolly Group, a catalyst behind the formation of the HSTR alliance and a Network World columnist.

Today, Token-Ring is barely detectable in the market as businesses have opted for switched Ethernet in the LAN. Ethernet has even spread into the WAN as telecom carriers now offer Ethernet services on a local, metropolitan area, regional and national basis.
Jim Duffy

Page Break

Cisco vs. Bay/3Com/Cabletron/Juniper

The subplot of the routing vs. switching debate was the fierce competition between Cisco and enterprise rivals Bay, 3Com and Cabletron. Over time, Cisco also deepened its focus on the service provider market, where start-up Juniper Networks, with the financial backing of several Cisco rivals in data and telecommunications, was becoming a formidable router alternative.

Perhaps only IBM faced a similar "gang up" mentality from other leading companies in its industry. But it was also a compliment, a show of respect and fear of Cisco's potency and eventual dominance in the market through a combination of technology and marketing savvy, aggressive pricing and an ambitious acquisition strategy that enabled Cisco to burst its way into hot markets from storage to video.

Bay was formed out of necessity from the merger of two smaller Cisco rivals -- SynOptics in enterprise hubs and switches, and Wellfleet in enterprise routers. But even the combination of two major players in their respective markets could not slow the Cisco juggernaut -- Bay was eventually acquired by Nortel, which remains a distant No. 2 or No. 3 to Cisco in enterprise networking.

Before that, however, Bay combined with IBM -- which had an ill fated LAN infrastructure operation called Networking Hardware Division (NHD) -- and 3Com to form the Network Interoperability Alliance (NIA). The group claimed it had formed to enhance interoperability between ATM switches; but observers noted that the vendors really got together to gang up on Cisco and try to thwart the company's increasing dominance in enterprise networking.

The toothless NIA folded with nary a whimper after three years, having made virtually no impact on the market or user buying decisions.

IBM, meanwhile, folded its NHD hand in 1999 by selling its routing and switching technology to Cisco.

Cabletron, meanwhile, committed a competitive faux pas by having its Cisco IOS router software license revoked after a trade show stunt. Cabletron resold a Cisco router blade in its hubs and switches. But after the company showed a marketing video at a trade show of the Cabletron Crusher beating the flabby Cisco Kid to a pulp, Cisco yanked the IOS license.

"We want healthy competition, but Cabletron has gone beyond that with the things they're doing," Alex Mendez, at the time Cisco's vice president of marketing for enterprise networking, said then.

Cabletron was gone a few years later, having split into four companies in 1999. What's left of the company's switching hardware operations -- Enterasys -- was eventually taken private by an equity investment after an accounting scandal toppled top management. It is also now a bit player in enterprise networking, ranking No. 7 in switched Ethernet revenue market share in 2006 with 1.5 percent, according to Dell'Oro Group.

Juniper is having better luck against Cisco in the service provider market than Cabletron, Bay, IBM and 3Com had in enterprise. The company stole one-third of Cisco's overall service provider router market share and remains a viable competitor and alternative to Cisco in that market. Juniper is also looking to broaden this success into the enterprise market, where it acquired VPN and firewall leader NetScreen technologies. The company is also expected to enter the LAN switching arena soon.
Jim Duffy

Page Break

Ethernet vs. MPLS in the WAN

They began as totally separate and distinct technologies: Ethernet as the standard in the LAN, and MPLS VPNs as an attractive alternative to frame relay and ATM services in the WAN.

But now they are starting to collide. Service providers are turning up Layer 2 Ethernet VPNs based on an MPLS derivative called Virtual Private LAN Services (VPLS) on a regional, national and eventually, global basis. And while some carriers say VPLS Ethernet is a complementary access or metro technology to MPLS national and global services, some acknowledge that users are reconsidering their Layer 3 VPN decisions.

VPLS is intended for businesses that prefer to maintain control of their routing, for security and staffing purposes, rather than share it with their service provider. Layer 3 MPLS VPN users choose to let the service provider manage the routing domain.

This will usually be the chief determinant of whether an enterprise selects a Layer 2 or Layer 3 VPN service, carriers say. But some users have also cited the ability to scale multicasts as a benefit of Layer 3 MPLS VPNs, especially as they migrate to service-oriented architectures.

MPLS VPNs have been around a few years longer than VPLS. But as VPLS continues to mature and become more functionally complete, the decision to pick one over the other will become harder -- and the ability of carriers to keep one from cannibalizing the other will be tougher as well.

"I don't think it's a zero-sum game," said Josh Holbrook of the Yankee Group. "I don't think one of those services will prevail over the other."
Jim Duffy

IPv4 vs. IPv6

The argument about how best to upgrade the Internet's main communications protocol raged in the Internet Engineering Task Force in the early 1990s. By then, experts realized that the Internet would eventually run out of address space with the original version of the Internet Protocol, known as IPv4.

The issue of what direction to take with the next-generation of IP came to a head at a 1994 IETF meeting in Toronto. Ultimately, the IETF decided to replace the 32-bit addressing scheme in IPv4 with a 128-bit addressing scheme in IPv6. The standards body tried to create other reasons to upgrade to IPv6, including built-in security with IPsec and easier management through autoconfiguration of devices.

Nearly a decade after IPv6 was finalized, the network industry has yet to embrace the new protocol. That's because a forklift upgrade to IPv6 is too expensive and time consuming for a carrier or enterprise, with little measurable return. Instead, the network industry anticipates a gradual transition to IPv6, which will likely run side by side with IPv4 for many years to come.

Now it appears that IPv6 is finally winning this argument. The American Registry for Internet Numbers recommended in May that the Internet community start migrating to IPv6.

In fact, some industry experts predict that there are only around 1,200 days left until the Internet runs out of IPv4 addresses. Leading the charge to IPv6 is the U.S. federal government, which has mandated that all agencies support the new protocol in their backbone networks by June 2008.
Carolyn Duffy Marsan

K56flex vs. x2 56Kbps modems

In this argument over technologies the winner is not a technology at all, it's its inventor Brent Townshend.

He ultimately prevailed in his legal claim as inventor of the modem that capitalizes on a vagary of the public phone network that makes it possible to send data to a dial-up modem at 56Kbps, faster than the 33.6Kbps limit on sending data from that same machine.

In 1997, when broadband Internet access hadn't been invented, the 56K modem was a major advance that industry heavyweights laid claim to, most importantly U.S. Robotics, touting its x2 technology, and Lucent, backing K56flex.

Page Break

Interest in their modems was sky high, but they didn't interoperate, so customers - particularly ISPs whose modem banks would have to support the technology in order for end users to reap its benefits - steered clear. No ISP wanted to bet on x2 or K56flex and wind up having to upgrade them later if it turned out they bet wrong.

The warring factions decided they were killing the market for the faster modems, and agreed to a common standard known as v.pcm. After that some 95 million of the modems were in service within two years.

It seemed like a simple solution, but all the issues weren't settled. Townshend still held five patents on the underlying technology and after protracted lawsuits, exacted royalties that at their peak amounted to US$2.50 per modem. That netted him enough money - do the math - to carry on defending the patent for years. As late as 2004, he was still settling with vendors who had been using his invention for years without paying royalties.

For all the trouble it caused him, he thinks it was less hassle than starting up his own modem business. "I said, 'This is an easy thing to do,'" he said in a 2004 interview with Network World. "I can just license this to people that are in the modem business. I don't have to start competing with them or set up my own distribution."
Tim Greene

SNA and OSI vs. TCP/IP

The long, slow and at times, colorful battle between TCP/IP backers and SNA stalwarts was one of the more acrimonious periods of time in the industry's history. IBM introduced the concepts and initial components of SNA in 1974 and by the mid-1990s it was locked in a do-or-die battle with IP's chief proponent - Cisco.

Cisco and the industry had some fits and starts in promoting TCP/IP and a real SNA alternative, at one point forming the Advanced Peer-to-Peer Internetworking (APPI) group to counter IBM's SNA advanced technology called APPN (Advanced Peer-to-Peer Networking). While APPI was ineffectual and folded in 1993, it planted seeds that would ultimately sow SNA's demise.

For its part IBM did just about everything wrong in upgrading older SNA users to APPN -- first it made APPN gear complicated and costly -- it even wanted royalties from potential third parties that could have helped establish the technology. In 1994, an IBM executive promised to "kill" vendors (mostly Cisco) and "eat our own young" before IBM would let any rival cannibalize SNA/mainframe business.

In addition, IBM threatened lawsuits over the use of some of its technology. While IBM had started losing customers, it still had more than 50,000 SNA installations, so its blustering had some power. But with Cisco licensing Big Blue's mainframe channel technology in the mid-1990s and the advent of Data Link Switching (DLSw) and tn3270 (which let SNA run over IP networks), the nails were in the coffin. Combine that with the fact that at about that time Gartner said users with SNA as their primary protocol (after about 1995) will spend a total of 20% more than IP users on training staff, hardware and software purchases, and administration.

You didn't need a Magic 8-ball to see the outcome. IBM sold most of its young to Cisco in 1999.

Open Systems Interconnect (OSI) technology backers were also making a lot of noise during this period. OSI products and services were based on an international set of standard protocols that were supposed to guarantee interoperability among all vendors and products. IBM and Digital Equipment Corp., in fact, were two of its chief proponents, despite the fact OSI ran completely counterintuitive to SNA and DECnet respectively. The U.S. government throughout the 1980s and 1990s mandated the use of OSI products but changed its tune in 1994 saying TCP/IP could be used instead. The TCP/IP juggernaut moved on and widespread use of OSI products never happened.
Michael Cooney

U.S. Department of Justice vs. AT&T

In the Ali-Frazier of last century's heavyweight telecom fights, the U.S. Department of Justice went toe-to-toe with AT&T after filing an antitrust action against the carrier in 1974.

Specifically, the Department of Justice accused AT&T of engaging in anti-competitive behavior and sought to break up the company. Invoking the Sherman Antitrust Act in its case, the government said that AT&T had monopoly power over America's telecommunications, and argued that the company should sell off some of its subsidiaries, such as manufacturer Western Electric and research arm Bell Laboratories, which would then be carved into even smaller companies.

Page Break

The government's actions set off a fierce public debate. Proponents of the Justice Department's trust-busting suit argued that breaking up AT&T would allow more companies to enter the market place, thus spurring greater innovation and competition. Opponents countered that AT&T should be exempt from antitrust rules in order to maintain uniform standards in telecommunications services. Breaking up Ma Bell, they said, could lead to a fragmented, disorganized telecom industry.

"AT&T provides the very cheapest service possible," Carl Glick, a telecom analyst, told Time Magazine. "Justice gets so wrapped up in its rhetoric about the advantages of competition that it loses sight of the economic implications of its moves."

After years of legal wrangling, AT&T eventually agreed to a settlement on the government's terms in 1982. Under the agreement, AT&T would be allowed to keep its long-distance operations, Western Electric and Bell Laboratories in exchange for divesting from its 22 local phone monopolies. Since then, AT&T has had incremental success at gaining back some of its previous clout, as the company was allowed to merge with BellSouth, one of the RBOCs that formed after AT&T had been forced to divest from its local phone services.
Brad Reed

CLECs vs. Bells

It's been more than a decade since Congress passed the 1996 Telecommunications Act with the intent of helping competitive exchange carriers compete against the regional Bell companies. In that time, however, competition for local phone services has gotten significantly weaker.

The 1996 telecom legislation mandated that incumbent carriers allow CLECs to use their existing infrastructure to provide last-mile connectivity to their customers. Additionally, the act charged the FCC with deciding what bundles the incumbent carriers would be required to provide CLECs, as well as the wholesale rates they could charge for access to their networks.

The Bells, however, were not ready to play nice. Employing a two-pronged strategy of simply refusing to comply with the act and of challenging the act's constitutionality in court, the Bells were largely successful in warding off competition from CLECs, which also suffered from poor market conditions in the early part of the decade. Between 2000 and 2003, roughly one-third of all CLECs nationwide filed for bankruptcy, leaving the incumbents firmly in command of most local telephone service markets.

CLECs have been reasonably successful in the more urban markets, however, where having potential customers clustered closely together made it profitable for CLECs to build out their networks. In the cases where CLECs have gained significant market share, incumbents have indeed felt obliged to offer more discounts for customers, just as the '96 telecom act envisioned. However, recent incumbent-friendly rulings issued by the FCC, which limit what network elements the incumbents are required to share, indicate that the going might not get any easier for CLECs for some time.
Brad Reed

PBX v. Centrex

Centrex may have lost the debate to PBXs but the service technology refuses to be silenced.

At first, Centrex was pushed as a fair tradeoff for a business-owned and managed PBX phone switch. The service was viewed as comparatively expensive, dependent on carriers to make management changes and, depending on the carrier, having limited features.

PBXs had none of these shortcomings, but customers had to lay out the cash to buy the switch, manage and maintain it and buy the trunk lines that connect corporate sites into a private phone network. And they had to pay for annual maintenance and periodic software upgrades.

Such PBX virtues also explain the tenacity of Centrex particularly among smaller businesses that have limited resources for a telecom staff to take care of their own phone networks. They simply can't afford an investment in running a private phone network.

With service providers providing and managing all the gear Centrex requires, capital outlay is minimal, as is the level of expertise needed to make the phones work.

Page Break

With the advent of VoIP, less expensive - free for open source versions - IP PBXs have emerged, and the debate is renewed. IP PBXs cost less than their predecessors and are designed to be simple enough that relatively non-technical people who run small businesses can operate them.

VoIP also means a morphing of Centrex services into managed IP PBX services, with actual IP PBXs being installed on site with management interfaces that put customers in control of features and configuration if they want it.

But having carriers manage devices on customer networks rather than managing devices within their network cloud opens up a separate set of problems. "It's one thing to be an expert on the cloud side, and quite another to look into a customer's Ethernet infrastructure, see which component is causing a problem and fix it remotely," says Ed Basart, the CTO of IP PBX vendor ShoreTel in a Network World debate on the topic

A counter argument is that IP PBX services make the cost of phone service more predictable. "Hosted solutions' flat monthly fees eliminate the guesswork," says Dan Hoffman, president and CEO of provider M5 Networks.

With the control they give and the intimate linking of voice with other applications they can provide, owning IP PBXs still wins the argument, but not once and for all. Centrex and its offspring will continue the debate.
Tim Greene

Net neutrality v.s. tiered services

In the mid-1990s, the comic book industry's two titans combined forces to release "Marvel Comics vs. DC," a short series that pitted such classic DC heroes as Superman, Batman and Wonder Woman against Marvel icons such as Spider Man, the Hulk and Wolverine.

The debate over network neutrality is very similar to this infamous clash of comic book stars. Why, you ask? Because seemingly every major power in telecommunications, politics and IT - from Congress to the Department of Justice to Verizon to AT&T to Google - has staked out a position and is waging a fierce battle for consumers' hearts and minds.

Network neutrality - often commonly referred to as 'Net neutrality- is the principle that ISPs should not be allowed to block or degrade Internet traffic from their competitors in order to speed up their own. Several consumers' rights groups, as well as large Internet companies such as Google and eBay, have led the charge to get Congress to pass laws restricting ISPs from blocking or slowing Internet traffic, so far with little success. The major telcos, meanwhile, have uniformly opposed net neutrality by arguing that such government intervention would take away ISPs' incentives to upgrade their networks, thus stalling the widespread deployment of broadband Internet. In order to keep maintaining and improving network performance, say net neutrality opponents, ISPs need to have the power to use tiered networks to discriminate in how quickly they deliver Internet traffic.

Yet despite all the heat this fight has raised, neither side looks as though it's backing down anytime soon. Although the Senate proposed new net neutrality legislation in January, there have been no significant advances on that front since then. Previous Congressional efforts at passing net neutrality legislation have proven so far unsuccessful. But net neutrality will almost certainly continue to make headlines, whether they come from forceful denunciations written by the Justice Department or from revelations that Comcast blocks some peer-to-peer traffic on its services.
Brad Reed

VoIP vs. TDM voice

TDM isn't dead yet as the backbone of corporate phone networks, but it is definitely in its last throes.

TDM, phone technology based on circuits switched by venerable PBXs, provided valuable services for a long time. Everybody got their own extension, a fancy corporate phone, voice mail and sometimes extra features such as conferencing, multiple call appearances and caller ID.

It was reliably maintained by well-trained telecom staffers who knew their switches like the backs of their hands.

Page Break

When VoIP came along, promising to run voice calls over IP networks for less money, customers were intrigued. Rather than send out a technician to move a phone extension from one office to another, the user just had to move their phone to a new jack and it would receive calls meant for the person it was assigned to, no technician required.

Rather than supporting two separate networks, VoIP riding on the data network would allow them to get by with one. If voice was just an application running on the network, the telecom staff wouldn't be necessary or at least wouldn't need to be so large.

The cost savings proved to be a red herring, but in the cases where business adapted VoIP, they found other reasons to like the technology. If a customer help line in one office was swamped, more agents from anywhere on the network could quickly and painlessly be routed into the call-distribution list and start fielding calls.

VoIP supports services TDM doesn't, such as popping up account information about customers on call agents' computer screens based on the caller IDs or enabling users to place calls by clicking on phone numbers within applications or finding out who is available to take a call transfer by using presence information that VoIP supports.

Big name converts to VoIP, such as Bank of America, have spoken out about VoIP's benefits. "People have asked me what is the killer application? I say by deploying a VoIP system we're getting just out-of-the-box features that people take for granted with VoIP. But when you come at it from the context of coming from a traditional key system, or PBX, this is all new," said Craig Hinkley, the Bank of America network pro who oversaw the project. "Being able to go into a Web page, and configure speed dials. And if I'm out of the office, the ability to go to a Web site and forward my phone to my cell phone. That's a quantum leap forward for a lot of people."

If these weren't reasons enough, PBX makers are phasing out TDM altogether. Sales of TDM PBXs to replace ones that have reached the end of their useful lives are insignificant compared with VoIP.
Tim Greene

Cable vs. DSL

While they didn't rack up a body count on the scale of a Gambino-Colombo (or even a Jets-Sharks) feud, the turf war between cable and DSL got pretty intense in the early part of the decade.

With consumers demanding high-speed Internet services to replace their clunky dial-up modems, the telcos and cable companies began offering triple-play services that wrapped voice, Internet and television services into one single-bill package. The cable companies' weapon of choice was the cable modem, which is designed to take advantage of unused bandwidth over cable television infrastructure. The telcos, meanwhile, hitched their wagons to DSL, which provides Internet connection through local telephone networks.

Service-wise, the battle between cable and DSL was something of a draw. Despite the fact that cable Ethernet cards limited download speed to 4Mbps, cable modems were fairly comparable in price to DSL services, and cable Internet was much more widely available than DSL in the early part of the decade.

As competition between the two services has intensified, global prices for broadband Internet have dropped. The first half of 2004, for instance, saw the price of cable fall by 16%, while DSL fell by 13% over the same period.

Getting a head start in offering broadband was crucial to cable securing an early advantage over DSL, and by 2004, a U.S. Telecom Association report showed that 64% of all American broadband Internet subscribers, while DSL accounted for only 26% of subscribers. By the end of 2006, though, DSL had substantially closed the gap, accounting for 44% of America's 51 million broadband subscribers, compared with 56% for cable. Thus, while cable continues to be the No. 1 broadband service in the United States, DSL has made rapid gains, and could be poised to overtake cable in the next couple of years. One potential challenger to DSL and cable has emerged in the triple play market in the form of Verizon's FiOS. But while FiOS gets a considerable amount of hype, the service is still not widely available, and it only reached its one-millionth subscriber in June of this year.
Brad Reed

Read Networking's greatest debates in Security
Read Networking's greatest debates in Software
Read Newtorking's greatest debates in the Data Center
Read Newtorking's greatest debates in Management