What challenges does the Internet service provider market face?

Where do we stand now and who has set out to improve this industry under stress?

By Robert E. Lee

September  1997
[Next story]
[Table of Contents]
Sun's Site

Poor performance. Lack of value-added services. Such are the major complaints from users in regards to their Internet service providers (ISPs). As the number of people dialing into the Internet increases exponentially, so do the problems they encounter. What's being done? We take the pulse of this market and examine some new initiatives directed at resolving the most troubling issues. (2,400 words, including sidebar)

Mail this
article to
a friend
Who is taking more heat about the Internet than the Internet service providers (ISPs)? It's hard to believe anyone else is, since ultimately all perception of Internet performance is placed onto the company providing the initial connection to the Internet.

There is a very definite differentiation occurring in the marketplace today (See sidebar "WorldCom and America Online set new ISP model.") As the mass market for access has become polluted with millions of consumer users drawn to the $20 per month pricing for unlimited access service, many business users are becoming frustrated with the performance of the Internet. Many mid- and small-sized businesses are confused by the need to pay hundreds of dollars more for dedicated or high-speed dial-up access while virtually everyone is becoming disillusioned with the performance associated with these high-speed technologies.

Speculation over the past few months since the NetWorld+Interop Spring 97 show has included the disappearance of smaller ISPs and major Internet providers like AT&T from the dial-up ranks. A spokesperson for AT&T at the show indicated that AT&T is committed to another year in the dial-up access market at the $19.95 price point, but was unable to confirm that AT&T could continue to deliver services at that price point in the future. At the heart of the issue is the intense demand for consistent throughput coupled with nearly flawless performance.

Because the Internet is such an amorphous beast, it is impossible for any one service provider to guarantee complete performance for its customers. As a consequence, there remains a fracture in the marketplace as to what the real potential customer base will be during the next few years. Between the requirements and need to pass Internet traffic between local access providers, regional cross-connect points, and the national and international backbone providers, no single service guarantee can be made by anyone for end-to-end packet transport. It is this issue that drives the cost of the Internet up and destroys the confidence and satisfaction of business and consumers alike.


Ugh -- Internet performance
Increasingly, industry members and outsiders are making calls for a more comprehensive approach to industry problems. Heading the list is perhaps the greatest challenge -- overall Internet performance. To that end, IOPS.ORG has been founded by nine major Internet service providers, including BBN, Earthlink, AT&T, MCI, GTE, Netcom, PSINet, UUNET, and ANS. At the heart of their mission is the development of a core infrastructure between the major Internet service providers to carry the traffic of tomorrow.

Tim Waite, an accounts manager at Exocom, an ISP in Anaheim, CA, is particularly concerned with the ability of new technologies, like cable and xDSL modems, to be deployed by cable TV companies and telephone companies. Just the magnitude of access bandwidth offered by these technologies far outstrips the capacity of even the largest ISPs in the market. Add to that the seemingly poor service performance, lost expertise from downsizing and rightsizing initiatives, increasing lead times, and incredible mixed messages being delivered to the market by these new suppliers, and smaller ISPs like Exocom are left wondering where they ought to head to stay in the market and exploit the niches left open when these companies fail.

Consider the continuing failures in the backbone of the Internet -- outages in late July and early August, where router failures caused domain name resolution services to be significantly impacted, isolated sections of the Internet for as much as a day while routers were rebooted throughout the network. While IOPS.ORG is not out to address the immediate issues of these types of failures, its long-term mission statement includes solutions for these types of problems. In the meantime, users remain frustrated as they see sites appear and disappear for no apparent reason and helplessly watch the ISP community only point fingers in all directions for the source of the failures.

One industry initiative, driven by the Cross-Industry Working Team (XIWT), is the definition of class of service profiles to identify the needs of applications for bandwidth across the network, internal and external. Broken into five classes, each class covers a wide range of services, defined in Table 1. With network speed spanning from 10 kilobits per second to 1 gigabit per second in the class 5 category, the applications covered are everything reasonably imaginable today. These class definitions will aid ISPs and users in matching application requirements to network capabilities, and allow the ISPs to implement technologies that will guarantee these classes of service.

Table 1
Class Bandwidth Type of Application
1 10 Kbps-100 Kbps Electronic Transactions, Voice, Text Mail, WWW Text, Music, Video Conferencing Individuals
2 100 Kbps-1 Mbps Text Mail, WWW Text, Music, Video Conferencing Individuals, Collaborating Workgroups, Interactive Games, WWW Graphics, Video Conferencing Groups, Remote LAN Access, Personal WWW Server, CD-ROM on Demand
3 1 Mbps-10 Mbps Video Conferencing Individuals, Collaborating Workgroups, Interactive Games, WWW Graphics, Video Conferencing Groups, Remote LAN Access, Personal WWW Server, CD-ROM on Demand, Video on Demand, WWW Video
4 10 Mbps-100 Mbps Video Conferencing Groups, Remote LAN Access, Personal WWW Server, CD-ROM on Demand, Video on Demand, WWW Video, Client Campus Applications, High Quality Imaging, Commercial WWW Server, Campus Applications Server
5 100 Mbps-1 Gbps Video on Demand, WWW Video, Client Campus Applications, High Quality Imaging, Commercial WWW Server, Campus Applications Server, Video Broadcast

Commodity vs. value-added services
Following closely on the heels of the infrastructure problem is the issue of what an Internet service provider actually offers its customers. The market is saturated with many small firms providing $19.95 unlimited access accounts -- a pure commodity type business that is driven strictly by price. On the other end are the firms that cater to high-end access needs, at the T-1 level and above. These companies represent a niche market that is growing because they have come to understand the technology and service issues required for Internet business.

The entry of new players over the last two years has been anxiously anticipated by users almost desperate for good service and performance. In California, Pacific Bell Internet has taken nearly two years to deliver a solution to the market. Leveraging the Pacific Bell brand name, over 100,000 customers have joined the dial-up service since its launch in early 1996. But the huge increase in customer base is marred with periodic outages on PBI's network, failures in the authentication servers, mail server outages, and the trials of adding a significant number of new employees to handle the increase in customers. Customers continue to leave the service during these outage times, while new customers move in believing the Pacific Bell brand implies a higher level of service over the not-so-well-known company down the street.

Smaller businesses, now considered the fringe of the market that most of the major ISPs are seeking, are torn between the costs of Internet service and the reliability and capability of an ISP to serve their needs. Recently, Dee's Seas, the resale arm of underwater photographer Dee Lawrence, began an evaluation of ISPs to host the Web site for his works. While a brand name, in this case AT&T, is very attractive from a service perspective, the flexibility and personalized support of a smaller and more local ISP is weighing in ahead of the larger company. The perception is that the smaller ISP will work more closely with the company, be more responsive, and have a more reasonable rate for the services provided.

ISPs will have to rise above the price wars and move mission-critical Internet services out of this no-man's land. Where quality and performance are an issue, users are going to have to be prepared to pay for what they get, and ISPs are going to have to demand that they be given a reasonable margin for their efforts -- in return for a valuable and reliable service end-to-end. ISPs also need to grow the higher end of this market by amortizing the cost of this improved infrastructure over a larger base of customers, making the entry cost more reasonable than today. For example, Speedgate, an ISP in the Orange County, CA, area has an innovative service based on Centrex ISDN. Under special telephone tariffs, Speedgate is able to offer 128 kilobit-per-second bandwidth to a limited geographical area. Its service costs $375 per month, yet the cost of the circuit is only about $80 per month. Even a reasonable margin for the access link plus the Internet backbone access should put this service in the $200 range, opening the door to more customers at a much lower price than competing frame relay technologies.

Everything centers on performance and value-added services. No one buying services is going to argue with reasonable prices, provided the Internet performs reliably and the ISP is able to restore services quickly upon failure. The ISP community recognizes these facts and some are moving to make this a reality, but most still see these issues as outside their control until new technologies or financial incentives bring them to the top of their to-do list.


About the author
Robert E. Lee is a technology consultant, speaker, columnist, and author who has been in the computer industry for 20 years. He specializes in networking, Internet strategies, systems analysis and design activities, and has participated in the Windows NT and Internet Information Server betas since the start of those products. In addition to several other recent feature stories, Rob wrote the June 1997 SunWorld news story, "Cisco throws its support behind Microsoft's directory service vaporware." Reach Robert at rob.lee@sunworld.com.

What did you think of this article?
-Very worth reading
-Worth reading
-Not worth reading
-Too long
-Just right
-Too short
-Too technical
-Just right
-Not technical enough

[Table of Contents]
Sun's Site
[Next story]
Sun's Site

[(c) Copyright  Web Publishing Inc., and IDG Communication company]

If you have technical problems with this magazine, contact webmaster@sunworld.com

URL: http://www.sunworld.com/swol-09-1997/swol-09-isp.html
Last modified:

SidebarBack to story

WorldCom and America Online set new ISP model

On September 8, 1997, WorldCom and AOL announced a complex transaction that shuffled core networking and users services components of AOL and CompuServe into the WorldCom and AOL organizations. For the past year H&R Block has been trying to find a buyer for the CompuServe division, as the company continued to loose money and market share to AOL. The quick recap: WorldCom acquires the network backbone and access technologies of both AOL and CompuServe, while AOL takes the content part of CompuServe and rolls that into a concurrent offering with AOL. Bertelsmann AG, the AOL European partner, will take on a greater role to service the CompuServe services in Europe. For more details, you can check out the press release at http://www.aol.com/corp/news/press/more/970908.html.

More important than just the numbers here, $1.2 billion transactions affecting 2.6 million users, are the implications of the model being produced. AOL recognizes the need to focus its efforts in order to grow the ISP business it is in from the 9 million (prior to the acquisition) users it has to the potential the market presents -- hundreds of millions of users. With the recent massive investments made by AOL to triple the size of its access network and backbones to deliver reasonable quality, it perhaps became apparent that building the network infrastructure and maintaining the content and services delivered over that infrastructure was more than any one organization should handle.

By splitting the networking divisions out of both CompuServe and AOL and turning them over to WorldCom, AOL hopes to let WorldCom take on the headaches and heartbreak of running the networks. This is a strategy that can make a lot of sense and ultimately benefit users at all levels, from the consumer to major corporations. With 100,000 more dial-in ports made available through the agreement with WorldCom to support AOL through its UUNET subsidiary, AOL expects to have 650,000 ports under its control by year-end. This strategy offers a fast fix to the needs of AOL to grow its port count and reduce the ratio of ports to users, which takes AOL one step closer to alleviating the complaints of many users that access numbers are always busy.

The second benefit of this structure is the ability of UUNET and WorldCom to further consolidate bandwidth onto larger pipes and to create a greater effective throughput in the network, plus increase reliability through the various network structures in place. Over the past two years, CompuServe and AOL have poured hundreds of millions dollars into their backbone networks, installing high-speed circuits between their dial-up access points. This network fabric, excellent in each case on its own, gives WorldCom two back-up networks right out the door, including all of the corporate customers that each network has already. In return, AOL gains a stronger backbone for its dial-up base.

On the content front, the CompuServe purchase brings with it a number of online services that don't completely overlap with AOL's offerings. This increased content base, combined with what must be an eventual merging of the user interfaces, under the guise of an Internet browser, will provide the type of incentive to the consumer to choose AOL over Microsoft Network or regional and local ISPs. By focusing on the content, AOL will create a differentiation in the market from the local ISP providers. Very few organizations can afford to create the depth of content AOL offers. CompuServe attempted to do this several times, with WOW being the last big bust, while AT&T bought the Ziff-Davis online venture that also went belly up when its original goals didn't pan out.

So, the services differentiation model discussed in the main body of this article above takes shape as WorldCom continues its focus on delivering the best access and transport technologies and AOL shifts its primary focus to content and membership, using WorldCom as the subcontractor to provide the access services its tremendous customer base requires. MSN went through a similar reassessment as it tried to build a private network with a proprietary content technology. MSN now employs the Microsoft Internet and Java technologies to deliver its content and uses network providers like UUNET to supplement its private network while it divests itself of that aspect. MSN intends to be a content provider. With the acquisitions, AOL now has the market strength to battle the resources Microsoft can muster to steal that market away.

Who will be next? This is a battle between Microsoft and AOL for content and the loyalty of the users, with less of an interest as to how these users arrive at the site. Whether the acquisition will pass the regulatory agencies is another issue, since Microsoft could certainly argue anti-trust due to the overwhelming market share this gives AOL -- but isn't that what they say about Microsoft and its operating systems? The goods news is that content should get better, the networks should get better, and the battle for supremacy on the Internet should get exciting with two major players ready to go after the top spot.


SidebarBack to story