Find You Find Me

Just as the Internet makes it possible to erase geographic boundaries in the learning and business environments, so too does it offer the potential to erase the geographic boundaries that limit traditional telephone directories. Services like Four 11 and switch-board.com have attracted enormous attention because of the simplicity they offer in finding people-wherever they happen to be. Telephone books are great if you want a local plumber or happen to know both the name and address of the person you seek. Otherwise, they make good doorstops.

Finding people and businesses on the Internet is likely to become big business. Uniform Resource Locators offer a crude but not ineffective directory-type feature. Even newcomers to the Internet and the World Wide Web soon learn that if you know the name of a company you want to reach, you can type http://www.companyname.com and there's a good chance you'll find the firm. In many ways, the Domain Naming System has become the first intuitive Internet directory service. It's just not as convenient if you need to locate a person.

Where are the Internet white pages-the directory service that provides people and organizational information? They exist in several forms, and new technologies are under development that promise a truly global-style service. In particular, the Access and Searching of Internet Directories working group of the Internet Engineering Task Force is developing standards for access and synchroni-zation protocols for an Internet directory service.

Several protocols are evolving that aim to offer Internet directory services. They include Internet X.500, WHOIS++, NETFIND, CSO, and RWHOIS. Most important, however, is a hybrid system that facilitates interaction among various services. The ASID working group is attempting to define, evolve, and standardize protocols, algorithms, and access methods for a directory service on the Internet.

The questions are, Do we really need a global Internet directory service? And if so, can it be done well? Who will put the information out there? Who will keep it updated? What about privacy issues? Will a global Internet directory service make the life of Joe-Internet-User easier? Or will it invite unwanted attention?

As the publisher of an Internet white, yellow, and blue pages directory, I am familiar with both the problems and the potential of developing and delivering online directory services. I asked three directory service pioneers what they thought about the need for and the future of Internet directory services and came away with some surprising answers. Here's what some of the experts have to say.

Tim Howes is directory server architect at Netscape Communications Corp. in Mountain View, California. Previously, Howes served as project director and principal investigator of the WINX project sponsored by theNational Science Foundation at the University of Michigan, where he researched directory service problems. As the university's directory service project leader, he was also responsible for designing, implementing, and running the campus directory service, one of the largest and most successful directory services in the world. He is one of the inventors of the lightweight directory access protocol (LDAP), and he led the team of programmers responsible for the widely used and freely available University of Michigan LDAP implementation. Upon completion of his Ph.D. at Michigan, Dr. Howes joined the team at Netscape, making LDAP and a truly global, integrated directory service a reality both at Netscape and on the Internet at large. He continues to be actively involved in the IETF, in which he chairs the Access and Searching of Internet Directories (ASID) working group responsible for LDAP as well as other Internet directory service protocols.

Scott Williamson is vice president of directory services at Network Solutions Inc.(NSI), a wholly owned subsidiary of Science Applications International Corp. In that capacity, he oversees Internet directory services technology development and deployment. The RWhois technology, of which Mr. Williamson is coauthor with Mark Kosters, is in use as the distributed information tool of the InterNIC domain and address registration worldwide. Previously, Mr. Williamson served as principal researcher for Thomson Corp.'s Thomson Technology Internet Lab. Prior to his joining Thomson Corp., Mr. Williamson served as program manager of network solutions and was responsible for Internet-related projects that included the InterNIC, DDN NIC, and CDPD NIC.

Einar Stefferud is founder and president of Network Management Associates, Inc., in Huntington Beach, California. His firm provides strategic technical advisory information services for management problems of computer/communication networks and workplace automation. He is currently a principal of First Virtual Holding Incorporated, where he serves as chief visionary in addition to maintaining his regular consulting practice. Over the years, Mr. Stefferud has been deeply involved in prestandards work on X.400 and X.500 in IFIP WG 6.5 and in poststandards profile work in the NIST OIW. He has also been involved in the ANSI OSI Registration Authority, US-NMTS-IG (the U.S. National Mail Transfer Service Interest Group), the IETF, and IFIP Working Group 6.5. He was instrumental in establishment of several Internet standards.

Estrada: Since the dawn of white pages and yellow pages directories, directory services have become something that most people take for granted. Companies are already producing paper versions of white and yellow pages Internet directories, but online directories are still evolving. In your opinion, do we need online Internet directory services? If so, what form will they take? And when do you think the need for them will reach critical mass?

Howes: We absolutely need them. And we need them yesterday or sooner. But one thing to realize about a directory is that, more often than not, it is not a goal in and of itself. Rather, a directory provides the means by which the real goal, such as contacting a friend or business partner or accessing a network resource, is achieved. Directories allow us to realize those goals faster, more easily, and more efficiently.

To make that happen, the directory we choose must work within other applications and be accessible directly by users over the Web. It must be structured enough to facilitate automated searching, but flexible enough to handle new kinds of data and fit into different environments.

Those requirements and others have led more than 50 companies, including every major software vendor, to endorse the LDAP as the standard Internet directory service protocol earlier this year.

Williamson: Directory services are needed to glue the pieces of the Internet together. As the level of technical ability continues to drop for the typical user of the Internet, a guidance mechanism that applications can communicate with on behalf of users will become critical.

The ideal directory service will maintain objects of all types-from network numbers to users, to Cu-See-Me reflector resources. I think that over the next six months we will see infrastructures being built to support emerging directory products. NSI has a team of highly skilled engineers working diligently to build a global infrastructure based at NSI and driven primarily, but not exclusively, to support the InterNIC information tree.

Stefferud: All meaningful directories are local to some kind of operational need, such as the Password file or the Domain Name System [DNS], which must be working in order for some system to operate at all. Consider the Italian Lookup! Many telephone directories are out-of-date. Calling information does not work. So what do you do if you need to find a number? You call a friend. Italy is proof of the need for fault-tolerant systems.

Consider the U.S. Postal Service [USPS]. When was the last time you asked your postmaster for an address? When was the first time? The U.S. Postal Service has not yet failed, probably due to the lack of a directory. It has problems, but lack of a user directory is not one of them. There are zillions of mailing lists that facilitate the spewing of tons of junk mail on all kinds of people. It's hard to imagine how a national USPS public user directory would benefit us.

Consider the Internet. Chaos reigns supreme. Is the Internet another proof of fault-tolerant systems? Very likely. The Internet has been doubling every year without a comprehensive directory other than the need for a directory of technical and administrative addresses for all connected domains.

My take is that the only reason for needing anything like a global user directory is to feed mappings to e-mail gateways, which are necessary because of the e-mail infrastructural mess that derives from the way e-mail grew up as a set of disconnected enclaves. Get rid of the gateways, and the need for the directory will largely disappear.

Estrada: Who do you think will be the major players in developing and deploying directory services? Will the IETF play a significant role?

Howes: The IETF has played a major role in the past and will continue to do so. LDAP was developed by the IETFD ASID working group. The IETF provided a vendor-neutral ground and technically sound development and review processes that made the unprecedented industrywide adoption of LDAP possible.

The IETF will continue to play an important role as LDAP evolves, even as the market begins to drive development of the standard. Right now, for example, the IETF is working on defining the next version of LDAP, which will better address security and internationalization issues, among other things.

Vendors also play an important role in the directory service realm, just as in other areas. We've seen this with HTML and other standards, and directory service will be no exception. Vendors are now on a faster track than the IETF itself, which means they will often lead the standards efforts, bringing new features to the user community before they are fully introduced to the standards process. This is a healthy process-the logical consequence of the IETF's rough-consensus-and- running-code approach.

Williamson: I think several players are struggling to build an offering in this area. I think the major players-other than NSI-will be Netscape, Microsoft, No-vell, Banyon, AT&T, and Bunyip. We at NSI are refining the RWhois protocol-a protocol that by nature reduces a query in the effort to determine whether it can locate a server to respond to the requested information. We believe that in the near future, the result of our efforts will allow for distributed, secure, and authoritative registration of network objects by network operators as well as independent applications. Using the referral mechanism and authority area, it is our intent to have RWhois glue several directory services protocols together.

In order to get successfully from point A to point B, there has to be a starting point somewhere in the Internet. Hierarchical data can be easily located by using DNS SRV resource records-when the change is made to DNS server software-or by following referrals from a hierarchical directory server. But the discovery of nonhierarchical information continues to be a challenge. The technology to accomplish this-common indexing protocol, or CIP-is currently being developed as part of IETF efforts in the FIND working group. The objective now is to get the directory service technology de-velopers-RWhois, LDAP, and Whois++-to agree on how to proceed from there.

Stefferud: I think that main sources of directory services are going to be entrpreneurs who gather them and sell them, like the folks who are selling those CD-ROM compilations of zillions of addresses. The scrounging technologies for collecting addresses into compilations will be the winning endeavors. Netfind and its friends will be used to populate these entrepreneurial directory services.

Someone may someday figure out how to put these directories online for sale by the query, but this requires very low prices for queries.

Estrada: How and when can we expect the current white pages implementations to be globally searchable?

Howes: We can't, although we will continue to see subsets of the global data searchable, just as they are today through large sites like Four 11, switchboard.com, and WhoWhere. A less centralized approach, such as that implied by the CIP being developed by the IETF FIND working group, may also have applications in certain environments, for example, among universities or research organizations. We will also see an increasing number of peer-to-peer agreements, allowing searching of white pages information between business partners. But privacy, legal, and data ownership issues will prevent a truly globally searchable white pages service from ever achieving worldwide scope.

Williamson: This is a difficult goal to accomplish. If our names were hierarchical, we would have this solved with the current RWhois or LDAP technology. But they're not, so building a common indexing protocol server, such as CIP or a similar technology, that is fed by these hierarchical directory services is critical. Scaling then becomes the major issue. Just try and find John Smith by name only. At Network Solutions, we are working to find a solution to at least offer a starting point to those who are searching directory services. I don't think we can solve the John Smith problem, but we will try. Our current focus is determination of attribute relationships in the common indexing area to allow for reduction of irrelevant referrals.

I think that support for nonhierarchical searching will roll out over the next year. The hierarchical infrastructure should be built first. It is in the hierarchical tree that objects should live-such as allowing add, mod, and del actions- while their nonhierarchical attri-butes should be fed to the CIP-type servers. White pages will require nonhierarchical search services.

Stefferud: "Pity the Poor Fanatic! When He Loses Sight of His Objective, He Redoubles His Efforts!"

Reference

ASID Working Group Documents at the IETF Web Server, http://www.ietf.org.