Co-chairs Owen Ambur and Marion Royal convened the meeting at 9:30 a.m. at the General Services Administration. Mr. Ambur introduced new attendees. The chairs asked for any comments on/corrections to the October 18 meeting minutes. None were forwarded and the minutes were approved.
Bruce Esposito and Larry Mooney of Novell presented Novell’s DirXML data sharing service to the WG.
Larry Mooney provided the WG with an introduction to DirXML. DirXML is how Novell is tying XML into directory sharing and storing strategies. DirXML is event driven, so the changes to your data are propagated through your environment in real-time.
Bruce Esposito discussed the specifics of DirXML. DirXML is an XML based solution that uses XML to share information across directories and systems and create one metadirectory of common data. DirXML can pull data from different sources residing on different systems and anyone in an agency/organization can query this data. The data is fed into the metadirectory from its native systems by DirXML and thus avoids the errors associated with manual entry. Where traditional metadirectories have utilized a good deal of customized code, DirXML has an open architecture that requires a small amount of custom code. The DirXML solution takes data from its native directory or database and shares it with another directory or database. The business logic between the two directories/databases is based on XML. DirXML is the only solution of its kind.
DirXML is based on Novell’s NDS eDirectory. DirXML drivers interact with applications or systems (e.g. Lotus Notes or PeopleSoft) to pull requested information. This data then enters the NDS database, where it can be queried by any subscriber and be pulled into another application via another DirXML driver.
An example of the utility of DirXML is the ease with which a new employee’s various work accounts (e-mail, payroll, insurance, etc…) can be established. Where traditional methods would require repetitious data entry, DirXML uses a set of XML rules to automate this process. The employee’s personal data need be entered only once into a directory and can be sent to all other relevant directories across the organization.
Novell chose to use XML as a rule language because of the flexibility it provides. For example, when attempting to match users across systems, scripting languages require the presence of a key field in each system. Novell has created an XML rule to match objects in a system where there is no one common data field.
Mr. Royal observed that this would require a centralized database that contains all possible data fields.
Mr. Esposito replied that that was true. Novell’s Directory Services builds a table of fields with which each system can communicate.
Dean Cutting (State Department) asked if NDS became the master copy of all data.
Mr. Esposito replied that while NDS is the connecting copy, it does not necessarily become the master copy. DirXML allows users to define Authority Data Sources in which an individual can control the data that enters the NDS database. For example, Novell’s work with the intelligence community has shown that the agencies are not inclined to trust one another with the master copies of their data. Thus, one person from each agency becomes the Authority Data Source and controls what data is available to the other agencies through DirXML. While other agencies can view this data, they have no control over it.
DirXML allows for data to be shared in systems with different schematic maps and methods of data representation.
Mr. Esposito addressed concerns about driver creation. There are two ways to write DirXML drivers. The user can utilize a wizard to create a driver for a common application/system or, if the user is familiar with XML, he can write his own driver in XML. The DirXML software comes with both capabilities. The Novell consultants, hired consultants (e.g. Deloitte, Andersen), or agency employees can write the drivers.
Mr. Crawford asked what the value of using XML in place of a scripting language was.
Mr. Esposito responded that defining an XML based script is simpler and that XML already has the basic ability to share data. It has already defined a set of rules as to how to share data. Otherwise, all users must agree on a common scheme that is less flexible and not as easy to edit "on the fly."
Mr. Crawford asked if Novell writes these rules as it creates new maps and if the rules are reusable across organizations.
Mr. Esposito replied that these rules are uniform enough to be applicable across systems as long as NDS is the central connection.
Mr. Cutting asked if there are any other products that can use DirXML scripting.
Mr. Esposito replied that currently there are no other products that can.
Mr. Crawford asked if a consumer could buy DirXML off the shelf, install it, and without any extra effort, connect to other systems.
Mr. Esposito responded that one could, so long as there is a driver that will connect the systems. The consumer must set up the drivers, as discussed previously. DirXML does have a series of default drivers that can be used with common programs (e.g. MS Excel) provided the customer does not have any custom code built into them.
According to Novell, the difference between DirXML and other applications is that it focuses on providing a directory service rather than a database. Directory services are designed to be read many times and at great speed. An example of when a directory service would be preferable to a database in situations is profiling browsers that arrive at a website. CNN.com uses DirXML to collect information about the browser while he surfs the site.
Mr. Royal asked if DirXML uses chains and referrals.
Mr. Esposito replied that it does not. In many cases, DirXML sends out data calls to other directories.
Mr. Royal observed that one of the best uses of this approach is to categorize PKI certificates, because different agencies use different kinds of certificates with different extensions and storage methods.
Mr. Esposito pointed out that DirXML is not meant to be the "be-all, end-all" solution. It is a very specific tool designed to allow directory-based sharing of information between systems. It works by routing data through a directory rather than directly linking two databases.
Mr. Cutting wished to know how DirXML sets up host applications to respond to events.
Mr. Esposito responded that event translation is defined in the drivers. In writing a driver, the user could leverage an event system already built into the native system (e.g. using a Netscape log). Some systems, such as Lotus Notes, do not have event engines, so the driver must write a program for it.
Mr. Ambur asked if Novell has had any contact with the CIO Council’s Directory Forum?
Mr. Mooney said that Novell and the CIO Council had made contact over an issue several months ago, but have not been working together.
Mr. Ambur observed that those personnel currently working on the creation of a government-wide white pages could make good use of DirXML. He wanted to know what parts of DirXML are proprietary and what parts are based on standards other than XML. Additionally, he asked if there are any other vendors supporting the concept of using XML with a directory. He wondered if Novell had any suggestions for issues the WG ought to consider in terms of adding to the value of DirXML.
Mr. Esposito replied that DirXML is not based on any standards other than XML. Novell is writing a custom driver directory that is itself standards based. The drivers have some proprietary information, but they support many open standards in directory access. With regards to vendor support, DirXML is a vendor-based solution. As such, you will never see Microsoft come out with a DirXML program. Additionally, Novell does not have to set up DirXML for the customer. The customer can use other consultants or its own employees to do so. Additionally, the customer does not need to find or create a common schema amongst its systems to make DirXML work. While NDS is needed to run DirXML, NDS is a cross-platform directory. As a result, customers do not need to be running the Netware platform to use DirXML. It will run off of many platforms (e.g. Linux, Windows, etc…).
Mr. Crawford asked if DirXML has ever been used across enterprises.
Mr. Esposito responded that cross-enterprise exchange could take place, provided both organizations are using NDS. In this situation, the two NDS directories will communicate.
Mr. Crawford asked if a single NDS could integrate two systems.
Mr. Esposito replied that the intelligence community is currently working on integrating its systems through one NDS directory.
Susan Turnbull (GSA) asked if the DirXML drivers are robust enough to accommodate all versions of various systems and applications.
Mr. Esposito replied that the current drivers are based on the current versions of the systems. If a version release of a system or application differs greatly from the previous one, then its associated driver will have to change.
Mr. Ambur requested that Novell advise the WG as to how it could best promote the use of XML for the kind of data sharing DirXML is involved in.
Mr. Mooney thanked the WG for its time, and directed members to go to http://www.novell.com/ products/nds/dirxml/ for more information on DirXML.
Due to time constraints, the WG agreed to skip its traditional mid-meeting break and continue on to the next item on the agenda.
Discussion of XML.gov Requirements and Goals (10:45)
Mr. Ambur began the discussion by stating that he believes the XML.gov website may be the WG’s most important product, especially its proposed registry. He invited all members to comment upon the site prototype.
Mr. Royal encouraged members to feel free to comment upon the current XML.gov Requirements statement (handed out to all WG members at the beginning of the meeting). The handout also contains the expectations for Phase Two of the XML.gov project. If anyone has any comments or ideas after the meeting, please e-mail him or post them to the listserv. He added that due to limited time and funding on the GSA’s part, the XML.gov team is hoping to contract out for professional assistance and server capability. As such, the requirements statement would be transmuted into a Statement of Work.
Bill LaPlant (Census) asked if XML.gov referenced any standard for metadata registries.
Tim Boland (NIST) replied that the team is working with one of his colleagues at NIST.
Terry Gower (DLA) asked if the website is using the Bobby Approved icon.
Mr. LaPlant suggested that XML.gov use the Web Accessible Informance (WAI) Conformance Level A icon. He feels that it is more robust and that Bobby is not up-to-date with the lest versions of html. Additionally, Bobby is intended to be a guide and educational tool. It is not the most robust indicator of compliance with accessibility statutes.
Mr. Crawford suggested adding a conformance clause or section to the XML.gov requirements.
Ms. Turnbull suggested working with other groups who are addressing conformance issues.
Bruce Bailey (Education) suggested using XHTML for XML.gov, since the point of the site is to promote XML. Additionally, XHTML allows for many different methods of presenting information, thereby making the site more accessible.
Ms. Turnbull informed the WG of the existence of the Human-Computer Information Management Coordination group. This group’s research focuses on universal access with a strong emphasis on XML. Ms. Turnbull could serve as a conduit to that group. Additionally, the group could be a source of funding for the WG or XML.gov. She suggested the WG host a presentation of the group’s work at its January meeting.
Dennis Drinkard of Broadvision asked when XML.gov would be up and running.
Mr. Royal replied that the team wants it up as soon as possible. However, the GSA cannot afford to pay for the entire site. There are still many outstanding issues regarding partnerships and funding. While GSA could host the site, Mr. Royal feels that perhaps hosting ought to be part of the work sent to a contractor (along with site maintenance, development, and consultation).
Mr. Ambur forwarded some possible uses of the registry. First, it could support information collection and reduction under the Paperwork Reduction Act. His discussions with staff-level OMB personnel have encouraged him to propose to the CIO Council that agencies use the XML.gov registry for data registration rather than the lengthy OMB paper-based filing process. He asked for the WG’s opinion of this idea.
Mr. LaPlant worried that if agencies were required to pay for their use of the registry, some would bear a greater burden than others would. For example, the Census Bureau files as many as 200 quarterly requests for data collection with the OMB.
Mr. Ambur asked if each of the Bureau’s requests was for a new set of data, or if it had to file again to collect new data for the same data elements.
Mr. LaPlant replied that the Census creates new surveys on a continuing basis, and with each one the Bureau must file a new request for OMB approval.
Mr. Ambur pointed out that by maintaining a data registry, XML.gov could make this process more efficient.
Mr. Niemann stated that FedStats.net is working with the Census to include its statistics on FedStats.net.
Mr. Royal added that the group working on the presidential directive to make 500 public forms available on-line has collected over 1000 forms and is looking for a place to host its registry.
Mr. Niemann observed that the National Archives and Records Administration (NARA) has begun working on a database of government forms and documents.
Dan Jansen (NARA) announced that NARA has begun its Electronic Records Archive Program. The program focuses on ingesting, storing, and providing access to government forms. Currently, the program is still in its reassert and demo phase and has no active projects employing XML. One problem the program faces is that several decades worth of documents lack any standard format. NARA is taking possession of tons of data and investigating a scalable, flexible system. He suggested giving the WG a full-blown presentation of the program.
Mr. Ambur suggested that the WG put together a business case for funding the use of a data registry at XML.gov to present to the CIO Council.
Mr. Royal recommended looking for funding opportunities in other electronic government committees. Working with the Census Bureau could give the WG an idea of how data administration works at the agency level.
Mr. Cutting asked if there is any legislation urging government to discuss how data is stored and shared across agencies.
Mr. Ambur replied that the Workflow Management Coalition (WfMC) is addressing standards for workflow across organizational boundaries, including XML specifications.
Mr. Crawford added that the Federal Commons is addressing these issues with respect to grants. It is the only such group that he is aware of.
Mr. Ambur reminded the group that it hosted a presentation on Workflow XML (geared towards the DLA) several months ago.
Mr. LaPlant suggested that the Census Bureau’s National Processing Center (the Kinko’s of federal survey processing) could be a useful source of information on form processing.
Ms. Turnbull noted that WG members were suggesting uses for XML.gov that both science agencies and the CIO Council might be willing to support. The WG ought to try to draw from both "funding pots."
Mr. Ambur asked when the WG ought to have its budget request submitted.
Mr. Royal responded that he has learned the WG will have fifty thousand dollars for work on XML.gov for the year 2001. This added to the one hundred thousand left over from this past year should be enough.
Mr. Ambur asked if there was a deadline the WG ought to meet.
Mr. Royal replied that the WG needs to submit a request only if it wants substantially more than the fifty thousand already set aside.
Ms. Turnbull noted that the Knowledge Management Working Group is looking to support three different knowledge management communities. The WG should consider becoming one of these communities; XML.gov could serve as a knowledge repository. This would provide more funds. She will speak to the KM WG.
Mr. Royal pointed out that the WG must identify XML.gov requirements before it can project its costs.
Mr. Niemann suggested that the WG consider the FedStats model. After OMB mandated FedStats’ creation, a task force wrote out a budget, and was able to draw support from many agencies. The WG should try to get a mandate from OMB that will require each agency to provide some financial support. True, this would mean the WG would have to report its progress to agencies heads, but in his opinion, it would be worth the effort. The WG has a level connection through the CIOs already, so it ought to take the initiative to put together a plan and budget to push through the Council.
Ms. Turnbull stated that the WG should also try to work with its counterparts in State governments. The Federal Government needs to be interoperable with the States.
Mr. Royal asked if the WG should be the leader for redefining data gathering processes.
Mr. Ambur responded that he believed the WG should be a catalyst for change. It can suggest the use of standards compliant with XML.gov XML registry. He has gotten some positive feedback from the staff at OMB about XML.gov’s potential role in the data gathering process.
Mr. LaPlant warned that there may be a good deal of agency resistance to the idea of making these processes visible to the outside world.
Mr. Ambur felt that it is in the interest of the taxpayer to provide visibility. The WG should not necessarily mandate this, just facilitate it.
Mr. Cutting expressed the belief that due to the emergence of the web as an instant source of information, taxpayers will soon demand visibility and instant access.
There was a suggestion that the WG not tie its work too closely to the Paperwork Reduction Act, as the Act has become a very sensitive subject in some agencies.
Mr. Royal urged the group to e-mail either Mr. Ambur or himself with any related suggestions. He also urged members to post their ideas on the listserv.
Next Meeting: January 17. The December 20 meeting is cancelled.
XML Working Group Attendance List
November 15, 2000
|Kevin Campbell||XML Solutions|
|William Selfridge||Data Networks Corporation|
|Mike Sinisgalli||XML Solutions|
|Jeff Todd||Infoteria Corporation|