Home > Publications > Newsletters

Are Repositories Ready for the Internet ?

Thanks to Data Warehousing, Internet technology, and Object Oriented programming, new types of software objects are being invented every day. As systems become more distributed and use more layers of software, the need to track and manage metadata grows larger.

Repository and tool vendors are moving to take advantage of the recent Internet technology. However, because repositories can hold many different types of information, it is difficult for the vendors to provide a generic view for the Internet. All of the repository products I have seen so far require some degree of custom programming to make the repository information available to the Web.

Most of the technology to web enable a repository is not created by the repository vendors. Some vendors may include templates or sample scripts for custom Internet programming in their product. Most leave it to the customer to add on the desired tools and develop custom programs.

Generally speaking, repository information can be programmatically accessed in several ways. Some repository vendors provide API calls or object class libraries that can be called from a custom written program. Others use OLE (Object Linking and Embedding) calls for communicating between programs. With these methods the vendors control the manner and view of the information retrieved from the repository. If you have extended the repository with custom elements, you may not be able to see them in a query.

If source code program calls are not provided by the vendor, you may be able to access the repository directly at the database level. However I would only attempt this approach if the repository vendor encourages it. It could be difficult to navigate a complex repository data model without vendor guidance.

If a repository is implemented using a relational database management system (RDBMS), metadata information can be accessed through SQL calls in a custom program. The program must navigate the internal structure of the repository and retrieve the desired information from the appropriate tables. This approach provides great flexibility but requires complex programming.

Some repositories are implemented using an object database management system (ODBMS). The repository vendor usually provides an object class library that can be accessed by custom programs written in languages like C++. The object class library contains functions that the program uses to query the repository.

If your custom program is written in an object oriented style, you can still access the relational database. There are some object-to-relational mapping programs available that convert object oriented database queries into SQL calls for a relational database.

Once metadata information is retrieved from a repository, the custom program must format the information and send it to a web server process using CGI (Common Gateway Interface) or other scripting languages. The web server process then transmits an HTML page through the Internet to the requester’s browser.

Some features to watch for include support for Java, Active X, C++, and CORBA/ DCOM. As the technology improves the programming task of enabling a repository for the Internet should get easier. However the issues of controlling user access, restricting repository views, and granting update privileges need to be addressed by the repository customer.

The following is a quick tour of some repository products and their current Internet capabilities.

Repository Vendors

Platinum
The Platinum Open Enterprise Edition (OEE) repository is a three tiered client / server based product. The repository can be implemented on an Oracle or Sybase platform. Although no direct API calls are available, the repository can be access directly using SQL.

A separate product called Data Shopper for the Web is offered through Platinum’s consulting group. The Data Shopper can be customized to provide access to Platinum’s PR/MVS mainframe repository through an ODBC connection. The web server portion of Data Shopper runs on a Windows NT Internet Information Server (IIS) platform. The product comes with some CGI scripts and HTML templates that must be customized. Future releases will provide access to Platinum’s client / server repository.

Data Shopper for the Web can only support one SQL query per HTML page. This means that some repository requests that need multiple queries may not be supported with this product. Data Shopper for the Web only allows ‘read’ access to the repository. Access to Data Shopper is controlled using Windows NT security.

TranStar
The TranStar repository product is implemented using a proprietary ‘semantic’ DBMS and can be accessed using C language API calls. With the API calls you can query and update the repository. The information in the semantic database is organized by topics. The semantic architecture provides great flexibility by allows topics that relate to each other to be linked together in numerous relationships. However in order to follow a thread of topics, a calling program must make a separate query for each link. Metadata from the repository can also be sub-setted and downloaded into a Sybase SQL Anywhere or a Microsoft Access database for individual use.

UNISYS
The Universal Repository (UREP) tool is built on the Versant ODBMS. The repository is accessible using the provided object class library or through OLE calls. Because the repository is so universal, UNISYS has left it to the customer to define the scope of objects to be browsed and the method of browsing. UNISYS provides a sample generalized meta model browser for HTML but it is primarily for demonstration purposes. The Adaptive Solutions toolset integrated with the Universal Repository comes with a Windows based repository browser.

Viasoft (Rochade)
The R&O repository is an independent product that runs on either a mainframe or a client server platform. It integrates with a wide variety of development tools and supports most database engines. The repository can store many types of documents including text, spreadsheet, sound, video, image, and HTML. Repository functionality can be accessed by other programs through a set of API calls and object classes known as Autopilot. The R&O repository can be web enabled by writing a custom program using the Autopilot API calls. One customer has already developed a web front end for the repository with browse and update capability.

MSP
MSP offers a mainframe and a client / server repository product. The client server repository is implemented using one of several relational database engines. The repository is accessible using SQL and predefined table views that come with the product. The repository metamodel is based on the IRDS standard. A Windows GUI front end provides user access to the repository. A web enabling custom program could be written to access the repository directly using the SQL views