Huron, the new applications system environment announced last week by Amdahl Corp, will have the same effect on the world of software applications as the original 470 IBMulator had on the mainframe world back in 1976, claims the Sunnyvale manufacturer. Under development for 10 years, the high-level interface for programming and data access is said to have three primary objectives. It has been designed so that it can be changed easily; it co-exists with old data and programs; and it tells a computer why and how to do things without being just another fourth generation language. Hyped as Amdahl’s answer to the still incomplete IBM AD/Cycle, it is also not dissimilar to Cincom System’s AD/Advantage. That offers object technology in the form of re-usable logic components, while Amdahl’s offering can re-use 70% of code, although the Huron database is much more complete since AD/Advantage takes information only from DB2 and Supra catalogues.
Kernel in C
The Huron kernel is written in C and runs under MVS, but Amdahl says that UTS and Unix workstation versions will be available and the system will operate in all environments without alteration. Amdahl emphasises that although Huron does have some object code, it is a strictly confined list, and the system is not object-oriented, unlike the rule-based object oriented application development system for IBM mainframes developed and marketed by Sapiens International (CI No 1,546). Huron has been installed in 17 beta sites worldwide, mostly in the US and Canada, but there are three in the UK, one of which is the Inland Revenue. Although it is available now, sales are to be limited to areas of the US, Canada, Europe, and the Pacific Basin where support personnel are already in place, and installations will be confined to IBM-compatible, large-scale systems. The company acknowledges that Huron will require a type and level of support different to that demanded by mainframes, and it is now recruiting applications system engineers with expertise in databases, data modelling and computer aided software engineering. Initial installation fees for a single licence will range from $525,000 to $1.4m depending on the size of systems involved. At the heart of Huron is the integration of a central on-line MetaStor facility with the physical data storage system and the programming environment. The MetaStor is integrated with an extended relational database; a high-level rules language and a set of workbench tools to define databases and to write and execute transactions in test and production. All tools are built in Huron using the Huron interface, and all transactions obey a two-phase commit protocol.
By Janice McGinn
Applications are highly data-driven since application objects are separated from data objects. The MetaStor maintains and controls definitions and links between objects, and whenever an object is changed, the MetaStor reflects that change whenever the object is used. It dynamically selects access paths and accesses objects at the moment of execution, and it permits synchronous read-write access to distributed data in DB2, IDMS, IMS, VSAM and Model 204. The Huron Rules Language consists of 18 statements, and the rules are claimed to be both easy to learn and suitable for all phases of application development and maintenance. They are created using a context-sensitive rules editor, and these rules, which are modular, conform to principles of structured programing. Impact analysis tools take account of the effect of changes before they are made. Set oriented data access statements are integrated into the language, and both data manipulation and processing algorithms are applied to multiple logical records. The extended relational model used for the native database supports identifiers, semantic definitions of data elements and event-driven processing. Data conversions, set operations, selections, sorting, defaulting, auditing, database validation and database integrity are carried out at data level. The database supports automatic key generation for data records, and to speed up
random accessing in large databases, there is an hashed index scheme to permit extensions to hashed space allocation as an alternative to the B-tree. Database operations are continuous, and journals are merged concurrently with on-line operations. Amdahl says that interruptions are not usually required for maintenance, and to separate or protect critical systems, the Huron store may be segmented, and the MetaStor facility is duplexed. The database system defines data relationally and hierarchically, while extensions to the relational data model are said to support the self-defining nature of Huron as well as the definition of other data. The database system enables developers to work with the logic of the application, and to re-use rules and other system objects as blocks for building new applications.
Consistency
Event-driven application execution provides a mechanism for maintaining database consistency, and when applications are being written, the MetaStor and data access method deal with triggers, validations, derived fields, required fields, default fields, conversions, set operations, sorting, selections and referential integrity. Amdahl says that applications developed in Huron are data-driven to an extent not possible with conventional systems since application rules are separated from the variable data which is normally embedded in application programmes. Tables are defined and referenced from application programs, and the data can be modified to reflect new requirements independent of the application programs. Also, as new applications come on-line, data can be migrated from existing databases into the native environment. The Workbench provides a single look-and-feel for specification, development, design and testing of data, programmes, screens and reports. All tools interface with the MetaStor and are built in the system itself. The list of tools includes a context-sensitive program editor; a full-screen data browser and editor; a single record data editor; a screen definer and painter; report definer and painter; query processor; table definer; text editor and script processor; object cross-reference and search tools; security manager; calendar and appointment book; and a problem and reporting tool. The Workbench supports ad-hoc processing features such as on-line query language, and the company says that users can access existing databasesto browse and query non-relational files in a relational manner.
