IBM mainframes

Computer platforms where 1 represents IBM PCs and compatibles, 386 or better 2, Unix-based work stations (or equivalent) including HP/ApoUo, SUN, and IBM RS6000 3, DEC VMS 4, IBM mainframe, MVS or VM, or compatible and 5, others including Cray, Data Gen., etc. 

MIMD Multicomputers. Probably the most widely available parallel computers are the shared-memory multiprocessor MIMD machines. Examples include the multiprocessor vector supercomputers, IBM mainframes, VAX minicomputers. Convex and AUiant rninisupercomputers, and SiUcon... 

Statistical analysis using spreadsheet format experimental design. Macintosh. Anonymous ftp at ftp.sas.com. SAS statistical and data management system on IBM mainframes, VAX, PCs (Windows), Sun, HP, and IBM UNIX workstations. 

A special type of network architecture that deserves mention is the Attached Resource Computer Network (ARCNet). Developed in 1977, it was not based on any existing IEEE 802 model. However, ARCNet is important to mention because of its ties to IBM mainframe networks and also because of its popularity. Its popularity comes from its flexibility and price. It is flexible because its cabling uses large trunks and physical star configurations, so if a cable comes loose or is disconnected, the network will not fail. Additionally, since it used cheap, coaxial cable, networks could be installed fairly cheaply. 

All three programs store the input information on floppy disk. Most of the input is done by external contributors who send the disks in certain time intervals to our institute for further processing. The data of the different input sources are uploaded into our IBM mainframe computer, converted to the same file format and stored in an intermediate file. After several additional automatic and manual checks the data are loaded under the database management system ADABAS (Figure 4). 

The DEC VAX series of computers has become dominant for use with chemical information and molecular modelUng systems. IBM mainframes are used for these applications in a number of companies however, due mainly to historical corporate influences. 

ORACLE from Oracle Corporation and INGRES from Relational Technology Inc. are two popular relational systems for the DEC VAX series of computers. SQL/DS and DB2 from IBM are popular for IBM mainframes. A large number of relational systems are available for the... 

BASIS - Text management system, hardware supported IBM mainframe, DEC VAX, etc., available from Information Dimensions, Inc., 655 Metro Place, Dublin, OH43017, U.S.A. 

MACCS — Chemical information system for in-house use, hardware supported DECVAX, IBM mainframe, and Prime, available from Molecular Design Limited, 2132 Farallon Drive, San Leandro, CA 94577, U.S.A. 

RS/1 - Data analysis system, hardware supported DEC VAX, IBM mainframe PC, etc., available from BBN Software Products Corp., 10 Fawcett St., Cambridge, MA 02238, U.S.A. 

The corporate file is currently run on an IBM mainframe and data are stored in Chemical Abstracts Service Registry-II format. An in-house developed graphical interface program allows searching in a very friendly manner. The technology of this software package will be replaced in the near future. This file currently contains approximately 500,000 compounds. 

The tree-based search technology can offer very good performance on large files for a broad range of computers from IBM-PC to IBM mainframes. 

On an IBM mainframe (3090/150) the average retrieval time for a substructure was under 10 seconds on a database of 1,200,000 compounds. The effect of database size on retrieval time has been studied and the results are very encouraging increasing the size of the database from 150,000 to 1,200,000 resulted in an increase in retrieval time of only 50 per cent. 

An RDP value of 30 corresponds to a computing power of nearly 30 MIPS which exceeds that of an IBM 3090-150 mainframe by a factor of at least 3 for a price which is only a small fraction of an IBM mainframe. 

The structure query input and query compilation functions were separated from the actual searching function. Query compilation converts the structure input representation as specified by a user into the screen and connection table forms needed by the screen and iterative search processes for the actual searching function. Query negotiation and query compilation were performed on the IBM mainframe computer to take advantage of already existing software. Also, these functions are not computeintensive and so were not an appreciable factor when considering the relative cost of computing resources. 

The control for an individual query is handled by an IBM mainframe process. This process broadcasts the query to all iterative processors and then waits until all of the iterative processors acknowledge that the query has been started. To provide that acknowledgement, the iterative process has to propagate the query on to the screen processor and receive an acknowledgment from it. 

Over the lifetime of the system the response time has remained at a constant five to six minutes per search although usage has continued to increase steadily. The cost of the computing equipment has generally dropped, especially as it has become available on the used equipment market. The architecture has proven to be readily operable. Availability and reliability have been generally comparable to the CAS IBM mainframe environment - currently availability is in the area of 99 per cent for the search machines. Some tuning and operational enhancements were implemented in the early years, but such activity has subsided considerably in recent years. 

The networking software capabilities have also been upgraded considerably to support complete process-to-process communication among IBM mainframe processes and UNIX processes using a local area network. Use of a high-speed local area network permits different design options. 

In 1986 a cost study was conducted to compare the cost of searching the CAS Registry File on a parallel bank of Motorola 68020-based minicomputers versus the cost of searching the file on an IBM mainframe. The cumulative costs over a five year period were noticeably higher on the mainframe if response times comparable to the five minute search machine response times were desired. If response times were reduced from about five minutes to about one minute for the average search, then cumulative mainframe costs were more than twice as great as the minicomputer costs. 

---