Job scheduling across multiple platforms

Dit artikel delen:

There are a lot of opportunities to introduce new technologies into IT systems at the moment. The Internet and e-commerce have not progressed as fast as many predicted (what's new?), but there has been some involvement in most organisations.

This has enforced an exposure to external communications networks and to Web based user interaction, which in most cases is new territory. The newer requirements may be either independent systems or new front-ends, integrated with older systems; in either case it follows that some new technology is inevitably involved. In some cases this may be new software with existing platforms, but it most likely involves new hardware and operating systems. This is a wonderful opportunity to get experience with Linux and other Open Source Software (OSS), Apache in particular.
One of the biggest choices to make is which technology to adopt to develop and run the Application Servers which integrate with legacy systems and the Appliance Servers which handle the users and their Web browsers. Today there are basically two options, Microsoft IIS and its associated products such as Active Server Pages (ASP), or Java based systems. The latter run on any platform, including Windows, while the Microsoft products tie the user into Windows for the foreseeable future. From a personal standpoint, being tied into Microsoft for servers after the bitter experiences of the Office monopoly is unacceptable, which eliminates IIS, etc. Being tied into Java based solutions as a result of eliminating Microsoft is not so bad because there are quite a few suppliers of such Application Servers. It would be even better if the Java standards were managed by W3C or a similar body, divorced from Sun Microsystems. We must be eternally grateful to Sun for giving us Java, but the future must be controlled by independent bodies.
Before Java matured with the release of J2EE there were some excellent products developed, such as Cold Fusion, Apptivity, SilverStream, etc. Today these products must have matured with Java or else they are unacceptable for the future. They must vie for market share with the leaders, BEA's WebLogic and IBM's WebSphere. Oracles's integrated products must also come into consideration. All these products are available on multiple operating systems and hardware platforms. In some cases these servers can be added to existing systems, e.g. MVS users can run WebSphere on an existing mainframe, but most organisations will finish up with a Unix machine or an Intel-based server with NT or Linux. It is likely that most organisations will already have experience with a variety of systems, but the key difference now is that these systems are integrated into one application and with that comes an increased need for management tools that span all the systems involved.
There has been a lot of investment over the past few years in management software; to realise the significance of overall management simply note that IBM saw fit to buy Tivoli so as to be in the driving seat. But as well as systems management the integration of various subsystems into a common applications system emphasises the need to look further and to consider automation of functions such as directory services, help desk, data transfer, etc. Workflow management has been important for many years although I feel that there is still little attention paid to automation tools, but an even more common need is the management of batch processes. Since mainframe systems are much older than Unix or PC systems, they have a serious history of batch processing and because of that scheduling tools, in particular JESS on IBM mainframes, are very powerful and well used. In contrast Unix, VMS, etc. are essentially multi-user, interactive systems, they have rather weak batch capabilities and thus have relatively simple (and frankly inadequate) job scheduling tools.
This is going to have to change with all these new integrated applications, because the scheduling and management tools must work across multiple platforms. It is not enough to do the mainframe part right and mess up on the other systems. One solution lies in IBM's hands by enhancing JESS and Tivoli products to extend beyond z/OS. The other approach is to use independent suppliers such as Cybermation who traditionally look beyond the mainframe and have multi-platform support. In any case something separate is needed for those users without mainframes!
It is still amazing how little attention has been given to Batch processing and job scheduling in particular, outside the mainframe arena. The e-commerce initiatives should provide the incentive to put that right.


Om te kunnen beoordelen moet u ingelogd zijn:

Dit artikel delen:

Stuur dit artikel door

Uw naam ontbreekt
Uw e-mailadres ontbreekt
De naam van de ontvanger ontbreekt
Het e-mailadres van de ontvanger ontbreekt

article 2001-12-07T00:00:00.000Z Martin Healey
Wilt u dagelijks op de hoogte worden gehouden van het laatste ict-nieuws, achtergronden en opinie?
Abonneer uzelf op onze gratis nieuwsbrief.