Selection criteria for evaluating ESBs for use in a mainframe-oriented environment
This paper looks at the selection criteria to use when evaluating ESBs (Enterprise Service Bus) for use in a mainframe-oriented environment.
As service-oriented architectures (SOAs) have cemented themselves at the top of the list of business and IT concerns for many companies, attention has become focused on achieving the promised benefits quickly and effectively while at the same time maintaining attractive levels of return on investment. SOAs promise a high degree of business flexibility and agility, and offer considerable opportunity for reuse of application components. IT becomes more closely aligned to business goals and strategies, and improved visibility of IT-based operations in a business context enable more effective and optimised business solutions.
One key component of many SOA strategies has become the Enterprise Service Bus (ESB). ESBs are rapidly establishing themselves as flexible, productive and robust standards-based solutions to many integration needs, and offer a strong foundation for SOA implementation. But if the ESB is to become such a critical element of operations, then vendor and product selection becomes a decision of major importance. Making the wrong choice is likely to seriously impact the pace and ultimate success of business integration and SOA initiatives. This selection choice was discussed in general terms in the Saint Consulting report entitled ‘Best of Breed ESBs’, a generic buyer’s guide to help companies choose the most appropriate ESB for their needs, available from the Integration Consortium website, www.integrationconsortium.org.
While the referenced guide has been reported to have been useful to many potential ESB buyers, it was intended to be of general use in a wide variety of situations and as such does not attempt to deal with specific issues related to specialised environments. One of the most specialised of these is that of the mainframe user – companies which have been heavily immersed in the world of IBM mainframes for years. Mainframes are architectural very different from the workstation and server-based worlds, and foster a culture that may seem alien to the non-mainframe world. For example, mainframe-oriented users are accustomed to high levels of availability, reliability and performance. Recoverability is assumed at all levels, as is transactional integrity. These mainframe factors tend to permeate all software buying decisions whether destined for the mainframe itself or elsewhere in the operational installation.
But, although mainframe-oriented users may be different, SOA and the use of ESBs still appeals greatly to this audience. To some, an ESB-based SOA provides an ideal migration environment, where business applications can be moved off the mainframe piece by piece until the mainframe itself can be replaced. However, to the majority these technologies promise a means to leverage mainframe investments by making mainframe transactions available enterprise-wide, while at the same time enabling mainframe applications to benefit from distributed investments.
This paper discusses ESBs in the light of these specialised requirements of mainframe-oriented users, and provides guidance and checklists on what functionality might be necessary for a successful ESB deployment in this environment. It does not contradict the generic guide but instead builds on it, reinforcing some of the points already made there and adding to them where necessary.