Is the core enterprise processing role handled by the mainframe enough? Yet, enterprises today often are running different types of workloads built using different app dev styles. These consist of compound applications encompassing the mainframe and a variety of distributed systems (Linux, UNIX, Windows) and different programming models, data schema, services, and more. Pieces of these workloads may be running on the public cloud, a partner’s private cloud, and a host of other servers. The pieces are pulled together at runtime to support the particular workload. Mainframe shops should want to play a big role in this game too.
“Mainframe applications still sit at heart of enterprise operations, but mainframe managers also want to take advantage of these applications in new ways,” says Brent Carlson, SVP at SOA Software. The primary way of doing this is through SOA services, and mainframes have been playing in the SOA arena for years. But it has never been as seamless, easy, and flexible as it should. And as social and mobile and other new types of workloads get added to the services mix, the initial mainframe SOA approach has started to show its age. (Over the years, DancingDinosaur has written considerably on mainframe SOA and done numerous SOA studies.)
That’s why DancingDinosaur welcomes SOA Software’s Lifecycle Manager to the mainframe party. It enables what the company calls a “RESTful Mainframe,” through governance of REST APIs that front zOS-based web services. This amounts to a unified platform from a governance perspective to manage both APIs as well as existing SOA assets. As Carlson explained: applying development governance to mainframe assets helps mainframe shops overcome the architectural challenges inherent in bringing legacy systems into the new API economy, where mobile apps need rapid, agile access to backend systems.
The company is aiming to make Lifecycle Manager into the system-of-record for all enterprise assets including mainframe-based SOAP services and RESTful APIs that expose legacy software functionality. The promise: seamless access to service discovery and impact analysis whether on mainframe, distributed systems, or partner systems. Both architects and developers should be able to map dependencies between APIs and mainframe assets at the development stage and manage those APIs across their full lifecycles.
Lifecycle Manager integrates with SOA’s Policy Manager to work either top down or bottom up. The top down approach relies on a service wrapping of existing mainframe programs. Think of this as the WSDL first approach to designing web services and then developing programs on mainframe to implement it. The bottom up approach starts with the copy book. Either way, it is automated and intended to be seamless. It also promises to guide services developers on best practices like encryption, assign and enforce correct policies, and more.
“Our point: automate whatever we can, and guide developers into good practices,” said Carlson. In the process, it simplifies the task of exposing mainframe capabilities to a broader set of applications while not interfering with mainframe developers. To distributed developers the mainframe is just another service endpoint that is accessed as a service or API. Nobody has to learn new things; it’s just a browser-based IDE using copy books.
For performance, the Lifecycle Manager-based runtime environment is written in assembler, which makes it fast while minimizing MIPS consumption. It also comes with the browser-based IDE, copybook tool, and import mappings.
The initial adopters have come from financial services and the airlines. The expectation is that usage will expand beyond that as mainframe shops and distributed developers seek to leverage core mainframe code for a growing array of workloads that weren’t on anybody’s radar screen even a few years ago.
There are other ways to do this on the mainframe, starting with basic SOA and web services tools and protocols, like WSDL. Many mainframe SOA efforts leverage CICS, and IBM offers additional tools, most recently SoftLayer, that address the new app dev styles.
This is healthy for mainframe data centers. If nothing else SOA- and API-driven services workloads that include the mainframe help lower the cost per workload of the mainframe. It also puts the mainframe at the center of today’s IT action.
Follow DancingDinosaur on Twitter: @mainframeblog