The first signs of cloud computing took the form of computationally intensive tasks. But some of these tasks closely resembled format transformations familiar in classic middleware. Now, iPaaS cloud computing architecture is starting to gain traction, although data and security issues still loom.
One of the first 'poster children' for cloud was the New York Times' use of Amazon's S3 in 2007 to convert 4 terabytes of TIFF data (an article archive going back to 1851) to PDF files for the Web. That required conversion familiar to experienced middleware hands, but the job was mostly about computation, not data integration, middleware messaging or application virtualization.
Since those early days, middleware integration has increasingly appeared as a cloud computing task – and challenge. For now, integration-centric cloud computing is called by a number of names. These include ''Infrastructure as a Service'' (''IaaS''), ''Platform as a Service'' (''PaaS'') or ''Integration Platform as a Service" (''iPaaS'').
"We normally speak about iPaaS to indicate a broad range of middleware services in the cloud,'' said Massimo Pezzini, analyst, Gartner. He said iPaaS is a set of development and runtime tools that you use for integrating data, application services and processes.
''It typically provides some governance capability,'' said Pezzini. Again, typically, it adheres to a pay-per-use model, he said, while combining a hybrid collection of data integration, enterprise service buses, managed file transfers and B2B gateways.
''Connecting enterprise applications is a typical use case today,'' he said.
Massimo explained that iPaaS cloud adoption is inhibited by lack of trust. Issues implementers face include security and data handling – which are sometimes interchangeable concerns.
The data-centric data center
From a data-centric point of view, cloud computing is often quite challenging. The data center is not going away overnight, and - in most cases - the data in the data center will reside on the private side of the public/private cloud hybrid divide.
"Integration with various data sources and 'middleware' is actually one of the most complicated things to achieve," said Christopher M. Moyer, vice president of technology, Newstex LLC, a content aggregator.
Data sources are prized assets, and not something corporations are ready to place ''in the ether.''
"Eventually I suspect things will get easier as more and more companies embrace the idea of cloud computing," said Moyer, author of the recent "Building Applications in the Cloud" (Addison-Wesley, 2011). "But until then, it's actually the most challenging thing about migrating to cloud computing."
Increasingly, enterprise managers were talking about infrastructure implementations on public and private cloud in terms of application virtualization, said Uri Cohen, vice president of product management, GigaSpace.
"It's fine to virtualize infrastructure but it only goes a certain length. If you are business unit and get infrastructure as a service, there are a lot of steps you have to take," he said.
Clearly, the provisioning of cloud middleware has been no less complex than the provisioning of middleware in the data center. Much of the early cloud effort has centered on automation middleware provisioning.
"People want to make a process where building a new application or updating an application is done in hours or days not months. If you have over 100 Web applications outbound and inbound, you want to virtualize all the apps. Each business owner has to deal with packaging the application and ''automagically'' launching [related] resources for load balancing, etcetera."
Where is it good to use iPaaS cloud integration now? Gartner's Pezzini points to ERP-SaaS integrations, B2B integration and federated SOA integrations as likely candidates. "If you need to handle a lot of messages and have complex transforms, iPaaS might not be the best today," he told an audience at last month's Gartner AADI Summit 2011 in Las Vegas.
What are your thoughts? Email us and let us know.