Want to peel yourself away from VMware, mainframes, or ditch legacy .NET applications? There’s now a whole bunch of AI agents to help.

So says AWS, which this week launched “AWS Transform.”

That’s a newly generally available (GA) offering for “large-scale migration and modernization of .NET, mainframe, and VMware workloads.” 

AWS made bold claims about the new proposition, including its ability to deliver: “4x faster .NET application porting from Windows to Linux” and “80x faster conversion of VMware network configurations to AWS.” 

GenAI is making this easier… 

Efforts to prise applications away from rivals and/or modernise deeply embedded and complex “legacy” software has a long history of course. 

But generative AI is making it meaningfully easier and cheaper, say a growing number of technology providers – which continue to make big claims on this front and try to demonstrate complex real-world examples.

Amazon had earlier touted its “AWS Mainframe Modernization” offering which “enables… replatforming and automated refactoring” back at its 2021 re:Invent conference. (Our view at the time? "It appears to rely quite heavily on the well established Micro Focus enterprise analyser, albeit bundling that into a suite of integrated tools spanning initial planning to post-migration cloud operations...")

More recently, in 2023 CEO Andy Jassy praised its “Q Transform” toolkit for AI-powered Java application modernisation.) 

But AWS Transform packs more AI trained on more migration data.

Agentic fabulousness?

It deploys “specialized AI agents to automate complex tasks like assessments, code analysis, refactoring, decomposition, dependency mapping, validation, and transformation planning” to speed up and automate complex modernisation and cloud migration efforts, AWS said. 

The cloud behemoth claims to be snacking on its own dogfood. 

“For VMware, in our internal tests, we used AWS Transform to translate VMware network configurations for 500 virtual machines to generate AWS networking configurations such as VPCs, subnets, transit gateways and internet gateways within one hour. This is 80X faster compared to the two work weeks taken with traditional, manual approaches!”

That’s according to AWS’s newly titled “VP, Agentic AI”, Swami Sivasubramanian, posting on LinkedIn this Thursday. (He oversees the hyperscaler’s database, analytics and machine learning services.)

"Non-technical employees"?

Sivasubramania added: “What’s great about AWS Transform is that it’s accessible to both technical and non-technical employees”. 

(Readers wondering what kind of CIO is going to point AI-armed non-technical employees at some of their most mission-critical and complex enterprise applications, don’t be so blooming cynical.)

AWS has brought in some of the big beasts of the systems integrator jungle to help customers consider making this shift using AWS Transform. It names Accenture, Capgemini, DXC Technology, HCL, Infosys, NRI, and Pega as the partners quote, “driving transformation, together.”

Unravelling business logic

One of the biggest challenges for organisations looking to modernise and migrate complex “legacy” applications is, of course, tortuous relationships between application and business logic as well as complex dependencies.

Accenture Associate Director Guru Rao described the problem tidily to The Stack late last year, as he detailed efforts to move a bank’s trade and risk applications off Sybase.

That popular and perfomant but rigid and ageing database was easy "for adding stored procedures seamlessly” (Stored procedures are, in brief, precompiled subroutines used to add new functions and consolidate application logic.) 

“For each [new] function, developers started adding one stored procedure. ‘Here’s a function I want to achieve with this application: add a stored procedure – and business logic inside the stored procedure.’”

Untangling this typically has involved first understanding the business logic and then rewriting each stored procedure in a way which is easily comprehensible and easily migrated to the application layer, he added.

As Shiv Pullepu, MongoDB Industry Principal, Financial Services, added in the same conversation “it's common for any business-critical application to require thousands of stored procedures underneath it [and] become a humongous monster. It's stable – but it's not scalable for the future.”

Anyone looking to explore the kind of migration that AWS Transform is selling will, of course, want to weigh up existing on-prem costs, hardware and software lifecycles, the cost to migrate to their new target platform, the benefits of the latter, and the net year-on-year savings, as well its ability to help tackle this very real and still very manual process of understanding

But if AI can help both illuminate this trade-off and untangle the laborious process of tackling legacy applications and AWS's AI-powered efforts to run "code analysis, refactoring, decomposition, dependency mapping" are meaningfully useful, AWS will no doubt have no shortage of takers.

The link has been copied!