- Many early cloud movers went with a "lift and shift" approach to their migrations
- But this strategy has proved inefficient and expensive, consultants told us
- Now, enterprises are redoing their deployments with a focus on efficiency and a little help from AI
Call it a mulligan. Companies that were on the bleeding edge of the transition to the public cloud a few years ago are starting over again at square one, hoping to fix mistakes that have kept them from realizing the full benefits of running in the cloud, sources from consulting firms Deloitte and PwC told Fierce.
PwC's Cloud Engineering Partner Scott Petry and Deloitte Principal Akash Tayal said one of the biggest mistakes these companies made was going with a simple “lift and shift” model, basically recreating their existing infrastructure in the cloud as virtual machines rather than truly optimizing and reengineering things for the new environment.
The thought at the time was that just moving operations to the cloud would yield cost savings. However, "That turned out to be horribly inefficient,” Petry explained. “You get to the cloud and they start running the meter on your applications in a consumption model and it turns out to be a lot more expensive than buying the hardware yourself and running this stuff on it.”
Now, early movers are actually creating new cloud accounts – often with the same hyperscaler – for a fresh start. And this time, they’re paying close attention to things like consumption and waste management as well as “re-architecting their applications leveraging cloud native technologies to help drive a better cost profile and higher resiliency,” Tayal said.
Petry said some examples of fixes being made include moving from virtual machines to cloud-friendly containers, putting systems in place to turn off environments that are not in use and setting up proper CI/CD pipelines. Basically, all of the things they didn’t spend any time doing during their initial migration.
Petry said he expects “almost everybody” in the early cloud migration cohort will redo their initial deployments.
“The ones that really did it wrong are having to completely start over and open a new account and move it all again,” he said. “I’m sure there’s a handful here and there that they got it right the first time, but I would also argue that even those people have a continuous evolution ahead of them.”
AI influence
Both Petry and Tayal noted that the arrival of artificial intelligence (AI) has also changed how companies approach cloud engineering tasks.
“With Gen AI, companies are re-inventing how they do the software development lifecycle and migrate/build their applications in the cloud autonomously,” Tayal said. “Gen AI is also changing the way infrastructure and cloud operations work, leveraging automation at a whole new level.”
Petry noted AI also has the ability to deliver much-needed efficiency and cost savings gains. For example, he said that by using AI, companies are seeing as much as 40% savings in effort required for app development and 10% savings in creating SAP deployments.
“There’s a lot of potential savings,” he said, spanning every area from discovery, to project management to testing and writing code. “To me, it’s the biggest change in efficiency of the methodology that I’ve seen in a really long time. I’ve been doing software development for my whole career and this is the fastest I’ve seen it change.”