Divanny Lamas and Corey Quinn joined forces again during re:Quinnvent for an animated and colorful discussion on Amazon CTO Werner Vogels’ AWS re:Invent keynote address. The good news? AWS seems to be giving special attention to a previously underrepresented area: operations and observability. The bad news? Doing so effectively may be a pipe dream.
Don’t get us wrong, there was a lot to “oh and ah” about. Vogels’ address signaled a shift for AWS— a real emphasis on “[simplifying] ops and reducing operational complexity” within infrastructure. What Lamas likened to a “nerd christmas,” AWS’ product announcements around observability and chaos engineering showed a newly focused attention on areas that “no one was talking about 5 years ago,” a great sign for champions of these operational processes.
"Werner pushing things that were traditionally part of the ‘reliability canon’ as core AWS principles— [that] feels like they are making more of a stand on this is what DevOps looks like [and] this is how you are supposed to do DevOps, which is new for them," Lamas stated.
But will AWS’ continued quest to “kill” already established SaaS providers in these areas take root? It’s unlikely, Lamas and Quinn agreed. “I'm skeptical that all these different services they’ve announced will be as good as an integrated offering from a company that’s entirely focused on this," remarked Lamas.
The announcement of DevOps Guru, a service aimed to help operations teams identify app performance issues, fit into this new narrative. Quinn questioned the tool’s effectiveness: “I’m highly skeptical that AI or machine learning is going to do an effective job at pinpointing more errors in production than the false flags that it sends.”
In a rebuttal, Lamas explained, “there is a place for machine learning and data analysis in modern operations.” However, she added an important caveat— “At the same time, we at Transposit are big believers of the human-in-the-loop approach. DevOps Guru can tell you where there is a configuration mismatch but not when there are issues with two different components in the infrastructure that are conflicting with each other and how you resolve that. It can give you guidance but that doesn’t mean it’s integrated into your environment. So we’ll still need humans that have intuition and a broader context in systems that can guide these machines and systems that enable humans to interact with and drive action on that data.”
Vogels’ keynote may indicate that AWS is ready to really listen to customers, address feedback, and help guide users with best practices, something both Quinn and Lamas lamented has typically been ”left in the dark.” The two spoke about the differences in philosophy between an extremely prescriptive Google and an incredibly hands-off Amazon when it comes to sharing how their infrastructure should be used: “AWS has been the place that will give you all the tools and let you stab yourself in many many many different ways but it doesn’t mean that they’ve told you how you should do things”, said Lamas.
Ultimately, “this was the year of ops.” So, the goal, the two agreed, should be meeting customers where they are. Whether successful in their endeavors or not, AWS’ shift to spending more time on managing infrastructure versus simply “spitting out more and more services” is exactly the right way to go.