How to Succeed at Failing

AWS Data Hero Anahit Pogosova doesn’t just handle failure, she gets a little excited about it. She’s a Lead Cloud Software Engineer and a dedicated advocate for cloud, serverless, and data technologies. Anahit talks openly about how architecting solutions is about learning and improving from inevitable failures in the real world where nothing is ever as simple as it seems. She gave a much anticipated talk at re:Invent 2023 on the complexities and benefits of using AWS Kinesis Data Streams with AWS Lambda for a serverless data approach. In this brief interview, Anahit talks about why embracing failure is at the heart of her advocacy and what to expect from her talk. Video of her talk at AWS re:Invent 2023 is below.

"It might seem straightforward right off the bat: the session is for those who work with data streaming and are using, or considering using Amazon Kinesis Data Streams and AWS Lambda, right? But I dare to say, almost anyone working with AWS services could benefit from it. Because, in reality, this talk is about embracing failures. And even though it is about two very specific services, by the end of this talk you might realize that, in the grand scheme of things, it could be about almost any service. Because all of them can fail, sooner or later."

“It’s a 300-level talk, so it does assume some familiarity with KDS and Lambda. However, I would argue that even if you’re completely new to Kinesis Data Streams, you can still gain valuable insights from this talk. I begin by providing some foundational information required in the talk, ensuring that a broader audience can benefit from the content, not just those with prior experience with KDS.“

To dive deep into the topic, check out Anahit’s two-part series on Mastering AWS Kinesis Data Streams:

"It’s a very tightly packed session as-is! But one topic I wish I could explain in more detail is how Lambda interacts with the stream behind the scenes, the consequences and limitations of the default behavior, and the possible alternatives and workarounds. But I suppose attendees would have to go read my blog posts for that. :) The reality, of course, is different. And once you understand what the default behavior is and connect it to the capabilities that come with the KDS, it all starts to make sense.”

"I like the “Why?” questions the most; they help us understand the underlying reasons behind observed behavior. One such question could be: Why does Lambda handle errors by default the way it does? Because without delving into the ‘whys,’ it might seem as if someone on the Lambda team simply decided to be mean and make it hard for us developers :)"

"In the past, I have worked with large-scale streaming data applications and have experienced everything I’m discussing in my session firsthand. It was also when I first realized that everything indeed fails, all the time. So, I became determined to get to the bottom of it!"

"I think it’s a bit too early to tell. It might be the case that Gen AI will be able to warn about the failures I talk about right away and suggest possible solutions. It’s not rocket science, after all. But, I would argue that even then it would still be very valuable to understand how the services actually work, how they interact, and how they fail."

About Anahit

Anahit brings over a decade of experience in full-stack and data solutions to her roles as an AWS Data Hero and a Lead Cloud Software Engineer at Solita, one of Finland’s leading digital transformation companies. Anahit passionately shares her knowledge with the community through various channels, including conferences and meetups, podcasts and YouTube, as well as blogging and online courses. Connect with Anahit on LinkedIn or X (Twitter) @anahit_fi.