scale.bythebay.io: Adelbert Chang Interview
Recording: scale.bythebay.io: Adelbert Chang Interview
you [Music] I'm Adelbert Chang I am a lead engineer at target and I work on deployment infrastructure for the data science and optimization team one thing that I do a lot is try to do automated deployment infrastructure so automatically deploy he'll scale out systems and this tends to be very tricky because of things like people are gonna be using different languages different languages will implement the same thing over and over how do we abstract that away and I think what we're seeing now in projects like kubernetes Nomad envoy is moving that functionality out of the language and sort of having certain patterns so that we can reuse the same infrastructure to do things that are critical to our system so before we had schedulers I was very specific for MapReduce jobs now we have just like generic schedulers I can schedule services just as well as they can scheduled batch jobs for instance I think that the trend that we're seeing and something I'm really excited about [Music] I think automation will be a big one as you as so as you get like Sorenson when you want to scale up for servers because you see increased or something you want to be able to very easily hit a button and have it scale out maybe preferably automatically and that tends to you need to have a notion of like alright where do I where do I get the machines where do I get the the power from in order to to scale up these services how do I provision them so that they all all of their logs get shipped to the same location that I expect how do I get the metrics from them and if this is a manual process then this becomes very tedious as it's been in the past for me so I think as we were saying this surge of like automated infrastructure then that becomes that becomes more and more of a solved problem but still we've probably stages but I think automation is definitely a big one in there so on my team we write a lot of Haskell code so we do a lot of functional programming in the team and it's helped us a lot because it's given us a lot of confidence and what our code does and the nature of our work is very math heavy so we we work in supply chain optimization which has a lot of is very heavy and when we go to implement its algorithms in Haskell because of the way Haskell looks and because of sort of the theoretical foundations that has both rooted in our implementation ends up looking pretty similar to the equation that you would see on paper so functional programming gives us assurance is both from a mathematical perspective but also from like a global program my perspective over you're able to look at a chunk of code and understand it reasonably well to be confident and deploying it to production my favorite part is I've been attending this conference since it was since the first iteration and back when I was at Intel and it's been really cool to see it grow what I particularly like this year is how Alexi has really recognized the conference does not just call a specific we're just like a variety of different tract of data to functional programming to systems and so I think the most exciting part about the conference is you have all of these very interesting topics all together in one conference and to my knowledge there's not many other countries out there that provide the same mixture of disciplines