AWS re:Invent 2018: [REPEAT 1] A Serverless Journey: AWS Lambda Under the Hood (SRV409-R1)

Serverless computing allows you to build and run applications and services without thinking about servers. Serverless applications don't require you to provision, scale, and manage any servers. However, under the hood, there is a sophisticated architecture that takes care of all the undifferentiated heavy lifting for the developer. Join Holly Mesrobian, Director of Engineering, and Marc Brooker, Senior Principal of Engineering, to learn how AWS architected one of the fastest-growing AWS services. In this session, we show you how Lambda takes care of everything required to run and scale your code with high availability

[@user72974]

I thought the beginning with Holly was fine. It was a nice intro to the deep dive, it's not like she has to get into all the technical detail right away.

[@diegoacmendes]

If you want to see the good stuff skip to 19:00

[@d00de]

Good talk but a pretty bad Q&A session. A lot of deferred questions and weird answers. At 53:12, someone asks if there will be a way to ensure a certain number of warm lambdas and Holly answers with (paraphrasing): "We are aware that latency is an issue and we like hearing from our customers that this is an issue, so it's great that you also just said it's an issue. Thanks!"

[@chris-ew9wl]

that made me cringe lol

[@meli7a-lee]

Yes - that jumped out at me as well as I would've liked to hear directly from them what the recommended practice is. Was hoping the asker would've pushed harder for an answer, or that Marc would've chimed in. But yea... awkward =)

[@haritadepalli959]

Great observation about and application of the law of large numbers to server load balancing. Intuitively, a finer granularity in the sizes of the workloads should provide for a better utilization of the resource (server). I think what is not clear here is - the gaussian patterns in the aggregation of the workloads assures that a majority of servers are loaded at the group average of the workloads. So, one could plan a majority of one's servers to be operating at a peak level close to the group average. It does not cancel the need to have some servers running under capacity (left tail) and under provisioned for some workloads (right tail) - for which, one will need servers with a different capacity bar.

[@arada123]

Not a good talk at all. Reading from script demotivate the listner.

[@brionl9377]

I like this speaker: Holly

[@ManuAnand79]

@35:36 Yes there are servers in Server less :)

[@jabberl.]

Great talk on the architecture but horrible Q&A session, especially the director dodged the most key question about ensuring warm instances and low latency, that's all we care.

[@joggyjames]

ha sure did... probably because it wasn't on any of her prepared slides.

[@serjoshua]

"The more workloads you put on a box, the more uncorrelated workloads (and that's very important) you put on a box, the better behaved they are in aggregate. So, this is something that is very powerful for us at scale."
that is some good sh*t

[@tkawai2008]

Great presentation and Q&A session, thank you very much.I understood Lambda Architecture very well.I found some miss dictations which is correct of ENI, check these please.

[@UberDoyle]

It's like listening to a text book... -_-;

[@ukaszlech4368]

Oh lord, part 1 was painful to watch.

[@edmundandbaldrick]

and the questions weren’t 🤣

[@EmbImg]

Director of engineering and she couldn't prepare to speak on her own but read off of slides about the engineering technology she is directing on? Unfit.

[@DShanos]

what a pretty voice