Skip to main content

AWS re:Invent Day One – Fast and Arcuri-ous

Posted on November 28, 2017 by CraigArcuriCraigArcuri

I’ve been to some big, high energy cities, such as Shanghai and NYC, but this is my first trip to Las Vegas and I can best describe it as unique. Not all good, not all bad but certainly unique and very interesting. And I learned one other thing after arriving Sunday night. The “campus” for re:Invent 2017, including the MGM Grand, Aria, The Venetian, and the Mirage, is HUGE. Most of the Linux Academy team is staying at Harrah’s and our booth at The Venetian, “right next door,” is a 20-25 minute walk.
I tend not to learn such lessons the first time. So I set out excited on day one of re:Invent having already been registered for 5 breakout sessions, with 4 of them being split between MGM and the Aria (“right next door” of course). It was a long and arduous day. My colleague, fellow AWS Course Author Trent Hayes, tracked that he walked 7.5 miles. Based on the location of my sessions, I’m estimating I walked between 10 and 11 miles on day one. I welcome booth duty on day 2. (Come visit us at the Linux Academy booths 737 and 738 at the Venetian!!).
I made it to 5 sessions on day 1 and Amazon did not disappoint. They were all excellent and very informative. The complexity of some of the content was such that I won’t be able to do each session justice in a short blog post, but I’ll try to give an overview of each session.

Optimizing costs

I set out in the morning for my first session at the MGM. I did not come out of the gate blazing as myself and another gentleman stood in a Harrah’s garage for 15 minutes waiting for the shuttle. It was obvious to both of us that there should probably be more than two of us waiting for the re:Invent shuttle. But I made it to my first session, titled Optimizing Costs as You Scale on AWS, in the nick of time. I was excited to get a seat for this session because optimizing cost on AWS can be an Art as much as a Science and it is often overlooked. Getting the proper mix of EC2 instances (reserved, on-demand, spot) alone can save a company a great deal of money. The presenter emphasized going from paying for what you have vs paying for what you need. He went on to detail ‘The Five Pillars of Cost Optimization in AWS:

  1. Right-sizing your instances – There is really so much opportunity for costs saving here and for you to be a hero to your company and/or client. In a nutshell, select the cheapest instances available while meeting performance requirements.
  2. Increase Elasticity –  effective use of auto-scaling being a biggie here. But the presenter detailed using Lambda and CloudWatch to turn off non-productive instances. Much room for savings here!
  3. Pick the right pricing model – configure your environment properly to start and save money immediately. Reserved instances can save up to 75% and Spot Instances up to 90% (with the caveat that spot instances are not right for every use case)
  4. Leverage the right storage and service
  5. Governance measures and targets for optimization – understand what Trusted Advisor can do for you and how to use it to your advantage. Check out utilization metrics and turn off under-utilized instances! I was expecting a highly informative session and it certainly delivered. Cost optimization is really an under-appreciated skill. You will always be given opportunities if you save people money!

The first session was great and I set out optimistically for my second session at The Aria, which of course was “right next door” to MGM. Finding my way out of MGM and getting pointed in the right direction was 20 minutes of walking and asking. AWS has people posted everywhere to help, but the trek is long, with many twists and turns.


Having said all of that, I considered the walk a success as I got to check off my second legendary west coast burger chain in 24 hours (Fatburger; I got to In-N-Out on Sunday night). This may be controversial, but I have to give the slightest of nods to the Fatburger. I made it to The Aria “next door” about 15 minutes late for my second session on Kinesis. This was an excellent session as well but I missed a quarter of it and there are a lot of moving parts. Some of the use cases discussed for Kinesis included: streaming ETL data, continuous metric generation (leaderboards), and actionable insights (real-time reactions). Real-time is a keyword for Kinesis. If you need to work with real-time data, start thinking about Kinesis.

Internet of things (IoT)

My 3rd session was scheduled back up the strip at the Venetian. I got in line for the shuttle and quickly realized the line was very long and that my 4th and 5th sessions were back down this way at MGM and The Aria – time for plan B. I start walking around looking for a session with the smallest of the dreaded “walk-up” lines. It’s Vegas after all and the walk-up lines are a gamble offering no guarantee of getting in. But the ladies marshaling all these apprehensive techies were fun and friendly and I did make it into the session on IoT. It was an interesting session, but I have to admit IoT isn’t in my wheelhouse. There was a speaker from Amway discussing how they use IoT as well as a speaker for a company named Pentair. His discussion on how IoT is used to help people get water in 3rd world countries was moving. IoT is being used to save lives!
Well, it had been a productive day so far. I was tired but I soldiered on for another lap “next door” between MGM and The Aria. By this point, I was concerned about the battery life of my electronic equipment and I remembered walking past a CVS but I wasn’t going back three-quarters of a mile. I figured I’d be able to find a notepad somewhere. That turned into a bit of an adventure. So, if you’re at re:Invent and you’re preserving battery life, and you desperately need a notepad to take notes in sessions and can’t find anything, is it okay to buy a self-help journal with a picture of a bunny riding a unicorn on the front? I’m asking for a friend…

Serverless file processing

Well, I made it in time for session 4 and I’m glad that I did. The title itself was worth the trip (How we built a mission-critical serverless file processing pipeline for over 100 million photos). I can’t do that title justice in a paragraph. But the presenter, from Expedia, detailed how they use S3 storage and Lambda to manage photos. Basically, they have a ton of 4000 x 6000-pixel photos stored in S3. They use Lambda to reduce these photos to a quarter of the original size. They have several other Lambda functions working on these photos and ultimately get the photos to their content delivery network and their customers. It was a very informative session and showed just one use case detailing the potential of Lambda. And one of the key points of the session was how Expedia gets predictable spikes in traffic and how Lambda can be used to minimize costs (there’s that cost optimization theme again).

Real world AI for the enterprise

Alright! Last session of the day: Real world AI for the enterprise. The speaker from AWS discussed how Amazon uses AI and Machine Learning on their site, and what was surprising to me, how they have been using it for 20 years! This was an enjoyable session and definitely sparked my interest in AI and Machine Learning. A second speaker from Capital One detailed how they use AI and Machine Learning and how to get started with it (talent recruitment and team makeup). It was a great session and a great way to end the day. Well, not technically finished. I shuttled back to the Venetian for a meetup with my great teammates at Linux Academy and some fine folks from none other than AWS. I ended the day around midnight, exhausted but fulfilled – and I now understand full well that “next door” means something entirely different in Vegas!
Come visit the Linux Academy team at our booth at the Venetian! You can find us at booths 737 and 738 – we’d love to hear the story of your AWS journey.


Leave a Reply

Your email address will not be published. Required fields are marked *