Earlier this month I had the pleasure of attending AWS re:invent 2016 in Las Vegas with 3 of my colleagues from tombola towers, all in different roles which meant we rarely were seen in the same talks (as expected). The event was on a scale I was not expecting, spread across 3 Vegas hotels (anyone who’s been to Vegas will understand how big the hotels are, those who haven’t… they’re big!). There were over 32,000 attendees this year the event was sold out (an indicator of just how big AWS is in the cloud world). The organisation of the entire event was brilliant and I can’t say I witnessed any problems encountered for the organisers. For anyone lucky enough to attend in future best advice is plan your chosen sessions well, allow sufficient time to get between talks (back to back talks just doesn’t work unless they are very close). Also wear comfy trousers and shoes as there’s a lot of walking involved.

 

Keynotes

The keynote speakers were from AWS CEO Andy Jassy and CTO Werner Vogel spread over two mornings. The highlights for me that were announced:

New EC2 instance types – More options to choose are always welcomed, often we are restricted to a larger than needed instance type purely because there isn’t an “in-between”. Some of the new types announced I’m sure will feature within our estate in the coming months.

Price reduction – Cost is a major consideration for an infrastructure manager so of course a price reduction is always welcome!

 

EC2 System Manager

A new feature released by Amazon to help us Infra folk to manage EC2 estates.

systems1

Some of the key features of EC2 System Manager which offer some great improvements whilst managing an EC2 estate include:

State Manager – configuration state, helping avoid config drift

Automation – A show of hands demonstrated how many AWS users are still building their own AMI’s. System Manager is aimed at providing an automation suite which will help in this process

Inventory – Our EC2 estate has only grown since moving to the cloud, an inventory tool is something which should significantly help keep track of what we have running where

Patch Manager – Security at re:invent 2016 had a lot of focus, key to Security in a server estate is patch management. A toolset to help keep on top of the vast number of patches released every month is a big bonus.

 

DDOS Mitigation / AWS Shield

As already mentioned Security was a big topic at re:invent 2016, Werner touched on this aspect of the Cloud in his keynotes talk. AWS had recently suffered a fairly significant DNS attack just a couple of months ago which did see its customers impacted so think it was only appropriate for DDOS mitigation to get a mention. AWS are probably one of the most attacked companies in the world due to their large global profile, AWS enable their DDOS mitigation product (AWS Shield) by default for all customers for free. It was quite clear that AWS wants to protect their customers and also make sure they don’t incur additional costs which is great to see from such a large corporation like Amazon.

 

X-Ray

A new product that was announced this year s X-Ray. Aimed at helping developers debug their code to detect and identify bottlenecks which are causing issues. Whilst there are 3rd party products out there which can do this (datadog etc.) it’s great to see this getting added to AWS’s tool belt without the need to go 3rd party.

 

How Netflix Achieves Email Delivery at Global Scale with Amazon SES

Netflix is the poster child of Amazon Web Services. If ever there was a Cloud success story, Netflix is it. They obviously work very closely with AWS and have achieved a lot in the Cloud. There were a number of talks relating to how Netflix did this and that etc.

Devika Chawla (Netflix Engineering Director) was joined by Kadir Rathnavalu from AWS to discuss how Netflix moved from traditional mass mail delivery from a SPOF / bottleneck to utilising SES to delivery Globally at scale. Devika’s talk was more of a high level real world example of how to utilise SES and was very insightful as to the methods used not just for email delivery but for the strategy around targeted email campaigns which would not only benefit the “techies” but also various Marketing teams. Kadir went on to explain the more technical detail around SES and how to implement in organisations.

 

Automation and Scaling Infrastructure

Being very much a Dev house here at tombola this talk touched on a number of common problems we encounter on a daily basis. This picture sums up the challenges we face most days:

automation

This talk focussed on how we can utilise AWS services to help enable the IT Admin to maintain the aspects they are concerned with whilst allowing the Developer the agility to be unblocked from doing their job. There are a number of tools provided in AWS which can help tackle these problems including Cloudwatch for monitoring, Cloudtrail to audit changes and Lambda functions for automating remedial actions. The guest speakers mostly focussed on config and security changes but the examples given could easily be adapted across to other areas.

 

Containers

There were a number of talks around containers throughout the event. Being relatively new to the concept I attended a couple of the introduction talks around Docker and ECS. Like a lot of people we’ve heard of containers and love the idea but never really took the plunge into investing any real effort in how we can utilise these in our production environment. Danielle Greshock from AWS talked about why you would want to use containers the benefits of containers, how you go about managing the AWS container clusters (ECS) and running services within ECS.

From a real world perspective I’ve seen firsthand how applications spiral into a monolithic application that does every which in turn makes changing these applications extremely challenging and risks not just the application you want to change but all the other applications that are linked within the monolith.

She also talked about the scalability of ECS and how they scaled up to 100 instances and latency remained consistently low which adds confidence in the stability of ECS at scale.

 

Cost Management

A big part of managing an efficient infrastructure is keeping control of costs. What was apparent throughout the talks and Expo was that there are a number of methods and practices to help keep track of and reduce costs. Something tombola are working through at the moment is adopting Reserved Instances (RI’s) so getting a good overview of how best to approach this as well as the ability to speak to AWS and Cloudability around best practice was great. The newly launch Convertible RI seems like a great feature for reducing costs whilst still keeping the ability to be flexible in future if our requirements on EC2 specs change.

 

Summary

For me it is clear that Amazon is the dominant force in the Cloud (sorry Google). There are new products released all the time and what better place to showcase these than in Vegas at AWS re:invent. We at tombola only use a small subset of what is available to us, whilst we are relatively new to the Cloud and still learning there is a whole host of technologies and tools available to us to improve our not only our infrastructure but also our product and security.

The conference itself is overwhelming and highly recommended, the number of vendors present at Expo is crazy and all geared up to offer you products that are designed to work and integrate with AWS.

There are a whole host of the talks available on Amazon’s youtube channel, I highly recommend having a watch of some if you get a chance: https://www.youtube.com/user/AmazonWebServices/playlists?view=50&shelf_id=19&sort=dd