AWS Summit Sydney 2015: Conference Report, Part 2

03 Jun 2015 AWS Summit Sydney 2015: Conference Report, Part 2

A couple of weeks ago Shine attended an exclusive Amazon Web Services (AWS) Partner Roundtable, AWS Partner day, and open-to-all AWS Summit day in Sydney. I’ve discussed the Partner Day and Roundtable in a previous post – in this post I’m going to talk about my experiences at the main AWS Summit.

Getting Started

The inclement weather continued for day two of my Sydney trip, though thankfully Amazon seemed to be well prepared. After a short dash from the taxi, I was under cover and managed to (mostly) stay dry for the rest of the day.

The main presentations were in the Horden Pavilion, while the Solutions Pavilion was in the adjoining Royal Hall of Industries. After checking in and getting my pass, I caught up with a couple of AWS staffers from the Partner Day then it was off into the Keynote.

The last time I had been to the Horden Pavilion was to see Primus during a Big Day Out, and as soon as I walked in memories came flooding back. The Horden was setup by Amazon very well; a large square stage in the middle with a massive screen on each side and tiered seating that allowed everyone a great view of the speakers.

Mike Clayville (VP of Marketing, Sales, Pdt Mgmt, & Support) delivered the main keynote, with various guest speakers taking over for short stints to discuss their experiences.

The first part of the keynote covered the basics: the growth of AWS, who is using AWS, and who is partnering with Amazon to provide services and technologies. AWS has an impressive stable of customers, from large Enterprises like Qantas, NAB, Comm Bank, Netflix, MLB and Kellogs, to Startups such as Pinterest, AirBnB, SoundCloud, Wix, Yelp, Foursquare, Reddit, Coursera, and Spotify.

One interesting thing I noticed was the number of video game companies listed: Rovio, Riot Games, Supercell, PikPok, FunPlus, Ubisoft, EA, and Nintendo. I guess the question is to what degree these companies leverage AWS, but regardless of whether it is powering their websites, their multiplayer backends, or perhaps even data analytics, it’s still cool.

Next up was the big theme slide:

I really like this concept. Whereas a few years ago businesses needed convincing to move to cloud services, it’s now their default position – businesses now think ‘Cloud First’. Of course, Cloud does require a different mindset to the traditional data centre approach, especially around infrastructure management and software development.

A classic example of this is the ‘Pets vs Cattle’ approach to servers, where organisations may have migrated apps to the cloud but they are still managing servers “the old fashioned way”. While its great that Cloud is the new normal, I think that there is still a maturation process for businesses to fully leverage its benefits.

Next up Mike began going through his ‘Patterns for the New Normal’, so I’ll use them to break up the rest of my keynote coverage.

Pattern 1: Start-ups Build Businesses From Scratch in The Cloud

Start-ups are building solutions completely in the Cloud as they have no legacy apps and no dependencies. They benefit from lower infrastructure costs and can move much quicker than if they used on-premises infrastructure. This one is a no-brainer. Unless a start-up has some very specific hardware requirements, they should be all-in with the Cloud.

During this pattern, Mike introduced Luke Anear, Founder & CEO of SafetyCulture, an Australian start-up disrupting the Workplace Health and Safety area. Luke was very engaging, and described his company’s journey from the garage (literally) to the successful company it is today. Interestingly, they didn’t begin on AWS but realised they couldn’t exist without it. This was also the first point in the day that the new buzz word microservices popped up. Keep an eye on this blog for Shine’s take on the current craze.

Pattern 2: Speed Is Not Just For Start-ups

With the tagline of Companies of All Sizes Will Move Faster Than Ever Before, Mike introduced Chris Taylor, CTO of Qantas. Chris was very open and his talk illustrated what a great positive impact AWS can have. One of the first points he made was that the barrier of entry for business cases was lowered due to the cost savings of AWS versus provisioning hardware to get anything started.

He also raised a personal favourite of mine: process wasn’t operating at the “speed of cloud” and so they required a rethink, especially around areas such as change control. I’ve seen this anti-pattern many times, where even though solutions are cloud-based, the business gets frustrated at the slow pace of delivery without realising its actually process rather than technology holding things up.

Finally, Chris noted that Qantas automated everything and didn’t put apps up in the Cloud until they had the platform ready. This enabled them to see benefits such as Route Planning being reduced from 4 months to 4 hours and an 80% cost reduction by moving bookings from batch to real-time. And of course, he mentioned that they are moving to microservices :).

Pattern 3: Customers Want Access To The Whole Lot

This ‘pattern’, though true, was pretty much just an opportunity for Amazon to tout the breadth and depth of their services (which is considerable).

An interesting statement from Gartner was that ‘IT organizations cannot treat IaaS providers like commodities’. While I guess this holds true for the providers and their attempts to differentiate themselves, I think that infrastructure is certainly moving more towards a commodity, similar to electricity or water. Using tools like Docker containers, an organisation should be able to easily switch between providers based upon price and value-add features at will.

Other than worrying about data protection and sovereignty, businesses care more about cost and uptime than who provides the infrastructure. This section was also used to introduce Elastic File System, an S3-like but with a structured file system as opposed to buckets.

Pattern 4: Companies Will Use Data More Expansively Than At Any Other Point In History

Mike began this section discussing AWS Big Data support including Redshift, Kinesis, and EMR. He then introduced Julian Bell, IT Director of Amaysim.

One of Julian’s first statements was ‘Data is the New Oil’. While I have no doubt the value of Big Data, especially to a company such as Amaysim, I’m not sure we’re going to see any Big Data Tycoons buying up English Premier League soccer clubs.

What I did find interesting about Julians talk was that they originally tried to do the hardware themselves, then found they couldn’t scale with demand. He went on to describe their shift to the cloud, focusing on operational transformation, automating the process and not allowing manual logins to servers.

Mike then returned to discuss machine learning, covering how Amazon went from ‘customers who bought this also bought’ all the way up to keeping the box open for certain customers because they know that they come back again during the day and buy more.

He finished off the section by announcing a fully managed AWS Machine Learning service that leverages all of Amazons “learnings” (see what I did there? :)). This service came about by Amazon wanting to provide an internal service for machine learning; Mike provided an example where a task that took 2 developers 45 days to complete could be performed by a single developer in 20 minutes with the new service.

Pattern 5: The Old Shackles Are Loosening

Focusing on databases and the desktop, this pattern is perhaps one of the weaker ones in the deck. Amazon Aurora is a good idea and seems built for the cloud, providing an enterprise grade database (compatible with MySQL) at a fraction of the cost but with an extensive feature set. However, RDS still features Oracle, MySQL, MSSQL and PostgreSQL; these databases are popular for reasons beyond ‘shackles’.

The second part of this pattern, the desktop, focused on physical vs virtual desktops. While I see the benefits of virtualised desktops, I think this misses the mark; it should perhaps be looking at Desktop vs Mobile. While virtual desktops can be accessed via tablet, it’s still the same paradigm – I think there’s an opportunity to really disrupt the whole desktop concept and switch things up to better suit mobility and BYOD.

Pattern 6: Reinvention Is Continuous

This was easily the most interesting pattern for me, covering Amazon’s Container Service and Lambda.

The Container Service provides everything you need to launch clusters of Dockerised apps. These apps run on EC2 instances, and can use EBS, ELBs, VPCs, IAM and CloudTrail. Scheduling tools are included, and the cluster supports auto scaling, auto recovery, and scheduled upgrades. Its also free; the only cost is that of the AWS resources consumed (e.g. EC2, EBS, etc). An API can also be used, and integrated into CI (such as Jenkins) this service provides a great pipeline for businesses who have adopted Docker.

The other part of this pattern, Lambda, is an event driven compute service that will run a single Node.js function in response to an event (Java coming soon). I think this is a very cool development, however I would like to see the event sources expanded; currently the only supported events come from DynamoDB, S3, SNS, Kinesis, CloudTrail and Cognito. There is also an API that can be used to manually fire a function.

While this range is great, especially for mobile devices that can fire events directly, I would like to see an option for consuming some sort of web feed automatically. As an example, if I want to consume a filtered POST call to the Twitter API so that I can perform some sentiment analysis on it, I need an EC2 server to schedule the calls and deposit the data in DynamoDB or Kinesis so that my Lambda functions will execute. It would be much cooler if there was no need for the EC2 server and I could just say ‘make this call every 5 minutes, parse the data and make a lambda call for each object’.

Pattern 7: Moving To The Cloud Is Not A Binary Decision

This pattern covers the Hybrid Cloud – the concept of having some services in the Cloud and some in more traditional data centres, connected via VPN or DirectConnect.

At Shine, we’re seeing this more and more. Many businesses have solutions running in existing data centers (or even on premise) but want to replace consumer facing web applications that integrate to them. In this case, it makes sense to move or create web applications into AWS while using VPCs and DirectConnect to safely connect to business critical legacy applications.

A Hybrid Cloud architecture brings the benefits of cloud (scalability, redundancy, cost, etc) to Internet facing applications that need it while leveraging existing time tested back end systems. Of course, the legacy systems need to be able to handle the load of any cloud-based applications, otherwise they just become a bottleneck. The last slide of this section made a good point: Hybrid is part of the journey, not the destination; the goal is still to be fully in the cloud.

Pattern 8: Customers & Partners Are All-in On The Cloud And AWS.

This was the final pattern in the keynote, and just covers the fact that businesses such as Netflix are running their IT completely in AWS. There was a great quote from Charles Philips, CEO of Infor; “Friends don’t let friends build data centers”. I really think this sums up the advice I would give to businesses that are new to the cloud. Creating new data centers now without a critical business requirement (e.g. legislation, etc.) is essentially creating infrastructure tech debt.

One last interesting thing from the keynote: Mike described a story where a chemical company modelled 205000 molecules on AWS using 156,000 cores for a cost of $33,000; the equivalent hardware would have cost them $65,000,000 to buy. Astonishingly, Mike stated that Amazon wouldn’t even notice that kind of usage.

The scale of what AWS provides is staggering. It’s almost like the concept of insurance, where the risk is spread across the many and the few who need to claim benefit the most by not bearing the full cost of their loss. In the case of IaaS, the cost of the hardware is spread across the many, and the heaviest users benefit the most.

So that wraps up the keynote. In my final blog post for this trip I’ll cover my experiences in the Solutions Pavilion and the breakout sessions that I attended.