A few weeks ago, I was lucky enough to head up to Sydney to attend the AWS Partner day and AWS Summit. In this post I’ll summarise how it all went and some of the impressions I had.
AWS Partner Day
The AWS Partner day is a business focussed event (rather than a technical one) for partners to get the latest updates and with plenty of time for networking. The focus this year was on partners getting more of their staff certified (particularly up to the professional level certifications), and also pushing more of the competencies.
The main reason for the latter was that there is so much in the AWS ecosystem now that it is hard to know everything there is. They are therefore encouraging partners to focus on areas where they are skilled, and to promote them with the help of AWS. Being able to set up a VPC, a few EC2 instances and some load balancers is a given these days, and so companies who focus on that side of things aren’t able to differentiate themselves at all.
There was a strong emphasis on the large ERP (e.g. SAP) implementations in the cloud but also on some of the less well known parts of AWS where few people have specialised up to now, with a particular push for much of the new machine learning aspects, which was also a theme for the Summit.
AWS Summit Sydney – Day 1
The Summit took place at the Hordern Pavilion in Sydney and was a well-attended event with approximately 10,000 visitors over the 2 days. It kicked off in true Amazon style with a set by ‘DJ Kanban’ who mixed music using a JIRA Kanban board. This kept everyone entertained until the keynote began.
The keynote was led by Werner Vogels and, similar to re:Invent last year, he used many analogies of super powers (some with quite tenuous links!) to highlight some of the old and new aspects of the AWS platform. Sadly, there were no new technical announcements, but it was an entertaining keynote.
Some high-profile guests including Brian Hartzer, the Managing Director and CEO of Westpac group. Brian discussed the 200th anniversary that Westpac are celebrating this year, making them Australia’s oldest company. He talk about how, having survived 200 years, to survive for 200 more, they must adapt to change and innovate constantly. He made a compelling argument, even though innovation and change aren’t words we usually associate with our big banks.
The day continued with many strong presentations. I’ve picked out a few of those that I made it to below to give you a flavour.
Versent presented on some technical tools and approaches to better manage multiple AWS accounts and credentials backed by Amazon’s Key Management Service. It was badged as a ‘level 500’, deeply technical session and although it had some good points, I think a few people were disappointed that it wasn’t more technical.
There was a strong theme of artificial intelligence throughout the conference with a practical test of Amazon Rekognition on show in the main pavilion (more on that later) but also in many of the sessions. One of the best was by Adam Lartner and Alastair Cousins (AWS Solutions Architects) who put together a demo using Rekognition, Polly and Lex built on a raspberry pi version of the Amazon Echo (aka Alexa). They demonstrated how these technologies could be used together, along with a few other tools, to pre-process video to identify whether there are faces in the images first, on the pi, before sending them to Rekognition.
Sri Nadendla and Daniel Hobson presented a talk on the various messaging solutions within the ecosystem – SQS, SNS, Kinesis and Amazon IoT. They presented the pros and cons of each, the restrictions that apply and the models that they provide. It provided some useful information and I am certainly wiser about the options, but I’m not sure I came away with a clearer understanding of when I should use one over the others.
In addition to the focus on AI, there was a strong emphasis on serverless computing, with several talks on Lambdas and how they can best be leveraged for data processing without the need to manage any servers or infrastructure directly. Big data also continued to be a topic, with several talks touching on aspects of Amazon Kinesis and its various incarnations (Streams, Firehose and Analytics) and how they can be used to analyse both batch and streaming, real-time, data.
Day one’s sessions finished with a talk on Apache Spark with a well-presented demo of what Spark can do and how to set it up easily on AWS.
After the sessions, there was some time for drinks and networking, and time to visit the booths around the hall presented by many of the sponsors. Sadly, as is typical of this sort of event, they were mainly staffed by marketers rather than technical staff so I found them of limited value, the exception being the AWS stand itself where I had some good discussions around Lex and Rekognition and also Lambdas.
One of the booths was dedicated to Amazon Rekognition where they were running an experiment which would ultimately feature in the keynote on day two. There was a camera which took your picture and allowed you to register some rather banal details such as where you are from and what your favourite animal is. Several other cameras were placed around the hall taking photos every 15 seconds and these images were fed into Rekognition to plot your movements around the hall.
In practice, it seems 15 seconds was too long and many people weren’t captured except when they registered at the booth. I was somewhat of an anomaly, not registering (though I wanted to) but having the booth recognise me fifteen times around the hall despite not knowing who I was – it caused some confusion as it was only supposed to recognise you if you had registered. Perhaps I look like too many of the other attendees or my doppelgänger was also there!
AWS Summit Sydney – Day 2
Day two’s keynote focussed around several companies that were using AWS solutions in novel and interesting ways. One that stuck in my mind was Elanation, who produce a Fitbit-style wearable for kids that promotes exercise by opening up new levels on their iPad games as they do more physical exercise.
There was also an innovative energy company, Mojo Power, who aim to drive down energy costs by allowing customers to access wholesale energy rates, but also dynamically telling them to reduce their usage when the demand on the grid, and therefore the price, is at a peak.
Generally, I found the talks on day two weren’t quite up to the standard of day one but there were some interesting high points. A talk from the two co-founders of Jemsoft on their use of machine learning and open source technologies to identify brands in streaming video and the challenges they faced continued the theme of the previous day whilst focusing on the hardware that AWS could provide to help rather than on the out-of-the-box services.
The talk on step-functions was an interesting dive into how they could be used, but was sadly caveated with “not available in our region”. I finished the day – and the summit – with a talk on Amazon’s new x-ray functionality for tracing calls through the stack and how it is now also available to help track lambdas.
Overall, it was an interesting and useful couple of days sending me home with many ideas for improvements in our approaches, which can hopefully be put into action over the coming months.