For the large part Mike Clayville may have been preaching to the already converted with Cloud Computing but for the curious and sections of the delegation that were less well informed about the benefits bestowed in using AWS Cloud offerings, the Keynote was a text book sales pitch from Amazon – identifying the need, highlighting features, benefits and endorsements from OpenMarket, Lonely Planet and NAB.
No surprise then that Clayville runs the worldwide sales operations for AWS and Melbourne was one of a number of summits being held around the world where the overriding message is the AWS obsession with it’s customers, listening to what people want and encouraging feedback. The keynote agenda for Melbourne as well as looking at some of the customer successes was to look at why people are going to the cloud, what are they are getting out of it and how people can get to the cloud themselves.
Clayville’s pitch for AWS is evidently well rehearsed and polished, “AWS has been the platform of choice for about eight years for start ups driving hard to innovate. Our focus is on ‘let’s deliver for the start ups that undifferentiated heavy lifting’ so they can focus their scarce resources, their time and their money on differentiating innovation that they can bring to the marketplace. That’s the reason why people like Pinterest, Dropbox and Freelancer have really standardised on our platform and we are very proud to have them there and will continue to drive new services for them so that they can then innovate on top of those and create new market opportunities for themselves.” AWS positioned start ups as their primary sector at Melbourne whilst also mentioning that there is a huge take up of their offerings at an enterprise level, “What we are seeing in the enterprise space is that this is an opportunity that they cannot pass up.” Clayville says that in his long career he has never seen enterprises take up new technology like they are taking up cloud today.
Continuing the pattern of inertia selling, the public sector has also taken up Cloud with the intent of adding value to their constituencies and not spending time on that (once again) “undifferentiated heavy lifting”. The ecosystem which AWS has built up is extremely rich bolstered by system integrators with access to a range of education services who work jointly with customers, allowing them to provide the best possible outcomes in transformation to the cloud.
One of the main reasons that organisations are beginning to stand up, take notice and drive adoption of the cloud, Clayville says, is agility. He went on to explain that during a quarterly update of one of their projects with the CIO from a large customer the most appreciated benefit from migrating to the AWS cloud was not saving money but the change in pace of the business. Previously, they had taken 24 months for project cycles, sitting around brainstorming, say, 50 ideas of how to drive business forward into new markets and be more competitive. The big tax is building an architecture around what that project would look like but more importantly spending time narrowing down ideas to two big bets before building the infrastructure, along with the business case and integrating it whilst always risking the cost of failure which is extremely high. The AWS Cloud offers an iterative process of innovation which reduces the cost of failure to nearly zero. For example, with AWS Redshift for $999 you can get a terabyte of data which means that the cost of failure for a year is $999 allowing you to contemplate the way you can approach failure and perhaps go after all 50 of your options very quickly and fail or succeed fast whilst bringing the business cycle right down to 3-6 months. To demonstrate this ‘culture of innovation’ which Clayville went on to explain as an iterative style of business innovation that is being underpinned by the cloud, Rick Klink, the CEO of OpenMarkets talked about their experience with AWS.
Rick Klink, CEO, OpenMarkets
OpenMarkets is a Melbourne based platform provider of online stockbroking services and the context of Rick Klink’s talk was what has happened to stockbroking over the last 20 years, broadly the change from open outcry trading to platform based trading in more traditional office environments but most relevantly the advent of cloud computing in the last couple of years and time to market for trade execution with automated trades.
OpenMarkets have a high reliance on cloud to get trades onto the market as cheaply as possible. Klink explained that trading has traditionally been the remit of the big banks in Australia, where the high cost of infrastructure set up and maintenance are made palatable by spreading various products across the same infrastructure and sharing the capital expenditure load. OpenMarkets business model relies on ease of scalability to provide very quick execution on stock trades. What this means is that they are able to log onto the AWS console and provision by looking at how much capacity they need on the fly.
Klink gave a really good example of why cloud services are golden, pointing out the dot com boom as an example of when organisations pre cloud would have had to have built the infrastructure to cope with the bull run some ten years ago. Unfortunately the boom did not last which left behind redundant data centres whereas if the market goes away whilst using the cloud, you do not suffer Capex spend but rather, tether back usage. This is also reflected in CPU usage each day where most trades are executed between 9 and 5 meaning that if there is no usage at night, OpenMarkets are not paying for redundant usage.
Klink says,”We only trade one third of the day, so two thirds of the day the computers sit there spinning, they really don’t do anything. The CPU savings to us are because we only turn them on when we need them so by 10 o’clock in the morning we will know how many computers to turn on or off so if we pay by the hour for the computing time, we don’t need to carry those huge troughs between days.” This variable computing load allows OpenMarkets to compete against the big banks.
Another reason that AWS sees a lot of people migrating is because of the breadth and depth of the platform in terms of it’s scalability.
Darragh Kennedy, Online Platform Manager, Lonely Planet
Lonely Planet were a brilliant choice of customer to endorse the AWS platform, not least because the LP journey from start up to present is such an endearing one but absolutely in line with AWS targeting start ups in particular, perhaps with a view that some will enjoy the same success as Lonely Planet. In fact Kennedy allowed delegates to get lost in the adventure whilst explaining how the founders met on a park bench in London in 1972 and went on to track their way to Australia by land through Europe, the Middle East and Asia after buying a min van for £150 before ending up in Exmouth in Western Australia having run out of money. Everywhere they went people were amazed by their travel story which convinced them that they should write a book about their journey, “Across Asia on the Cheap” of which they printed 1,500 copies which sold out in weeks.
After a brief history of Lonelyplanet.com, Kennedy explained how LP had to undergo a huge agile transformation whilst looking to the cloud. “What we found is building a physical data centre is quite difficult to do in an agile fashion, it’s a very linear process which is not suited to agile however when you consume infrastructure as a service you no longer have those same constraints. You can break that work down into small pieces and begin delivering value immediately.” Lonely Planet put together a cross functional team to build their first environment and treated it similar to a software project writing infrastructure as code, checking it into a repository, building tests and using pairing to share knowledge. Kennedy goes on to say, “The first environment took us quite a while to build as it was quite a laborious process but what we found is that we were able to reuse that code and rapidly accelerate across multiple environments. We can now provide an environment through a repeatable reasonable process to any of our product owners or developers in less than ten minutes.”
LP have also used the flexibility of the cloud to split out the compute away from to physical data centres in Australia to regions around the world where is suits their business most. LP now use a broad range of AWS services starting with EC2, EBS, built many flavours of database on top of EBS. Flexibility allows Lonely Planet to try as many tools as they need in order to reach, as Kennedy puts it ‘that sweet point’. Capacity management is a completely different paradigm in the cloud. Cloud also allows you to fail fast and using the elasticity of cloud we can run four to five times as many builds per day and that increases the amount of different products they can get out. Cloudalso allows you to fail cheaply – if a product is successful you can scale out to reach that demand, if it is not successful you simply turn off that computer and with new campaigns they can have capacity sitting there ready to take the load.
With AWS another area of use is where people supplement existing workloads by building extensions of their business into the cloud.
David Broeren, Head of Technology, Digital and Online Channels, NAB
Broeren also gave a brief history of NAB, granted, a lot more condensed than those from OpenMarkets and Lonely Planet but the context was the same in giving an idea of the sweeping innovations NAB has been through to be able to compete effectively today. Nowadays digital is central to NAB’s operation to enhance the customer experience and Broeren spoke about implementing agile approaches and replacing some of their older tools to establish effective delivery pipelines.
Last year NAB connected up with AWS and Broeren gave a very good pictorial idea of how AWS offerings have been used since they decided to remove their own infrastructure modelling. Although the example he gave was for nab.com.au, Broeren explained, “the reality of today is a pattern for UBank, MLC and a whole bunch more in the pipeline so we are now genuinely running at scale.” “Using those delivery tools we have been able to apply that (scale) to Amazon using Cloud Formation, the ability to script everything and build EC2 instances, Elastic Load Balancer, CloudWatch alerting, autoscale environments and so on.” Broeren spoke about the business sponsor that was giving them the money to be able use AWS and explained how within one hour of showcasing AWS they went from opening an Amazon account to having two fully redundant data centres up and running and ready to take servers which peaked the interest of the sponsor.
From here, they were able to build out the instances and commission forty servers in two minutes using things like Puppet before overlaying the software stack, components and templates which meant that within an hour and two minutes they went from ground zero to having a fully functioning NAB site.
In order to test the load, NAB used beeswithmachineguns which is a utility to load test web applications in a development environment and see how AWS was able to cope with spiked activity and then tether back the capacity used after these attacks, “we turned on to see if Amazon Web Services has the ability to heal itself and make sure we always have that 100% expectation and we can meet it so it’s in production, it’s running right now and on average it kills about six servers a day whether that’s peak time, in marketing campaigns, we continually test so that if we are developing in this environment we can make the best use of those differentiating capabilities that AWS offers.” The driving factor for Broeren to use AWS then became the welfare of his teams, instances in which things go wrong or “go bump in the night” from a DevOps perspective and knowing that cloud reliability is supporting NAB’s Cloud efforts from AWS seems to have secured the presence of their products in the organisation and with testing this in a production environment, Broeren bet his job that it would.
