Last night I was fortunate enough to attend the Docker Meetup Groups event. Docker in this case being the forum for a cloud-centric discussion -- The Big Debate: AWS v Azure vs Google Cloud vs EMC Hybrid Cloud(external link) -- where AWS, Microsoft Azure, Google and EMC Hybrid squared off.
Debate Teams:
AWS:
• David Cheal, CTO, Strut Digital
• David Taberner, Cloud Solution Architect, Reckon
Azure:
• Michael O'Keefe, Open Source Product Manager, Microsoft
• Tom Hollander, Cloud Solutions Architect, Microsoft
Google:
• Nicholas Tan, Chief Architect, News Corp
• Win Woo, Cloud Solutions Engineer, Google
EMC Hybrid:
• Matthew Zwolenski, CTO, ANZ, EMC
• Danny Elmarji, Snr Mgr, Systems Engineering, ANZ, EMC
Debate Moderators:
• Anthony Shaw, Head of Innovation (Cloud, Group), Dimension Data
• Stephen Joseph, CTO, Streethawk
• Scott Coulton, DevOps Solutions Architect, Healthdirect
Pre-face
The following are my thoughts only, at a high level and you will also note that I have generally not attributed direct quotes to individuals (less chance of litigation). If you are really interested in any further particulars you are welcome to reach out and ask me personally. This blog also does not aim to be a complete play-by-play of the event or an endorsement of any of these platforms.
The event took place at Sydney's CommBank office. And there were at least 300 people in the audience. Not a bad effort for a Monday. I was fortunate enough to get there early and secure a ringside seat directly behind the moderators.
Everyone on stage was knowledgeable, experienced and from an operational and technical background. The conversations got technical and detailed which was good, I was not looking to hear sales-speak. These were honest professionals rather than gold chain-wearing salespeople.
Breaking down the debate
The debate lasted a solid few hours and was very well run by host Gerhard Schweinitz. The questions put to the panel were relevant to the trends in cloud platforms and would be good ingredients in anyone's cloud platform comparison exercise.
Right off the bat, the teams were given 60 seconds to explain why their platform was the best choice for cloud, with all teams coming out punching.
Questions
Questions followed around High Availability (HA) and Disaster Recovery (DR) – a big part of the assessment criteria of any hosting platform – cloud or otherwise. The debaters touched upon availability, reliability, redundancy and automated failover and how their tool sets eased this pain.
Geographic data centres
All of the debate participants had similar things to say about their platform's multiple geographic data centre offerings, as well as a growing array of provided service offerings that automate a lot of the mechanics of High Availability for particular services.
Solution architecture
Attention was also given to detail here as the debaters correctly referenced that your solution architecture does make a difference to your application's resilience in the cloud. You can make build choices that will pay dividends in availability and robustness – “Cloud Native” is the magic phrase here.
Data sovereignty
The audience also put forward some questions around data sovereignty and how that plays out for these USA-homed platform providers. No massive surprises here in terms of answers and some cautious language was used by some on stage around solutions that “comply to all regulatory requirements” - which you can take how you will.
Cloud-agnostic architectures
There was an interesting line of discussion around cloud-agnostic architectures and how easy or feasible it was to migrate from one platform to another. There was a lot of talk of deployment and management tools like Puppet, chef and Ansible. However, in reality, I agree with David Cheal from Strut Digital that moving between cloud platforms is not easy or terribly feasible at the moment. Danny Elmarji from EMC described this as the challenge of “data gravity” which is a very eloquent way of describing the realities of distributing your workload across a number of cloud providers or physical regions.
Cost management
Cost management was also touched upon. It is clear that this space is more mature in the AWS world than others. As much as organisations like to talk about spend optimisation, our experience has been that there are usually internal and organisational challenges to the implementation of cost-saving measures on cloud spend. Definitely a solvable problem but one that takes the correct mix of change management, technical knowledge and business drive.
Machine Learning and data science
The audience posed a few questions to the experts on the feature offerings around Machine Learning and Data Science application services, of which Microsoft, Google and AWS have existing offerings.
Wrapping up the debate
To round things up, the organisers arranged an informal sales pitch game where each of the teams had 60 seconds to pitch to a hypothetical CIO their platform, with the audience each having one vote (by hand raise) for their preference.
After the CIO pitch, the other member of the team had 60 seconds to pitch to a potential startup client. Once again we voted with our hands.
Clever. And while AWS came out on top in both exercises, it was far from a whitewash, with Google and Microsoft close behind in both of the competitions. In all honestly, EMC was a little disadvantaged here based on the nature of the questions with a slightly different offering, but they still gave it their all and are certainly one to watch in the hybrid/private cloud space.
Some phrases heard a number of times by various debaters:
-
“ … all depends on your architecture … ”
-
“ … using services vs using infrastructure … ”
-
“ … depending on the workload … ”
-
“ … essentially, … ”
-
“ … cloud native … ”
Next time let's hope that someone is up there representing OpenStack(external link), as every new release brings more and more exciting features to the fold.