Why Amazon is growing into the sky and how you can dust off with cloud computing when it counts.
Installing software updates?
Not too long ago, the employees of many companies occasionally (perhaps every few weeks) received a message telling them that IT would be unavailable for a few hours at night on the weekend because new software needed to be installed.
At that time, one could actually imagine a single system administrator who would be in danger of having to spend the night in the event of any unexpected incidents in order to manually insert disks in the noisy and stuffy server room and “feed” the physical server.
Amazon can install up to 86,400 software updates per day or even more on the fly without ever having to shut down a service.
Why has Amazon grown so much?
There are probably several reasons why Amazon has grown so massively over the last few years. Many are business related:
- For example, the fact that Amazon has never distributed profits to its shareholders, but has always reinvested them in its own growth. Which is in contrast to the usual shareholder value concepts.
- Or that Amazon is known for its frugality, i.e. the sparseness in office equipment and buildings – and in employee salaries (of course).
- Or that presentation slides (Microsoft Powerpoint!) are forbidden at Amazon, and instead employees have to write before meetings “narrative” text documents, which must all be read by the other meeting participants before the meeting. At first glance, this costs time and is more reminiscent of an essay competition at a middle school than a tough American company under time pressure. But this makes the meetings efficient and everyone is on the ball.
- Or the fact that Jeff Bezos, the richest person on this planet, likes to make his children pancakes in the morning without hurrying, likes to have his first important meetings (only) at 10am, doesn't want to make important decisions after 5pm and makes sure he gets enough sleep.
- And finally, the consistent support of our core competence logistics with IT skills, which does not shy away from developing and opening up new concepts and approaches.
- And which has led to a clean mapping of individual service areas by IT across companies and (!) the strict coupling of these areas through interfaces (Application Programming Interfaces – APIs). The takeover of the organic food wholesaler Whole Food was also handled without any problems and IT implementation was completed within a short time.
Consistent strategic IT management
Perhaps not all, but certainly some business success criteria are backed by a well thought-out IT architecture and strategic IT principles. One, if not _the tangible technical reason and the actual basis for Amazon's rapid growth is, of course, the fact that Amazon has the technical ability to expand and adapt its IT as needed.
Ambidextric and agile IT management
This and the above-mentioned agile provision and adaptation in a matter of seconds is only possible because operations and software development are not separated from each other, but are closely interlinked in terms of organization, or almost seamlessly integrated. Amazon invented this concept, also known as “DevOps” (as a combination of “development” and “operation”) and paved the way for this form of ambidextric DevOps IT and culture.
Ambidextric literally means “two handedness”. What is meant by this is the combination of stability and flexibility, which is also known from Asian martial arts: Being robust – rooted to the ground – standing there and yet being able to act and move at lightning speed, regularly collapse there as an ideal goal of the sport. «Maneuverability» also plays a role in real combat, as the Energy-maneuverability-Theory developed by John Boyd proves.
Amazon has perfected such virtues with its Amazon Web Services.
Agile is not necessarily lean or DevOps
The agile software development as a general trend was certainly also helpful for this: If the software is already developed agile, it can also be provided more agile. But this is not enough at all: Agile “software development” is simply … Software development” and does not say anything about how the software is delivered. You can develop software agile but still deploy it eventually.
Agile software development in the form of Scrum is also clocked in weekly or even monthly intervals (Timeboxing). The result is “released” only after these time periods have expired. Before these intervals expire – in direct contrast to DevOps – a deployment is simply not planned.
What AWS can do through the example of Zoom
The Corona pandemic caused Zoom to experience an increase in sales that even analyses of this magnitude did not expect: “The biggest beat we have witnessed in covering software for over 20 years." (“The biggest beat we have witnessed in covering software for over 20 years."), an analyst is quoted by J. P. Morgan.
Instead of the expected approximately $200 million, Zoom had earned 328 million US dollars in the quarter, and a further increase to $500 million is still expected. The high number of users of the free Zoom license, which only runs for up to 40 minutes per session, is seen as a key success factor and has contributed significantly to the resounding popularity and awareness during the crisis.
The ease of use and high communication quality of Zoom is certainly another success factor. The really exciting question, however, is how Zoom managed to provide the massively increased user traffic and the many free users during the pandemic without external indications of weaknesses or service failures.
As CEO Eric Yuan in the quarterly stock market briefing announced, there was a twentyfold (!) increase in meeting minutes: “We had an approximately twentyfold increase in our metric of annualized meeting minutes run rate, which jumped from 100 billion at the end of January 2020 to over 2 trillion meeting minutes, based on April 2020's run rate.
How was it possible for Zoom to increase its capacity at almost any time during the pandemic and to fully exploit this business success without downtime?
The explanation is – you guessed it – Amazon's AWS and the so-called elasticity of cloud computing.
Without building data centers and installing servers, Zoom was able to quickly purchase additional resources from Amazon and withstand this onslaught without major problems – and to get rid of these resources at any time afterwards without being bound to fixed costs in the expected reduction (through superfluous server farms). The magic word for this means: the elasticity of the cloud. This means that the required performance of the cloud is virtually scalable to a certain extent.
The magic word for this means: the elasticity of the cloud.
As Ben Thompson correctly writes (subscription) this would not have been possible in this extreme form just a few years ago. Due to the consistently marginal costs by other users, Zoom was even able to increase the demand by generous free subscriptions (!). And besides that, Amazon was able to support other customers in this situation – and guarantee (!) its own operation without having to report failures and interruptions.
The elasticity of the cloud is of course also being used by Amazon's “market companions”, above all Microsoft with its teams and Office and Google, which, despite an equally high level of stress during the pandemic, showed no sign of weakness whatsoever.
DevOps is therefore more than just a buzzword, but the prerequisite for being crisis-proof with IT and growing successfully.
Zoom not only withstood this onslaught, but also responded accurately and quickly to the security concerns that arose in the midst of the situation and immediately launched a 90-day program to focus on improving security aspects – DevOps at it's best!
- Berls, Bruce: Zoom's Pandemic Success Story
- Thompson, Ben: Zoom's Earnings, Zoom and AWS, Zoom and End-to-end Encryption
- Ebenhoch, Peter: How safe is Zooming?
- Ebenhoch, Peter: Zoom conforms to data protection
- Zoom Inc: Zoom End-to-End Encryption Whitepaper