The Future is Serverless

“I believe it is the biggest shift we have seen in the last 20 years, or maybe even more. It completely changes how we design, build and operate infrastructure services,” Tomasz Stachlewski says. The Amazon Web Services (AWS) Head of Technology for the EMEA and CEE is referring to the paradoxically named ‘serverless computing’. This approach to the cloud eliminates the need for companies to spend considerable resources, monetary or otherwise, on managing a server when building functionalities. They can simply upload a code or run an application, paying only for what they use – it’s the user who is serverless, not the application. “Amazon Lambda, the most popular serverless solution, has 1 million free requests per month,” he goes on. “I know companies that are operating on serverless and they are not paying anything, pretty much. With serverless, the prices are going down – that’s why it’s so great. It would not be possible to do that with a typical infrastructure.”
To fully understand why he considers serverless to be the next step in the evolution of cloud infrastructure services, we need to talk first about the cloud. By his own account, the Novartis user case grounds the nebulous service in practical reality. In 2013, the pharmaceutical heavyweight wanted to run a project that involved the virtual screening of 10 million compounds against a common cancer target. When crunching the numbers for the operation, it estimated that its own servers would need to work round-the-clock for 39 years, while cutting that time short would require some US$40 million in infrastructure investments. Novartis turned to the AWS cloud instead: 39 years of computational chemistry were completed in 9 hours by spinning 10,600 servers for just US$4,232.
Our interview is taking place in a ballroom at The Venetian, the main venue for the AWS re:Invent 2022 which sprawls across five other hotels on the famous Las Vegas Strip. Even though the room is cordoned off and only accessible to attendees wearing press tags, it is bursting at the seams: many interviews are happening concurrently, others with spare time are industriously typing on their laptops, and some are catching the World Cup match between Croatia and Canada in the TV corner. At the other end of the room, the hotel staff is rolling in carts with an all-you-can-eat buffet – I could not have thought of a more apt (and serendipitous) analogy for the cloud.
Stachlewski started at AWS as a solutions architect seven years ago and, during that time, he has gathered an astonishing amount of 14 AWS certifications: industry credentials that validate a level of AWS cloud expertise. While he jokingly says, “It is like collecting Pokémon,” there is a practical element to his being a philomath: in his role, he works closely with AWS clients and helps them build their application systems, so he needs to keep up with the rapid changes in the cloud industry. “I want to know what I’m talking about,” he notes succinctly. Having worked on numerous projects in the past seven years, then, Stachlewski has observed that the value of the cloud extends well beyond efficiencies in economics and operations. He compares the over 200 AWS cloud services – computing, storage, databases, machine learning, analytics, IoT, you name it – to LEGO blocks that can be stacked on top of each other in imaginative ways to build new and exciting solutions at incredible speeds. “This is actually the beauty of the cloud. And you wouldn’t have access to these LEGOs in your own data centre,” he says. On top of that, employees appear happier when working for companies that have migrated their operations to the cloud. It all comes down to people enjoying playing with the cloud’s new, cool toys while feeling a sense of security knowing their employers are looking into the future.
“Now, I’d like to play the devil’s advocate,” I say. Almost two decades after AWS released the Amazon Elastic Compute Cloud, industry knowledge of the cloud’s value proposition has matured, and a growing number of mid-sized companies, for which speed is no longer of the essence, claim that cloud costs weigh heavily on their balance sheet. One of these companies is the software development firm Basecamp. The company’s co-founder, David Heinemeier Hansson, who also developed the popular Ruby on Rails framework, had written in an op-ed that, for companies like his own, operating on the cloud is like “paying a quarter of your house’s value for earthquake insurance when you don’t live anywhere near a fault line.” For Stachlewski, on the other hand, while costs are an important consideration for any company, companies must also calculate in their decision whether they want to wrestle with maintaining an on-premise data centre or servers in general.
In 2008, Netflix experienced a major database corruption and had no means to ship DVDs to its members, which forced them to make some hard choices – they could no longer rely on their data centre. So, they decided to move to the reliable, scalable, and distributed systems of AWS’s cloud. “They wanted to work with a partner that had the biggest experience and was best in this,” Stachlewski mentions. In 2014, the streaming company decided to leverage AWS Lambda. By going serverless, Netflix was able to automate the encoding and organisation of the thousands of files that publishers uploaded on its platform daily, validate whether those files were backed up properly, and ensure that the thousands of stop-and-start processes were construed and configured in accordance with the system’s rules and regulations. Most importantly, Lambda removed the responsibility of managing its servers from Netflix and allowed it to focus on expanding its video-on-demand platform. By 2016, Netflix had eight times more subscribers, with overall viewing increasing by three orders of magnitude. “I have seen too many projects where 50% of the time was spent on managing the infrastructure,” Stachlewski says. “And of course, you could set up your data centre and could be happy, but you will not have easy access to those LEGOs that are the difference between you and your competitors.”
Besides Netflix, Disney+ has also decided to outsource the maintenance and management of its servers to AWS. And as Amazon Prime battles the two on who is going to take the biggest piece of the streaming pie, this brings up the question of whether the boundaries between competitor and collaborator are collapsing.
“I wouldn’t go that far,” he says. “Everything depends on relations and here the relations are very good. The evidence from the last 15 years shows that AWS is pretty much the best partner to work with. And that’s why so many companies, including competitors, are using us to build and to host their application systems.”