Serverless and the Future of Computing
Co-Founder of Lune, Advisory Partner at Crane
I had the pleasure of moderating a London Enterprise Tech Meetup (LNETM) panel discussion about serverless and the future of computing. Ian Ellis assembled a top roster of panel speakers: Chris Swan (DXC Technology), Melika Golkaram (Google Cloud), and Paul Johnston (serverless advocate). To set the scene, we heard flash talks from Mark Hinkle at TriggerMesh and Taavi Rehemägi at Dashbird. With such an incredible group of people discussing a topic as exciting as serverless computing, I wanted to share some of the things I learned.
1. Serverless is a philosophy
Firstly, the definition of serverless is often misunderstood. It is the word “server” that throws people off. Serverless is more of a philosophy of how to approach technology. It is about building applications by focusing on the things that drive maximum business value over the entire application lifecycle. It is somewhat of a mental shift from “how easy is this for me to build?” to “how easy is this for my team to build and maintain over the coming years?”. Paul has written about serverless at length on his Medium page and specifically about what serverless is.
In practice, this means that developers gain more control of code deployment, there is no infrastructure maintenance required, apps are highly scalable, built on event-driven architectures, and they only cost to run when actually in use. Most importantly, however, serverless means that dev teams take a longer-term view and focus on delivering business value, rather than secondary activities.
2. Serverless is the hot glue gun for IT integrations
Whilst serverless computing is a nascent technology and has not yet seen mass adoption, many companies are actively reaping the benefits of it. For example, at DXC part of the technology doctrine is to think serverless first. If a specific problem can be solved by using serverless, it should be. It is the preferred option, after which engineers deploy to containers, then VMs, and lastly mainframes or other legacy infrastructure. Part of the reason for this hierarchy is that serverless is an extremely useful tool to integrate services together. When building microservices-based and event-driven architectures, there is always a need to connect countless services and systems.
Mark Hinkle and Sebastien Goasguen had this realisation and founded TriggerMesh to help companies build cloud native applications by offering an integration solution to connect legacy infrastructure with any cloud and serverless functions. They recently launched the beta version of EveryBridge, the only true cloud native integration platform in the market.
“Serverless is the space-filling hot glue for all kinds of IT integrations.” — Chris Swan
Take the media industry, which is riddled with complexity, with a myriad of content types (linear TV, streaming services, live events etc) that all have different requirements. Not only that, but there are multiple steps to make a stream consumable by consumers — from transforming the media format to encryption to CDNs and so on. When there is an event, such as a new episode becoming available, these services need to talk to each other and get triggered to run at the right time.
In addition, the media industry is all about front end experiences. The front page is changed constantly, whether it is because a new hot movie is out, the Olympics are kicking off or there is some other current event. As the front ends are updated, they need to ensure that any user, no matter where they are in the world, gets access to the right version of the app at the right time.
Finally, when dealing with content such as live sports, the infrastructure needs to cater to huge sudden demand spikes. This is the bread and butter of serverless computing, and many media companies have already realised how powerful it is for their use cases.
As the number of services and integrations grow, fully understanding IT environments — the performance, where errors are happening and why — becomes increasingly difficult. This is where Dashbird’s serverless monitoring platform helps companies continue serving their customers as well as possible.
3. A mental shift is required, but we are getting there!
Today we are still at a stage where many companies say they are serverless but as you look closer they eg. still have minimum-spend pricing models. Building really good serverless solutions require both a different skillset and a different mindset. It requires an understanding of how to build good apps today (eg. using serverless for integrations) but also building so that others can continue building on top, adding integrations in the future. This mental shift will take time, but we are well underway.
“There will be a culture war and the word serverless will lose meaning.” — Paul Johnston
Once more people embrace this new way of thinking, we will start seeing powerful applications. Melika cites the example of not deploying a database in containers anymore, but rather deploying a serverless database and calling an API from a service account in order to continuously capture data for analysis at a later stage. In the same spirit, one of our portfolio companies, Axiom, has built a serverless data insights platform, enabling companies to store and query unlimited amounts of data, all the time. Other exciting startups in this space include Vercel, for front end development, and Webiny, a serverless CMS.
To get started with serverless, companies should consider going cloud native, embracing the longer-term and business value driven thinking, and having a hierarchical technology doctrine that starts with serverless. Startups like TriggerMesh and Dashbird are here to help in this transition towards a modern IT architecture. We are still just scratching the surface!
I highly recommend checking out future LNETM events, they are still virtual and always high quality! Keep an eye out for future posts on the website. Thank you for a fantastic event Ian, Chris, Melika, Paul, Mark and Taavi!
At Crane, we are extremely fortunate and excited to have the opportunity to work with and support companies like TriggerMesh and Axiom, building the future of the enterprise technology stack.
If you are building a company in and around the serverless space or are excited about this space in general, please get in touch. We love nerding about next-gen infrastructure and cloud native solutions!