The Need to Roll up Your Sleeves for WebAssembly
We already know how putting applications in WebAssembly modules can improve runtime performance and latency speeds and compatibility when deployed. We also know that WebAssembly has been used to improve application performance when running on the browser on the backend. But the day when developers can create applications in the language of their choice for distribution across any environment simultaneously, whether it’s on Kubernetes clusters, servers, edge devices, etc. remains a work in progress.
This status quo became that much more apparent from the talks and impromptu meetings I had during KubeCon + CloudNativeCon in April. In addition to a growing number of WebAssembly module and service providers and startups offering support for WebAssembly, it’s hard to find any organization that is not getting down to work to at least see how it works as a sandbox project in wait of when customers will ask for or require it.
Many startups, established players and tool and platform providers are actively contributing to the common pool of knowledge by contributing or maintaining open source projects, taking part in efforts such as the ByteCode Alliance or sharing their knowledge and experiences at conferences, such as during the KubeCon + CloudNativeCon Europe’s co-located event Cloud Native Wasm Day. This collective effort will very likely serve as a catalyst so that WebAssembly will eventually soon move past its current status as just a very promising new technology and begin to be used for what it’s intended for on a massive industry scale.
Indeed, WebAssembly is the logical next step in the evolution from running applications on specific hardware, running them on virtual machines, to running them in containers on Kubernetes, Torsten Volk, an analyst at Enterprise Management Associates (EMA), said. “The payout in terms of increased developer productivity alone justifies the initial investments that come with achieving this ultimate level of abstraction between code and infrastructure. No more library hell: No more debugging app-specific infrastructure. No more refactoring of app code for edge deployments. In general, no more wasting developer time on stuff other than writing code,” Volk said. “This will get us to a state where we can truly compose new applications from existing components without having to worry about compatibility.”
Work to Be Done
But until we get that point of developer-productivity nirvana, work needs to be done. “Now we need all-popular Python libraries to work on WebAssembly and integrations with key components of modern distributed apps, such as NoSQL storage, asynchronous messaging, distributed tracing, caching, etc.,” Volk said. “Luckily there’s a growing number of startups completing the ‘grunt work’ for us to make 2024 the year when WebAssembly really takes off in production.”
Collaboration, alliances and harmony in the community, especially in the realm of open source, will be critical. “The one thing I’ve learned from the container wars is that we were fighting each other too early in the process. There was this mindset that the winner would take all, but the truth is the winner takes all the burden,” Kelsey Hightower, principal developer advocate, Google Cloud, said during the opening remarks at KubeCon + CloudNativeCon Europe’s Cloud Native Wasm Day. “You will be stuck trying to maintain the standards on behalf of everyone else. Remember collaboration is going to be super important — because the price for this has to be this invisible layer underneath that’s just doing all of this hard work.”
At the end of the day, those writing software probably just want to use their favorite language and framework in order to do it, Hightower said. “How compatible will you be with that? Or will we require them to rewrite all the software?” Hightower said. “My guess is anything that requires people to rewrite everything is doomed to fail, almost guaranteed and that there is no way that the world is going to stop innovating at the pace we’re on where the world will stop, and implement all the lower levels. So, it is a time to be excited, but understand what the goal is and make sure that this thing is usable and has tangible results along the way.”
During the sidelines of the conference, Peter Smails, senior vice president and general manager, enterprise container management, at SUSE, discussed how internal teams at SUSE shared an interest in Wasm without going into details about SUSE’s involvement. “WebAssembly has an incredibly exciting future and we see practical application of WebAssembly. I personally think of it as similar to being next-generation Java: it is a small, lightweight, fast development platform and, arguably, is an infrastructure that lets you write code and deploy it where you want and that’s pretty cool,” Smails told The New Stack.
In many ways, WebAssembly proponents face the chicken-before-the-egg challenges. After all, what developer would not want to be able to use the programming language of their choice to deploy applications for an environment or device without having to worry about configuration issues? What operations and security team would not appreciate a single path of deployment from finalized application code to deployment on any device or environment (including Kubernetes) security without the hassles of reconfiguring the application for each endpoint? But we are not there yet and many risks must be taken and investments made before wide-scale adoption really does happen the way it should in theory.
“We have a lot of people internally very excited about it, but practically speaking, we don’t have customers coming to talk about this asking for the requirements — that’s why it’s in the future,” Smails said. “We see it more as a potentially exciting space because we’re all about infrastructure.”
Get the Job Done
@ShopifyEng's @saulecabrera discussed Winch, a baseline compiler in Wasmtime. He says a baseline compiler is significantly faster than 'ahead-of-time' compilation optimization. Benchmarks to come.. #Wasm day #KubeConEU #kubeconeurope pic.twitter.com/Vq4wdiH6fe
— BC Gain (@bcamerongain) April 18, 2023
Meanwhile, there is a huge momentum to create, test and standardize the Wasm infrastructure to pave the way for mass adoption. This is thanks largely to the work of the open source community working on projects sponsored in-house or among new tool providers startups that continue to sprout up, as mentioned above. Among the more promising projects discussed during the KubeCon + CloudNativeCon co-located event Cloud Native Wasm Day, Saúl Cabrera, a staff developer, for Shopify, described how he is leading the development of Winch during his talk “The Road to Winch.” Winch is a compiler in Wasmtime created to improve application performance beyond what Wasm already provides. Offering an alternative to overcome the limitations of a baseline compiler, WebAssembly Intentionally-Non Optimizing Compiler and Host (Winch) improves startup times of WebAssembly applications, Cabrera said. Benchmarks result that demonstrates the touted performance metrics will be available in the near future, Cabrera said.
