El Segundo, CA
When their team said that they needed access to high-powered Dedicated Servers to compare web frameworks on, we immediately offered them a hand.
“With Deft, we didn’t have to worry about the hardware at all. It was magically ready to go. All we had to do was install our software and get data.”
For us, lending free server space and expertise was a way to give back to the open source community, which has given us so much knowledge.
No one was making money off the project, but everyone was benefitting from the data that the tech company was publishing.
Our cloud transformation team already uses the company’s framework comparison to learn what the apps they re-architect are capable of, and to find the fastest web framework for a particular application. More importantly, now, anyone else can do the same. Supporting open-source framework performance benchmarks gives others the chance to build better performing apps, too.
When those high performance web frameworks meet up with the compute power Deft can deliver, that’s when we can do our most eye-popping work.
How better framework performance helps everyone — and why it has to be open source
The tech company is a software consulting firm, so testing the performance of various frameworks is certainly not core to their business. As with Deft, though, speed matters, and the team was taken with the idea of giving back to the open source community.
“We see scenarios all the time where a business is successful, but the software team can’t catch the application up to deal with additional load,” said their CTO. “Horizontally scaling usually isn’t an option forever — it’s just necessary to make the application faster. And that can be challenging when the framework limits the performance.”
To prevent the problem before it reaches their desks, they decided to test framework performance for the most popular web development languages. That way, dev teams could make performance a part of their decisions from the very start of the project.
“We want to direct people to an architecture that gives them the greatest headroom for performance,” the CTO said. “If you choose a high-performance framework to begin with, your capacity to grow is much greater.”
“We realized people can make more informed decisions if they have performance data at the start, so we set out to generate reliable data. Deft gave us the hardware we needed to get it done so anyone has the opportunity to make their applications faster.”
While they could have theoretically charged for its framework performance results, the team saw it as an opportunity for them to give back to the open source community. That proved valuable in its own right, giving them a Github presence and helping out with recruiting.
It was also the only way to get people to trust their performance benchmarks.
“In order for this to have credibility, it needed to be reproducible,” they said. “Making it open source allowed anyone to download and run the tests themselves, and to contribute improvements.”
What raw server power and a 10-gig connection will do to your web framework comparison
From the beginning, the company had always planned to provide a web framework performance comparison in two environments: virtual and physical. In part that was personal preference — the CTO and his teammates find unconstrained physical hardware appealing.
But it was also a practical consideration. You can’t really see what a high-performance web framework is capable of until you can give it enough room to truly run.
“Our in-house hardware was limited by our network. Deft gave us the compute power to truly exercise the frameworks, without a 1-gigabit network constraining the top performers. Deft’s 10-gig network, with its ultra-low latency and great performance, shows us just how fast applications can run.”
In their tests, bare-metal performance still substantially outperformed most cloud environments for the same spend.