We’re excited to announce Turbo Replay! Turbo is our biggest release since Replay for Windows last November. Replays will start faster. Print Statements will evaluate faster. In short, Replay’s been turbo-charged 🚀
Before we jump into how Turbo works, we first have to share some context into how Replay works. When you’re viewing a replay, we’re replaying the runtime in our backend. Chrome is running as it normally does, except instead of talking to the OS, it’s talking with the recording.
We run replays on some of the most computationally intensive EC2 instances (100 cores, 1,000GBs of ram). Each recording runs in a Docker container that can use many threads and up to 50 GBs of ram.
In an ideal world, you would distribute the computation across many containers. The problem is that software execution is not easily split. This is where Turbo comes in!
Turbo lets us run recordings across multiple containers and distribute the compute across several EC2 instances. This leads to 3X faster Start times and significantly faster Print Statements.
We’d love to hear what you think. Let us know where Replay feels slow and we’ll fix it!
With Turbo Replay shipping we thought it would be fun to share a long term roadmap for replayability. We often talk about replayability from the perspective of Time Travel Debugging, but the implications are far broader.
Node.js 16X we rebased Replay Node onto v16.14 last week. Node is still in beta, so if you play with it, we’d love to hear what you think in Discord.
