{
  "id": 11017608,
  "title": "5x faster Edge Functions: V8 isolates to Firecracker MicroVMs",
  "url": "https://urgent.news/2026/09/30/5x-faster-edge-functions-v8-isolates-to-firecracker-microvms",
  "topic": "tech",
  "section": "Tech",
  "published": "2026-09-30T18:17:45.000Z",
  "source": {
    "name": "Hacker News",
    "slug": "hacker-news",
    "url": "https://www.netlify.com/blog/edge-functions-firecracker-microvms/"
  },
  "original_language": "en",
  "account": "Every day, approximately one billion Edge Functions run on Netlify, fulfilling various tasks such as personalization, routing, and authentication across hundreds of thousands of websites. The primary challenge is to minimize latency while handling tens or even hundreds of thousands of edge functions per second. To address this, Netlify has rebuilt its infrastructure in collaboration with Unikraft, resulting in Edge Functions running on MicroVMs (Micro Virtual Machines) within their edge network, delivering a 5x improvement in performance compared to the previous approach. This change enhances security, reliability, and enables complex compute at the edge without altering how Edge Functions are written or used. The infrastructure consists of an edge node, a compute node, and a MicroVM, all working together to process requests with minimal latency.\n\nWhen a request arrives at the edge node closest to the client, it is converted into a specification and routed to an available compute node. If the compute node already has a MicroVM associated with the service, the request is forwarded to it. The compute node checks whether the required images are already present on disk. If not, it fetches them from the edge node, ensuring that only the necessary images are retrieved for the region. The edge node creates a specification for the machine that will run the function, defining the required images, CPU, memory, and connection limits. This specification, along with its hash and site-specific information, becomes the service ID, allowing for isolation between different deployments. The service ID prevents multiple deploys with different code or environment variables from sharing a MicroVM, enhancing security by containing potential failures.\n\nTo optimize performance, the edge node uses rendezvous hashing to assign each service to a specific compute node, ensuring that the same MicroVM is used consistently for the same service. This method creates a caching strategy, minimizing cold starts and ensuring that hot spots do not form, which would lead to resource contention. If the traffic for a service becomes too heavy on a single node, the system redistributes the service across multiple nodes to maintain optimal performance and prevent any one node from becoming overloaded. The final step involves pulling in the function's code onto the compute node, which is already warmed up if it has been previously invoked.",
  "summary": null,
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}