EDGE COMPUTE / v4.2
Every edge on Earth,in 41 seconds.
One command deploys to 312 cities. No cold starts, no config files. Fast, by default.
zsh — ~/hello
$ npx kernel deploy› building …… 1.8s› uploading to 312 locations› health checks passed✓ live at https://hello.kernel.app (41s) - TYOTokyo6ms
- SINSingapore14ms
- SYDSydney21ms
- FRAFrankfurt11ms
- LHRLondon12ms
- IADVirginia9ms
- SFOSan Francisco10ms
- GRUSão Paulo18ms
p50 latency 18ms
01312edge cities
020mscold start
0318msglobal p50
0499.999%SLA
01 / CODE
You write the function. That’s it.
TypeScript, Python or Go. Keep your language, forget the infrastructure.
import { edge } from '@kernel/sdk'
export default edge({
regions: 'all',
cache: { ttl: '5m' },
async fetch(req) {
const user = await req.kv.get(req.ip)
return Response.json({ hello: user ?? 'world' })
},
})from kernel import edge
@edge(regions="all", cache={"ttl": "5m"})
async def handler(req):
user = await req.kv.get(req.ip)
return {"hello": user or "world"}package main
import "kernel.dev/edge"
func main() {
edge.Serve(edge.Config{Regions: "all"}, func(r *edge.Req) any {
user, _ := r.KV.Get(r.IP)
return map[string]any{"hello": user}
})
} 02 / ARCHITECTURE
The path of a request.
Every hop is measured and shown on your dashboard.
- 00User
- 01Nearest edge
- 02Function runtime
- 03KV / cache
- 04Origin (optional)
03 / CHANGELOG
Changelog
v4.2Python runtime is generally availablev4.1Region-pinned deploymentsv4.0KV reads are 3× fasterv3.9Go SDK released
Ship hello world to the world.
$ npx kernel deploy