Ship pixels and APIs
at the edge—everywhere.
Celeris routes users to the closest edge location via global Anycast, caches intelligently, and fails over automatically—guided end-to-end by Celeris AI.
Routing you can see. Policies you can trust.
Clients connect via global Anycast to the closest healthy edge, requests flow through caching and routing layers, and traffic automatically fails over when issues arise.
api.example.com
Drag to simulate traffic weight changes between revisions
Cache Invalidation
Purge by path, wildcard, or tag—propagates globally in ~50ms
Health-weighted routing: Traffic automatically routes to the closest healthy region based on active health checks. Failover happens at zone → region → global levels.
Simulate Failure
Click a region on the map to manually fail it
Incident Timeline
Everything you need at the edge
Smart caching, instant purge, origin protection, and multi-region routing—all orchestrated by Celeris AI.
- Cache key controls (query, headers, cookies)
- TTL, negative caching, SWR/stale-if-error
- Compression + HTTP/2, HTTP/3
- Purge by path, wildcard, or tags
- Propagation in seconds globally
- Batch invalidation support
- Shield region for cache-fill locality
- Request coalescing (collapse misses)
- Circuit protection at edge
- Signed URLs/cookies for access control
- Range requests, large file delivery
- Short-lived tokens, rotation policies
- Closest healthy selection + health checks
- Active-active + weighted routing
- Cloud-native L7 load balancing
- REST, gRPC, WebSockets, SSE
- Long-lived connections + backpressure
- HTTP/2 + HTTP/3 where available
Cache keys, TTL, SWR, HTTP/2+3
Purge by path, wildcard, or tags
Shield region, request coalescing
Signed URLs/cookies, range requests
Closest healthy, L7 load balancing
REST, gRPC, WebSockets, HTTP/3
Ask for performance. Get a policy.
Celeris AI proposes caching policies, explains cache misses, simulates changes, and generates optimized configurations—all from natural language.
Edge Performance Assistant
Try asking:
Click a prompt above to start
—
Hit Ratio
+12%
—
Origin RPS
-35%
—
p95 Latency
-18ms
AI-generated dashboards per app
EU Latency Deep Dive — Generated Jan 7, 2026
EU p50 Latency
38ms
EU p99 Latency
142ms
Shield Hit Rate
89%
AI Summary: EU latency improved 23% after enabling origin shielding in eu-west. The p99 spike correlates with /api/search cache misses (query param variance). Consider normalizing cache keys for this endpoint. Shield is reducing origin load effectively at 89% hit rate.
Celeris AI goes beyond suggestions
Generate custom analysis dashboards per app, detect bottlenecks automatically, explain cache miss patterns, and produce deployment-ready configurations—all with governance and approval workflows built in.
I'll create a caching policy that:
- ✓ Caches static assets (JS, CSS, images) for 7 days with immutable headers
- ✓ Bypasses cache entirely for /api/* paths
- ✓ Enables stale-while-revalidate for HTML pages
Estimated improvement: +15% hit ratio, -40% origin load
I analyzed EU cache miss patterns and found:
Issue: Vary: Authorization on /assets/*
This creates unique cache keys per user, defeating caching for public assets.
Recommendation:
- • Split public assets to
/static/*without auth headers - • Configure 30-day TTL for /static/*
- • Keep /assets/* for authenticated content
For peak traffic, I recommend enabling origin shielding:
-
⬡
Configure
us-centralas the shield region for NA traffic - ⬡ Enable request coalescing to collapse concurrent misses
-
⬡
Set
stale-if-error: 60sfor graceful degradation
Expected improvement during peak:
Origin RPS reduced by ~65%, p99 latency improved by ~40ms
I'll create a signed URL policy for private downloads:
-
🔒
Apply to
/downloads/*and/private/* -
🔒
Token expiry:
15 minutes(configurable) - 🔒 Include IP binding for additional security
- 🔒 Automatic key rotation every 30 days
This policy supports range requests for large file resumable downloads.
Top cache-miss reasons (last 24h):
💡 Normalizing query params on /api/search could improve hit ratio by ~8%.
Edge analytics that actually helps
Real-time cache hit ratios, traffic distribution, origin health, and AI-powered miss explanations—all in one place.
Cache Hit Ratio
87.3%Requests Served from Edge
2.4M
↑ 12% vs yesterday
Bandwidth Saved
847 GB
↑ 8% vs yesterday
Avg TTFB
42ms
↓ 5ms vs yesterday
Top Cacheable Paths
/static/app.js
HIT 99%
/api/products
HIT 45%
/images/hero.webp
HIT 97%
Why Miss? Breakdown
Requests by Region
Protocol Distribution
Rate Limit Events (24h)
1,247 blockedTop blocked IPs and paths are available in the detailed dashboard.
Origin Latency
127ms
↓ 15ms from shield optimization
Error Rate
0.02%
Well within SLO target
Shield Effectiveness
73%
Origin requests reduced
Request Coalescing (24h)
ActiveRequests Coalesced
142K
Origin Calls Saved
89K
Failovers (30d)
2
Both auto-recovered
Invalidations (30d)
47
Avg propagation: 1.2s
Config Deploys (30d)
23
0 rollbacks
Recent Events
Zonal failover: eu-west-1a → eu-west-1b
Jan 5, 2026 at 14:32 UTC • Duration: 2m 15s
Cache purge: /api/products/*
Jan 6, 2026 at 09:15 UTC • Propagation: 0.8s
Config deploy: Shield enabled for us-central
Jan 6, 2026 at 16:42 UTC
AI Summary: What changed? What recovered?
The zonal failover on Jan 5 was triggered by increased error rates in eu-west-1a (network congestion). Traffic automatically shifted to eu-west-1b within 15 seconds. Full recovery after upstream issue resolved. Suggested guardrail: Consider enabling circuit breakers with 5% error threshold for faster detection.
Cache Inspector
/static/app.js
Request Headers
Response Headers
Cache Key
Served From
TTL Remaining
Protection at the edge
WAF, DDoS mitigation, and rate limiting integrate seamlessly with Gateway and Zero Trust. TLS termination with automatic certificate lifecycle happens at the edge—before traffic reaches your origins.
WAF + DDoS Protection
OWASP Core Rule Set, custom rules, and volumetric attack mitigation at the edge
Rate Limiting & Bot Protection
Per-IP, per-path, and custom dimension rate limits with bot score integration
mTLS & Certificate Management
Automatic cert issuance/rotation, HSTS enforcement, and origin authentication
Audit Logs & Compliance
Full request logging, config history, and compliance exports (SOC2, HIPAA)
Config as code. Instant preview.
Define edge policies in your repo, preview per PR, and deploy globally with confidence—all through GitOps workflows.
Preview Environments
Test edge policies per PR
Every PR gets its own preview edge policy. Test caching rules, routing changes, and failover configurations before merging.
pr-847.preview.edge.celeris.io
CLI Commands
Works With Your Stack
Standard cache headers, SDK middleware optional
GitOps Native
Config in repo → PR review → Global rollout. Same workflow as your application code.
Connect your origins, instantly
Bring your services, storage, and frameworks. Celeris Edge works with any HTTP origin.
Bring Your Origins
Kubernetes services, Functions, VMs, or any HTTP endpoint as your origin.
Storage Origins
Serve static assets directly from S3-compatible storage with automatic caching.
Framework Support
First-class support for modern frameworks with automatic cache headers.
DNS & Certificates
Bring your DNS provider or use Celeris DNS. Auto TLS certificates included.
Observability Export
Stream edge logs and metrics to your existing observability stack.
One-Click Enablement
Enable edge delivery for any service in your catalog with a single click.
Browse Marketplace
Make the edge part of your platform,
not a pile of configs.
Routing, caching, shielding, failover—guided by Celeris AI.
Trusted by teams at