Monitoring & Observability Features
Every check your stack needs.
HTTP, TCP, DNS, SSL, and keyword assertions. 5 regions. Checks every 30 seconds. AI pipeline observability in early access. One incident is one too many.
Currently in early access for engineering teams
One Platform, Two Endpoint Types
Every check below runs through the same pipeline, infrastructure or AI.
Modern infrastructure has two kinds of endpoints: the kind that always existed, servers, APIs, regions, and the kind that didn't five years ago: LLM APIs, inference endpoints, vector databases, agent pipelines. Most monitoring tools were built for the first kind. Teams shipping AI in production end up running a second, separate stack to watch the second. up0 doesn't split the two: the same 5 regions, the same incident pipeline, and the same status page cover both.
Infrastructure endpoints
AI endpoints
- LLM API latency & token throughput: per model, per provider
- Vector DB & embedding service health: the layer between the model and the data
Same regions. Same incidents. Same status page.
HTTP & HTTPS
Full HTTP monitoring with body assertions
Check status codes, response time, and keyword presence in a single request. A 200 that returns an error page is still a failure; up0 catches it.
- Assert on status code, keyword, or JSON path
- Custom headers, auth tokens, and POST bodies
- SSL certificate expiry alerts from 30 days out
- Redirect follow or block: your choice
- IPv4 and IPv6 probe support
DNS & SSL
Catch certificate failures before users do
SSL expiry and DNS resolution failures cause hard outages. up0 monitors both continuously, alerting you at 30, 14, and 7 days before expiry, and the moment a DNS record stops resolving.
- DNS A, AAAA, CNAME, MX, and TXT record checks
- SSL expiry alerts across configurable thresholds
- Certificate chain validation and issuer tracking
- Domain hijack detection via record change diffs
Multi-region coverage
5 regions. Zero single points of failure.
A single-region check misses regional BGP failures, CDN edge issues, and localized DNS problems. up0 confirms failures from 3 independent locations before alerting, eliminating false positives that wake your team at 3am.
AI Endpoints · LLM & Inference
Your LLM API is an endpoint. Monitor it like one.
AI inference endpoints have unique failure modes, token limits, context window errors, provider-side latency spikes, that a standard HTTP check won't catch. up0 monitors LLM APIs and the gateways in front of them the same way it monitors any other endpoint: from 5 regions, on the same check interval, through the same incident pipeline.
- LLM API latency and token throughput tracking, per model and provider
- Model gateway and proxy endpoint checks
- Inference queue depth alerting before it becomes user-facing latency
AI Endpoints · Vector & Embedding
Vector databases fail differently than a web server.
A vector database can be reachable and still return degraded query results. An embedding service can time out while the model serving it stays healthy. up0 checks vector databases and embedding services as their own endpoints, not as a side effect of the model that calls them, so an incident points at the layer that actually failed.
- Vector database health checks and query-latency monitoring: Pinecone, Weaviate, Qdrant, and similar stores
- Embedding service availability, monitored independently from the models that depend on it
Alerts & Integrations
Alerts that reach the right person, fast
Confirmation from 3 regions before firing. Multi-channel delivery. Escalation policies. Every alert lands with full context: region, latency, status code, and a link to the incident timeline.
Multi-channel delivery
Email, SMS, Slack, PagerDuty, Microsoft Teams, and webhooks. Configure per monitor or per team.
Escalation policies
Page on-call if the first responder does not acknowledge within 5 minutes. Fully configurable.
Alert confirmation
3 independent regions must agree before an alert fires. No more 3am wakeups from a blip in one datacenter.
Status pages
Public and private status pages with your domain. Subscribers get email alerts when incidents are posted.
Incident timeline
Every alert links to a full timeline: when each region failed, when it recovered, and total downtime.
Webhook payloads
Structured JSON payloads with full context. Wire up any tool: linear, jira, opsgenie, or a custom endpoint.
CLI & Terraform
Monitors as code
Define monitors in the same pull request as the infrastructure they watch. The up0 Terraform provider and CLI support full CRUD for monitors, channels, and status pages. No clicking through dashboards to onboard a new service.
- Terraform provider with full resource support
- CLI for scripting and CI/CD pipelines
- REST API with 10,000 req/min rate limit
- Official SDKs for Python, Node, and Go
monitor created id: mon_abc123 url: https://api.acme.com/health interval: 30s regions: us-east-1, eu-west-1 status: active
status: up uptime: 99.98% (30d) p50: 42ms p95: 89ms last_check: 8s ago
FAQ
Common questions
What types of monitors does up0 support?
up0 supports HTTP/HTTPS, TCP, DNS, and SSL certificate monitors. Each check can assert on status codes, response body keywords, response time thresholds, and SSL expiry windows.
How often does up0 check my endpoints?
Checks run every 30 seconds. All checks run from multiple regions simultaneously to eliminate false positives.
Does up0 support AI pipeline observability?
Yes, it's a core part of what we're building alongside uptime and latency monitoring, currently in early access. Dedicated monitor types for AI inference endpoints, LLM APIs, and vector database health track latency percentiles, token throughput, and embedding service availability separately from standard HTTP monitors.
Can I manage monitors with Terraform or the CLI?
Yes. up0 ships a Terraform provider and a CLI. Both support full CRUD for monitors, notification channels, and status pages. Treat your monitoring configuration the same way you treat infrastructure-as-code.
Get early access to up0
Join the waitlist and we'll email you when it's time to add your first endpoint, pick your regions, and get alerted the next time something breaks.