Managed OpenSearch & Prometheus
Managed OpenSearch and Prometheus on dedicated bare metal in Europe. Ingest what you like, query it as often as you want, add seats freely. The bill is the node rate times the node count, and nothing else.
Ten days, one Basic cluster, no card required.
Two services, one platform
Ship your logs here from anywhere, then search them with Dashboards included, plus k-NN and hybrid search for retrieval-augmented workloads on the same cluster.
Metrics and alerting for your infrastructure and services. Point
remote_write at it in one line, and your existing
PromQL queries and dashboards keep working.
What production costs
$199.98 / month
Two Standard nodes at 4 vCPU and 32 GiB each, a terabyte of storage across the cluster. Coordinator nodes and Dashboards included. That's the whole bill.
Never on your invoice
Four ways in
The console is the fastest way to a running cluster. Everything it does, the Terraform provider, the REST API and the MCP server do too: same organizations, same clusters, same credentials. Nothing is console-only.
resource "clusternest_opensearch" "logs" {
name = "production-logs"
organization_id = data.clusternest_organization.main.id
tier = "standard"
version = "3.7.0"
opensearch_whitelist = ["203.0.113.0/24"]
}curl -X POST \
https://api.clusternest.com/cluster/opensearch/ \
-H "Authorization: Bearer $CLUSTERNEST_TOKEN" \
-H "Content-Type: application/json" \
-d '{
"name": "production-logs",
"organization_id": 42,
"tier": "standard",
"version": "3.7.0",
"opensearch_whitelist": ["203.0.113.0/24"]
}'claude mcp add --transport http \
--client-id mcp --scope user \
clusternest https://api.clusternest.com/mcp
claude "rotate the admin credentials on
production-logs and show me the new ones"
rotate_opensearch_cluster_credentials
get_opensearch_cluster_credentials
Rotated. The new admin password is in the
reply above, and nothing else changed.Works with your agents
Ask in plain language and the agent writes the query. PromQL, OpenSearch DSL, PPL and SQL have been public and documented for years, so it gets them right without examples from you. The control plane is the same: a public OpenAPI spec, a Terraform provider, and an MCP server that puts every operation in front of an agent as a tool. Connecting is a server URL and your normal login. Nothing to install, nothing to run alongside it.
$ claude "was the website slow yesterday?"
production-metrics · PromQL
histogram_quantile(0.95, sum by (le) (
rate(http_request_duration_seconds_bucket[5m])))
production-logs · PPL
source=logs-* | where status >= 500
| stats count() by span(@timestamp, 1h)
p95 held at 180ms until 14:20, then sat above 1.4s
for 35 minutes. 502s from checkout spiked across the
same window.If you'd rather not do it alone
Index and retention design, the migration itself, dashboards and alerts rebuilt, then an ongoing retainer if you want one. Quoted per engagement, on top of the same clusters everyone else runs.
Talk to usEnd-to-end reviews of your application or service: architecture, capacity, cost, scalability, security and instrumentation, wherever it runs. Plus on-prem Kubernetes built and kept current, bare-metal migrations, and support for estates you already have. Independent of the managed service.
ConsultingTen days on a Basic cluster, no card. Upgrade whenever the trial ends, or don't, and nothing happens.