Comparisons
How Cardinal stacks up.
Vendor details were checked against public documentation and reviewed in August 2026. Product availability and pricing can vary by plan; if we've gotten something wrong, tell us and we'll fix it.
Data ownership
Where telemetry ingested into the product is stored at rest.
- Cardinal
- Your own object storage (S3 / GCS / Azure Blob)
- Datadog
- Datadog-managed cloud storage
- New Relic
- New Relic-managed cloud storage
- Splunk
- Splunk Cloud, or self-managed Splunk Enterprise
- Grafana Cloud
- Grafana Cloud backends, or self-managed Grafana stack
Customer-operated infrastructure
Infrastructure a customer operates to run the observability backend; not a total-cost comparison.
- Cardinal
- Bulk telemetry in object storage; stateless workers can use spot/preemptible compute. No SSD telemetry-data tier required.
- Datadog
- Managed SaaS; no observability backend infrastructure to operate.
- New Relic
- Managed SaaS; no observability backend infrastructure to operate.
- Splunk
- Splunk Enterprise can be self-managed; storage and compute are operator-managed.
- Grafana Cloud
- Grafana Cloud is managed; a self-hosted Grafana stack is operator-managed.
Instrumentation
Supported ways to collect and send telemetry.
- Cardinal
- OpenTelemetry-native; use your preferred collector
- Datadog
- OpenTelemetry supported; Datadog Agent and DDOT Collector options
- New Relic
- OpenTelemetry supported; New Relic agents and OTLP endpoints
- Splunk
- Splunk OpenTelemetry Collector; Universal Forwarder in some environments
- Grafana Cloud
- OpenTelemetry, Prometheus, and Grafana Alloy
Retention
How long data is retained and what controls that period.
- Cardinal
- Full fidelity in storage you operate; retention follows your policy
- Datadog
- Metrics: 15 months by default; log and trace retention varies by product and plan
- New Relic
- Configurable by data type and plan
- Splunk
- Configured per index; hot/warm/cold tiering
- Grafana Cloud
- Retention differs by signal and plan; logs have 14-day free and 30-day paid minimums, with extensions available
Sampling
Documented controls for reducing telemetry volume before or at ingestion.
- Cardinal
- No sampling by default; you choose whether and where to sample
- Datadog
- Trace-sampling controls for managing ingestion volume and cost
- New Relic
- SDK and collector controls for managing ingest volume
- Splunk
- Collector processors can filter, transform, or sample before export
- Grafana Cloud
- Adaptive Metrics and collector filtering help control volume
High-cardinality metrics
Documented pricing, limits, and controls for many unique metric series.
- Cardinal
- Indexed in your object storage; no per-tag surcharge
- Datadog
- Custom-metric billing is based on distinct metric/tag combinations
- New Relic
- Cardinality budgets and limits; additional capacity can incur cost
- Splunk
- Metric time-series limits and subscription entitlements; high cardinality can increase usage
- Grafana Cloud
- Active series and data points per minute affect metric pricing
AI and agent interfaces
Publicly documented AI assistance and interfaces. Availability can vary by product and plan.
- Cardinal
- Reusable troubleshooting skills and an MCP gateway
- Datadog
- Bits AI and a documented Datadog MCP server
- New Relic
- New Relic AI and MCP integrations
- Splunk
- AI Assistant and a documented MCP server
- Grafana Cloud
- Grafana Assistant and Cloud MCP connections
Deployment model
Where the managed service or self-hosted deployment runs.
- Cardinal
- Entire platform runs in your cloud against your object storage
- Datadog
- Managed SaaS
- New Relic
- Managed SaaS
- Splunk
- Splunk Cloud or self-managed Splunk Enterprise
- Grafana Cloud
- Grafana Cloud or self-managed Grafana stack
Pricing shape
How the bill scales with your footprint.
Read our pricing philosophy →- Cardinal
- Flat plans in your cloud (Free / Production / Enterprise) or per CCU in Cardinal's cloud — never per-GB
- Datadog
- Product-specific: hosts, containers, and/or telemetry usage
- New Relic
- Data ingest plus a user or compute model; add-ons may apply
- Splunk
- Workload-based or ingest-based, depending on subscription
- Grafana Cloud
- Signal usage (for example, active series, DPM, or GB); some products include host-hours
| Dimension | Cardinal | Datadog | New Relic | Splunk | Grafana Cloud |
|---|---|---|---|---|---|
Data ownership Where telemetry ingested into the product is stored at rest. | Your own object storage (S3 / GCS / Azure Blob) | Datadog-managed cloud storage | New Relic-managed cloud storage | Splunk Cloud, or self-managed Splunk Enterprise | Grafana Cloud backends, or self-managed Grafana stack |
Customer-operated infrastructure Infrastructure a customer operates to run the observability backend; not a total-cost comparison. | Bulk telemetry in object storage; stateless workers can use spot/preemptible compute. No SSD telemetry-data tier required. | Managed SaaS; no observability backend infrastructure to operate. | Managed SaaS; no observability backend infrastructure to operate. | Splunk Enterprise can be self-managed; storage and compute are operator-managed. | Grafana Cloud is managed; a self-hosted Grafana stack is operator-managed. |
Instrumentation Supported ways to collect and send telemetry. | OpenTelemetry-native; use your preferred collector | OpenTelemetry supported; Datadog Agent and DDOT Collector options | OpenTelemetry supported; New Relic agents and OTLP endpoints | Splunk OpenTelemetry Collector; Universal Forwarder in some environments | OpenTelemetry, Prometheus, and Grafana Alloy |
Retention How long data is retained and what controls that period. | Full fidelity in storage you operate; retention follows your policy | Metrics: 15 months by default; log and trace retention varies by product and plan | Configurable by data type and plan | Configured per index; hot/warm/cold tiering | Retention differs by signal and plan; logs have 14-day free and 30-day paid minimums, with extensions available |
Sampling Documented controls for reducing telemetry volume before or at ingestion. | No sampling by default; you choose whether and where to sample | Trace-sampling controls for managing ingestion volume and cost | SDK and collector controls for managing ingest volume | Collector processors can filter, transform, or sample before export | Adaptive Metrics and collector filtering help control volume |
High-cardinality metrics Documented pricing, limits, and controls for many unique metric series. | Indexed in your object storage; no per-tag surcharge | Custom-metric billing is based on distinct metric/tag combinations | Cardinality budgets and limits; additional capacity can incur cost | Metric time-series limits and subscription entitlements; high cardinality can increase usage | Active series and data points per minute affect metric pricing |
AI and agent interfaces Publicly documented AI assistance and interfaces. Availability can vary by product and plan. | Reusable troubleshooting skills and an MCP gateway | Bits AI and a documented Datadog MCP server | New Relic AI and MCP integrations | AI Assistant and a documented MCP server | Grafana Assistant and Cloud MCP connections |
Deployment model Where the managed service or self-hosted deployment runs. | Entire platform runs in your cloud against your object storage | Managed SaaS | Managed SaaS | Splunk Cloud or self-managed Splunk Enterprise | Grafana Cloud or self-managed Grafana stack |
Pricing shape How the bill scales with your footprint. Read our pricing philosophy → | Flat plans in your cloud (Free / Production / Enterprise) or per CCU in Cardinal's cloud — never per-GB | Product-specific: hosts, containers, and/or telemetry usage | Data ingest plus a user or compute model; add-ons may apply | Workload-based or ingest-based, depending on subscription | Signal usage (for example, active series, DPM, or GB); some products include host-hours |
Ready to compare in your environment?
The best comparison is your own data. Run Cardinal alongside your incumbent and see the difference before you switch.