Enterprise observability platforms
A scenario-led evidence matrix for six multi-signal platforms, with explicit capability states, three cost workloads, portability constraints, strengths, cautions, and unresolved commercial evidence.
Read the page as a transparent path: establish context, inspect the evidence, then test the published stance.
Observability technology map
Coverage, cost, and market breadth—in one view
Y capability families · X total monthly reference cost · Bubble breadth index
View all plotted technologies as a table
| Technology | Capabilities | Total monthly cost | Breadth | Relative: capability / cost / breadth | Latest evidence |
|---|---|---|---|---|---|
| SolarWinds Observability | 12/12 | $1023/month | 98/100 | 90% / 86% / 89% | Market signal |
| Grafana Ecosystem | 12/12 | $480/month | 98/100 | 90% / 63% / 89% | Market signal |
| Grafana OSS | 3/12 | $90/month | 98/100 | 16% / 42% / 89% | Market signal |
| Grafana Cloud | 11/12 | $760/month | 97/100 | 84% / 84% / 88% | Report |
| Prometheus + OpenTelemetry + Grafana Stack | 9/12 | $300/month | 96/100 | 59% / 52% / 80% | Market signal |
| Amazon CloudWatch | 10/12 | $410/month | 96/100 | 67% / 68% / 80% | Market signal |
| OpenTelemetry | 3/12 | $120/month | 96/100 | 16% / 63% / 80% | Market signal |
| Elastic Observability | 10/12 | $150/month | 94/100 | 67% / 21% / 75% | Market signal |
| Datadog | 12/12 | $1620/month | 93/100 | 90% / 93% / 72% | Report |
| Azure Monitor | 10/12 | $460/month | 92/100 | 67% / 72% / 67% | Market signal |
| BMC Helix | 8/12 | $1050/month | 92/100 | 51% / 91% / 67% | Market signal |
| Prometheus | 3/12 | $105/month | 91/100 | 16% / 56% / 65% | Market signal |
| Fluent Bit | 2/12 | $70/month | 90/100 | 0% / 56% / 57% | Market signal |
| Datadog Agent | 3/12 | $75/month | 90/100 | 16% / 35% / 57% | Market signal |
| Google Cloud Observability | 9/12 | $180/month | 89/100 | 59% / 28% / 56% | Market signal |
| Prometheus Blackbox Exporter | 2/12 | $60/month | 88/100 | 0% / 42% / 49% | Market signal |
| Datadog Vector | 3/12 | $90/month | 88/100 | 16% / 42% / 49% | Market signal |
| New Relic | 12/12 | $300/month | 88/100 | 90% / 35% / 49% | Report |
| Prometheus Alertmanager | 3/12 | $60/month | 86/100 | 16% / 28% / 46% | Market signal |
| Splunk Observability | 11/12 | $750/month | 86/100 | 84% / 83% / 46% | Market signal |
| Grafana Loki | 2/12 | $105/month | 86/100 | 0% / 75% / 46% | Market signal |
| Dynatrace | 12/12 | $650/month | 84/100 | 90% / 78% / 42% | Report |
| Grafana k6 | 2/12 | $80/month | 84/100 | 0% / 63% / 42% | Market signal |
| OpenSearch | 8/12 | $180/month | 83/100 | 51% / 33% / 41% | Market signal |
| OpenObserve | 10/12 | $275/month | 82/100 | 67% / 41% / 36% | Market signal |
| IBM Instana Observability | 12/12 | $212/month | 82/100 | 90% / 26% / 36% | Market signal |
| VictoriaMetrics | 6/12 | $225/month | 80/100 | 41% / 59% / 33% | Market signal |
| Coralogix | 11/12 | $200/month | 80/100 | 84% / 27% / 33% | Market signal |
| LogicMonitor | 10/12 | $530/month | 77/100 | 67% / 77% / 31% | Market signal |
| Sumo Logic | 10/12 | $1438/month | 76/100 | 67% / 94% / 28% | Market signal |
| Papertrail | 3/12 | $20/month | 75/100 | 16% / 15% / 25% | Market signal |
| Grafana Pyroscope | 2/12 | $75/month | 75/100 | 0% / 59% / 25% | Market signal |
| Oracle Cloud Observability and Management | 8/12 | $855/month | 74/100 | 51% / 89% / 23% | Market signal |
| ClickStack | 9/12 | $240/month | 72/100 | 59% / 40% / 19% | Market signal |
| Better Stack | 9/12 | $300/month | 72/100 | 59% / 52% / 19% | Market signal |
| Logstash | 2/12 | $60/month | 72/100 | 0% / 42% / 19% | Market signal |
| Cribl | 8/12 | $500/month | 72/100 | 51% / 80% / 19% | Market signal |
| Datadog Cloudcraft | 3/12 | $49/month | 70/100 | 16% / 22% / 17% | Market signal |
| Grafana Mimir | 3/12 | $120/month | 68/100 | 16% / 63% / 15% | Market signal |
| Grafana Tempo | 2/12 | $90/month | 68/100 | 0% / 69% / 15% | Market signal |
| Chronosphere | 10/12 | $900/month | 66/100 | 67% / 88% / 14% | Market signal |
| OpenNMS | 4/12 | $180/month | 65/100 | 35% / 69% / 12% | Market signal |
| Honeycomb | 7/12 | $150/month | 61/100 | 47% / 32% / 11% | Market signal |
| Grafana Alloy | 2/12 | $75/month | 58/100 | 0% / 59% / 9% | Market signal |
| ScienceLogic SL1 | 7/12 | $1050/month | 56/100 | 47% / 96% / 7% | Market signal |
| BigPanda | 6/12 | $900/month | 55/100 | 41% / 96% / 5% | Market signal |
| Apica | 10/12 | $0/month | 55/100 | 67% / 0% / 5% | Market signal |
| ITRS Geneos | 6/12 | $950/month | 52/100 | 41% / 99% / 4% | Market signal |
| Digma | 5/12 | $40/month | 50/100 | 39% / 16% / 1% | Market signal |
| Grafana Pages | 4/12 | $0/month | 20/100 | 35% / 0% / 0% | Market signal |
Audience, scenario, and cohort
This comparison is for an enterprise platform team selecting a multi-signal observability operating model for applications, infrastructure, user experience, and reliability workflows. The initial cohort contains Datadog [1], Dynatrace [2], Elastic Observability [3], Grafana Cloud [4], New Relic [5], and Splunk Observability Cloud [6]. Each has documented multi-signal collection, exploration, alerting, and production support. The cohort does not include collectors, databases, consulting providers, GitOps tools, mobile-only products, or workflow engines because those entities do not satisfy the same substitution test.
Evidence boundary
The matrix below records whether current vendor documentation supports a capability. It does not verify performance, usability, correctness, availability in a particular plan, or suitability for a workload. Supported means usable primary documentation was found. Partial means scope, product, export surface, plan, or operating model limits a general statement. Unknown means evidence is insufficient. Unknowns are not scored as zero or replaced by a midpoint.
Cost and total ownership
A credible cost comparison requires telemetry volume by signal, host and container counts, session or device volume, retention, cardinality, query use, archive, transfer, premium features, support, infrastructure, engineering labor, currency, region, and pricing date. The interactive model uses small, growing, and enterprise workloads to expose cost structure. Its rates are illustrative assumptions and are not vendor quotes.
Product-specific totals remain Quote required / research pending until each product can be calculated under the same dated workload. This prevents a public list price for one signal from being compared with a negotiated bundle, a free allowance, or a self-hosted software license that excludes infrastructure and labor.
Ease-of-use study design
The next measured round will time the same tasks for every product: first useful telemetry, service discovery, trace-to-log investigation, alert creation and tuning, role-specific dashboard use, upgrade or configuration maintenance, raw-data export, and a standard exit test. Results will record evaluator experience, environment, sample size, failed attempts, context switching, and uncertainty. Until those tasks are run, ease-of-use statements remain documented capability or editorial assessment rather than measured fact.
Portability and exit tests
OpenTelemetry intake is useful but narrower than platform portability. A complete exit test must export recent and archived telemetry, reproduce representative queries, dashboards, alerts, service-level objectives, workflows, access controls, and integrations, then measure time, loss, fees, and operational interruption. The portability matrix separates these assets so open ingestion cannot silently become a claim of zero lock-in.
Reading the result
The paired strengths and cautions panels apply the same evidence threshold to every platform. A strength in one scenario may be irrelevant elsewhere; a constraint may be acceptable when its operational benefit is greater than its switching cost. No centered, average, or highest composite score is treated as inherently correct.
Capability heatmap
Evidence state, not winner badges
| Platform | Multi-signal ingest | Telemetry optimization | Exploration and correlation | Cost controls | LLM and AI observability | Digital experience | Alerting and response | SRE workflows | Self-hosted option | Raw telemetry export |
|---|---|---|---|---|---|---|---|---|---|---|
| Datadog ↗Service reviewed 2026-08-10 | Supported | Supported | Supported | Partial | Supported | Supported | Supported | Supported | Not supported | Partial |
| Dynatrace ↗Latest service documentation | Supported | Supported | Supported | Partial | Supported | Supported | Supported | Supported | Partial | Partial |
| Elastic Observability ↗Current Elastic documentation | Supported | Supported | Supported | Supported | Supported | Supported | Supported | Supported | Supported | Supported |
| Grafana Cloud ↗Service reviewed 2026-08-10 | Supported | Supported | Supported | Partial | Partial | Supported | Supported | Supported | Not applicable | Supported |
| New Relic ↗Service reviewed 2026-08-10 | Supported | Partial | Supported | Partial | Supported | Supported | Supported | Supported | Not supported | Partial |
| Splunk Observability Cloud ↗Service reviewed 2026-08-10 | Supported | Partial | Supported | Partial | Partial | Supported | Supported | Supported | Not supported | Partial |
Scenario cost model
Three workloads, explicit assumptions
20 hosts, 60 containers, 100 GB logs/month, 50 GB traces/month, 13-month metrics retention, five users.
| Input | Managed model | Composable model |
|---|---|---|
| Telemetry processing | $0.12 / GB | $0.01 / GB infrastructure allowance |
| Host allowance | $8 / host | $18 / host infrastructure |
| Named users | $12 / user | Included in labor model |
| Support allowance | 12% of direct cost | 5% of direct cost |
| Operations labor | 4 h × $85 | 16 h × $85 |
These rates are transparent editorial assumptions for comparing cost structure—not Datadog, Dynatrace, Elastic, Grafana, New Relic, or Splunk prices. Taxes, discounts, egress, archive, premium features, regional differences, and contract minimums remain excluded. Product comparisons must replace this model with dated evidence or display “Quote required.”
Portability and lock-in
Switching constraints by asset
| Platform | OpenTelemetry ingest | Non-proprietary instrumentation | Raw telemetry export | Query portability | Dashboard portability | Alert portability | Deployment control | Exit-cost evidence |
|---|---|---|---|---|---|---|---|---|
| Datadog | Supported | Partial | Partial | Partial | Partial | Partial | Not supported | Unknown |
| Dynatrace | Supported | Partial | Partial | Partial | Partial | Partial | Partial | Unknown |
| Elastic Observability | Supported | Supported | Supported | Partial | Partial | Partial | Supported | Unknown |
| Grafana Cloud | Supported | Supported | Supported | Partial | Supported | Partial | Partial | Unknown |
| New Relic | Supported | Partial | Partial | Partial | Partial | Partial | Not supported | Unknown |
| Splunk Observability Cloud | Supported | Supported | Partial | Partial | Partial | Partial | Not supported | Unknown |
Open ingestion reduces one form of coupling. It does not automatically move queries, dashboards, alerts, workflow definitions, historical data, access policy, or commercial commitments. Proprietary capability may add value while increasing switching work.
Ease-of-use task protocol
Comparable work, measured before scoring
| Representative task | Persona | Start state | Success evidence | Time | Errors | Assistance |
|---|---|---|---|---|---|---|
| Connect an OpenTelemetry Collector | Platform engineer | Fresh tenant and sample service | Traces, metrics, and logs visibly linked | Not measured | Not measured | Not measured |
| Investigate a latency regression | On-call engineer | Alert plus 30-minute telemetry window | Affected service, cause hypothesis, and evidence trail | Not measured | Not measured | Not measured |
| Create an SLO and burn alert | SRE | Defined availability objective | Valid SLI, windows, notification, and test | Not measured | Not measured | Not measured |
| Reduce high-volume telemetry | Telemetry owner | Cost anomaly and sample data | Documented policy with measured retained value | Not measured | Not measured | Not measured |
| Export data and saved assets | Migration lead | Representative dashboards, alerts, and history | Portable files plus declared manual gaps | Not measured | Not measured | Not measured |
A result is published only after the same starting data, persona assumptions, success criteria, stopwatch rule, error taxonomy, documentation access, and assistance policy are used for every platform. “Not measured” remains visible and cannot be converted to zero or a favorable score.
Reference architectures
Control boundaries change the work
Managed platform
Customer owns instrumentation, policy, identity integration, and response. Provider owns the managed control plane and backends within the contracted boundary.
Composable stack
Customer gains component and deployment control while assuming sizing, upgrades, backup, cross-component identity, availability, and integration operations.
| Boundary | Managed platform | Composable stack |
|---|---|---|
| Backend operation | Provider under service terms | Customer or chosen operator |
| Data path control | Collector plus provider ingestion rules | Collector, pipeline, and storage policy |
| Scaling and upgrade | Mostly provider; customer validates change | Customer coordinates every component |
| Exit work | Export, saved assets, history, contract | Component formats, schemas, operations, migrations |
Datadog
Evidence base ↗Evidence-backed strengths
- Broad documented product coverage across infrastructure, applications, logs, user experience, security, and AI workloads.
- Unified managed operating model reduces backend ownership for teams that accept SaaS control boundaries.
Cautions and constraints
- Product breadth creates many plan and billing dimensions; a workload model is required before cost comparison.
- Saved queries, dashboards, monitors, workflows, and historical data require an explicit exit test rather than an OpenTelemetry-only portability claim.
Dynatrace
Evidence base ↗Evidence-backed strengths
- Documented full-stack observation, topology, causal analysis, digital experience, and automation are integrated around a shared data platform.
- Managed and customer-controlled deployment choices exist, subject to exact edition and architecture.
Cautions and constraints
- DQL, Grail, dashboards, workflows, and Davis-derived context can increase switching effort even when ingestion is open.
- DPS pricing and entitlements need a dated workload calculation; quote terms remain unknown.
Elastic Observability
Evidence base ↗Evidence-backed strengths
- Documented SaaS, serverless, and self-managed deployment paths provide comparatively broad infrastructure control.
- OpenTelemetry intake, tiered storage, search, and multi-signal use cases are documented in one platform.
Cautions and constraints
- Elastic Stack operation, upgrades, index lifecycle, and capacity become customer responsibilities in self-managed deployments.
- ES|QL, data mappings, Kibana content, alert rules, and Elastic-specific processing remain migration work.
Grafana Cloud
Evidence base ↗Evidence-backed strengths
- Strong documented alignment with Prometheus, Loki, Tempo, Pyroscope, Grafana, and OpenTelemetry ecosystems.
- Managed backends can reduce operating load while preserving several open protocols and content formats.
Cautions and constraints
- Grafana Cloud is not identical to a self-operated LGTM stack; managed-service limits, features, and support require separate review.
- PromQL, LogQL, TraceQL, dashboards, alerting, and hosted retention still create migration and operations work.
New Relic
Evidence base ↗Evidence-backed strengths
- Documented application, infrastructure, browser, mobile, logs, and AI monitoring share a managed platform and NRDB query model.
- OpenTelemetry and native agents provide multiple ingestion approaches.
Cautions and constraints
- NRQL, entity models, dashboards, alerts, workflows, and historical data require an explicit migration plan.
- Usage, user, retention, and product entitlements must be modeled at a dated plan.
Splunk Observability Cloud
Evidence base ↗Evidence-backed strengths
- Documented infrastructure monitoring, APM, RUM, synthetics, and incident intelligence cover a broad managed scenario.
- OpenTelemetry Collector distributions support an open collection entry point.
Cautions and constraints
- Splunk Observability Cloud and Splunk log products are distinct; comparisons must state which products and contracts are included.
- SignalFlow, detectors, dashboards, integrations, and cross-product data paths require migration testing.
Source record
Bibliography
Datadog, ‘Datadog documentation.’ [Online]. Available: https://docs.datadoghq.com/. Accessed: Aug. 10, 2026.
Dynatrace, ‘Observe,’ Feb. 6, 2023. [Online]. Available: https://docs.dynatrace.com/docs/observe. Accessed: Aug. 10, 2026.
Elastic, ‘Elastic Observability solution overview.’ [Online]. Available: https://www.elastic.co/docs/solutions/observability. Accessed: Aug. 10, 2026.
Grafana Labs, ‘Grafana Cloud documentation.’ [Online]. Available: https://grafana.com/docs/grafana-cloud/. Accessed: Aug. 10, 2026.
New Relic, ‘Get started with New Relic.’ [Online]. Available: https://docs.newrelic.com/docs/new-relic-solutions/get-started/intro-new-relic/. Accessed: Aug. 10, 2026.
Cisco Splunk, ‘Splunk Observability Cloud documentation.’ [Online]. Available: https://help.splunk.com/en/splunk-observability-cloud. Accessed: Aug. 10, 2026.
Known limits
Unresolved unknowns
- Dated normalized product prices are not yet available for the three reference workloads.
- No common performance benchmark has been completed.
- Contract minimums discounts support terms and regional availability are unknown.
- Exit tests for saved queries dashboards alerts workflows and historical data remain outstanding.
Public review
Discuss and correct
Comments are public. Proposed factual corrections must be incorporated into version-controlled research before they are treated as published.
Local preview does not have Giscus configured. Native comments and Giscus are never presented as competing primary systems on the same page.