Your app runs across three AWS regions, a managed Postgres instance, a Stripe integration, and an OpenAI dependency. Something slows down at 2 AM. Is it your containers? A regional network blip? A vendor degradation you have no control over? "Cloud monitoring" is supposed to answer that question, but the category is so broad that two tools both claiming the label can solve completely different problems.
Some tools ingest host metrics, traces, and logs from your infrastructure (CPU, memory, container restarts, slow queries). Others watch your services from the outside: endpoints, SSL, response bodies. They correlate failures with the third-party clouds you depend on. Most teams need both layers, and the mistake is buying one when you needed the other.
We compared nine tools across the dimensions that actually decide the purchase: infrastructure metrics, APM and tracing, log management, external endpoint checks, multi-cloud coverage, and the one that surprises finance teams: how the bill scales. This comparison maps each tool to the job it actually does, whether you're on a single provider or spanning AWS, Azure, and GCP. Every pricing model below was checked against official pages in July 2026.
| Tool | What it monitors | Best For | Model | Free Tier | Entry Price |
|---|---|---|---|---|---|
"Cloud monitoring" spans two problem domains that rarely live in the same product tier, so we scored each tool on both. Infrastructure telemetry: can it ingest metrics, traces, and logs from cloud hosts, containers, and managed services? External and dependency monitoring: can it check your endpoints from outside your cloud, validate responses, and tell you when a vendor (AWS, Stripe, a payments API) is the actual culprit? We also weighted multi-cloud coverage (does it work equally across AWS, Azure, and GCP, or is it locked to one provider?), alerting and correlation (how fast, and does it group related failures into one incident?), and pricing predictability (flat, per-host, per-GB, per-user, or consumption units, and what happens at scale). A tool that nails infra metrics but can't tell you a third-party API degraded is only doing half the job.
| Feature | DevHelm | Datadog | CloudWatch | Google Cloud Monitoring | Azure Monitor | New Relic | Dynatrace | Grafana Cloud | Better Stack |
|---|---|---|---|---|---|---|---|---|---|
| External endpoint checks | Yes | Yes (Synthetics) | Yes (Synthetics) | Yes (Uptime checks) | Yes (App Insights) | Yes | Yes | Yes (Synthetic) | Yes |
| Response body assertions | Yes | Yes | Limited | Limited | Limited | Yes | Yes | Yes | Keyword only |
| Third-party dependency correlation | Yes (resource groups) | Partial | No | No | No | Partial | Partial | No | Partial |
| Multi-cloud (AWS + Azure + GCP) | Yes (cloud-agnostic) | Yes | AWS only | GCP only | Azure only | Yes | Yes | Yes | Yes |
| Config-as-code (Terraform/CLI) | CLI, Terraform, SDK | Terraform, API | CloudFormation, Terraform | Terraform | Bicep, Terraform | Terraform, API | Terraform, API | Terraform, Jsonnet | Terraform |
| Built-in status page | Yes | No | No | No | No | No | No | No | Yes |
| Free tier | 50 monitors | Limited | Basic metrics | Generous allotment | Basic metrics | 100 GB/mo | Trial only | Yes | 10 monitors |
| Pricing model | Flat monthly | Per-host + usage | Pay-as-you-go | Pay-as-you-go | Pay-as-you-go | Data + per-user | Consumption | Usage-based | Per-seat + usage |
| Host / infra metrics | No | Yes | Yes (AWS) | Yes (GCP) | Yes (Azure) | Yes | Yes | Yes | Limited |
| APM / distributed tracing | No | Yes | Limited (X-Ray) | Yes (Cloud Trace) | Yes (App Insights) | Yes | Yes | Yes (Tempo) | No |
| Log management | No | Yes | Yes | Yes | Yes (Log Analytics) | Yes | Yes | Yes (Loki) | Yes |
| Container / Kubernetes monitoring | No | Yes | Yes (Container Insights) | Yes (GKE) | Yes (AKS) | Yes | Yes | Yes | No |
DevHelm monitors cloud-hosted applications from the outside and correlates their failures with the cloud services they depend on. It is not a CloudWatch replacement for host metrics. It is the black-box layer that answers "is my service actually working for users, and if not, is a vendor the reason?" Every monitor, including the free tier, supports HTTP checks with custom headers, request bodies, response body assertions, and SSL certificate monitoring.
The differentiator for cloud teams is resource groups. When Stripe degrades, your checkout monitor, billing webhook monitor, and subscription monitor would normally all fire separate alerts. Group them by the shared dependency and you get one alert that says "checkout is degraded, and the common factor is Stripe." Not seven pages at 2 AM. This is the alert-correlation problem that pure infra-metrics tools leave to you.
Because checks are defined through a CLI, Terraform provider, and SDKs (monitoring as code), your configuration lives in version control next to the infrastructure it watches:
devhelm monitor create --type http --url https://api.yourapp.com/health --interval 30s
devhelm monitor create --type http --url https://api.stripe.com/healthcheck --assert-status 200
Key strengths
Pricing
| Tier | Price | Monitors | Check Interval | Key Features |
|---|---|---|---|---|
| Free | $0/mo | 50 | 5 min | Response assertions, SSL monitoring, 1 status page |
| Starter | $12/mo | 75 | 1 min | All Free features + faster checks |
| Pro | $29/mo | 250 | 30 sec | All regions, resource groups, PagerDuty/Opsgenie |
| Team | $79/mo | 500 | 30 sec | SMS alerts, team management |
| Business | $249/mo | 2,000 | 30 sec | Unlimited team, white-label status pages |
Cost traps
Limitations
Best for: Teams running apps in the cloud who want fast external checks, response validation, and vendor-dependency correlation at flat pricing, layered on top of (or instead of) heavyweight infra tools.
Datadog is the default answer when someone says "cloud monitoring platform." It covers infrastructure metrics, APM and distributed tracing, log management, synthetic checks, real user monitoring, and security across AWS, Azure, and GCP with 800+ integrations. If you want one pane of glass for everything your cloud does, Datadog is the most complete option on this list.
That completeness is also the catch. Datadog's pricing is a stack of independent products, each metered differently: infrastructure per host, APM per host, logs per GB ingested and retained, synthetics per run, custom metrics per series. A mid-size cloud footprint routinely produces a five-figure monthly bill. The line items that grow fastest (custom metrics and log retention) are the hardest to predict in advance.
Key strengths
Pricing
| Component | Annual Price | Notes |
|---|---|---|
| Infrastructure | $15/host/mo | Entry point; per-host |
| APM | $31/host/mo | On top of infra |
| Log management | $0.10/GB ingest + retention | Retention billed separately |
| Synthetic API tests | $5/10k runs | Location multiplier applies |
| Custom metrics | Usage-based | Grows silently with cardinality |
Cost traps
Limitations
Best for: Well-resourced teams that want one platform for infrastructure, APM, and logs across multiple clouds and can manage the billing complexity.
Amazon CloudWatch is the native monitoring service for AWS. If your workloads run on EC2, Lambda, ECS, EKS, RDS, or virtually any AWS service, CloudWatch collects their metrics and logs with zero integration work. The data is already there. Alarms, dashboards, Logs Insights queries, and Synthetics canaries all live inside the AWS console and IAM model you already use.
The strength is also the boundary: CloudWatch is AWS-first. It can ingest custom metrics from anywhere, but its reason to exist is deep AWS coverage. Monitoring Azure or GCP through it is not the intended path. For an AWS-only shop, it's the lowest-friction option. For multi-cloud, it becomes one silo among several.
Key strengths
Pricing
| Component | Price | Notes |
|---|---|---|
| Basic metrics | Free | Standard AWS service metrics |
| Custom / detailed metrics | ~$0.30/metric/mo | First 10k, then tiered down |
| Dashboards | $3/dashboard/mo | Beyond the free 3 |
| Logs ingestion | ~$0.50/GB | Plus storage and Insights query cost |
| Synthetics canaries | ~$0.0012/run | Per canary run |
Cost traps
Limitations
Best for: AWS-only teams that want native, low-friction infrastructure metrics, logs, and alarms without adding a third-party vendor.
Google Cloud Monitoring (part of the Cloud Operations suite, formerly Stackdriver) is GCP's native observability layer. It collects metrics, logs (via Cloud Logging), and traces (via Cloud Trace) from GKE, Compute Engine, Cloud Run, and the rest of Google Cloud, with uptime checks for external endpoints and SLO monitoring built in.
Its uptime-check and SLO tooling is genuinely good. You can define service-level objectives, track error budgets, and alert on burn rate without bolting on another tool. As with the other hyperscaler tools, the trade-off is gravity: it's built to monitor Google Cloud, and using it as your primary monitor for workloads on other clouds fights the grain.
Key strengths
Pricing
| Component | Price | Notes |
|---|---|---|
| GCP metrics | Free allotment | Most Google Cloud metrics included |
| Chargeable metrics | ~$0.2580/MiB | Beyond the free allotment |
| Cloud Logging | ~$0.50/GB | First 50 GiB/project free |
| Uptime checks | Free tier | Then usage-based |
Cost traps
Limitations
Best for: GCP-centric teams that want native monitoring, logging, and SLO tracking without leaving Google Cloud.
Azure Monitor is Microsoft's native monitoring stack, combining platform metrics, Log Analytics (Kusto/KQL queries), and Application Insights for APM. For workloads on Azure VMs, AKS, App Service, and Functions, it collects telemetry natively and feeds a shared alerting and dashboarding layer.
Application Insights is the standout piece. It delivers real APM, distributed tracing, and availability tests for apps instrumented with its SDK. The cost center to watch is Log Analytics: nearly everything interesting flows through it, and it's billed per GB ingested, so verbose logging or broad diagnostic settings can drive the bill up faster than teams expect.
Key strengths
Pricing
| Component | Price | Notes |
|---|---|---|
| Platform metrics | Free | Standard Azure resource metrics |
| Log Analytics ingestion | ~$2.30/GB | The primary cost driver |
| Application Insights | Billed via Log Analytics | Ingestion-based |
| Alert rules | Per rule/mo | Metric and log alerts priced separately |
Cost traps
Limitations
Best for: Azure-centric teams that want native infrastructure metrics, APM via Application Insights, and KQL-powered log analytics.
New Relic rebuilt its pricing around data rather than hosts: you pay for gigabytes of telemetry ingested and for the number of full-platform users, not per monitored host. That change made it one of the more approachable all-in-one platforms for cloud monitoring. Infrastructure, APM, logs, synthetics, and browser monitoring under one meter, across all major clouds.
The free tier is unusually generous (100 GB of ingest per month and one full user), which makes New Relic viable for small teams doing real observability at zero cost. The two things to model carefully are data volume (high-cardinality custom events and verbose logs eat the GB allowance) and the per-full-user charge, which is where multi-engineer teams accrue cost.
Key strengths
Pricing
| Component | Price | Notes |
|---|---|---|
| Free tier | $0 | 100 GB/mo ingest, 1 full user |
| Data ingest | $0.35/GB | Beyond the free 100 GB |
| Standard full user | ~$99/user/mo | Core (limited) users cheaper |
| Data Plus | $0.55/GB | Higher-compliance ingest tier |
Cost traps
Limitations
Best for: Small-to-mid teams that want all-in-one observability with pricing that scales on data volume rather than host count.
Dynatrace targets the enterprise end of cloud monitoring with a heavy bet on automation and AI. Its OneAgent auto-discovers services and dependencies, and the Davis AI engine does automated root-cause analysis. Instead of a wall of correlated alerts, it aims to tell you the single underlying cause of an incident. For large, complex, multi-cloud estates, that automation is the core value proposition.
Pricing moved to a consumption model (Dynatrace Platform Subscription) metered in granular units for full-stack monitoring, log ingestion, and synthetics. It's genuinely capable and genuinely expensive. Dynatrace is rarely the choice for a small team, and there's no free tier (evaluation is a 15-day trial).
Key strengths
Pricing
| Component | Price | Notes |
|---|---|---|
| Full-stack monitoring | ~$0.08/host-hour (8 GB) | Consumption units (DPS) |
| Log management | ~$0.20/GB ingest + retention | Metered separately |
| Synthetic monitoring | Per-request units | Consumption-based |
| Free tier | None | 15-day trial only |
Cost traps
Limitations
Best for: Large enterprises with complex multi-cloud environments that want AI-driven root-cause analysis and can commit to consumption-based pricing.
Grafana Cloud is the managed version of the open-source observability stack: Prometheus-compatible metrics (Mimir), Loki for logs, Tempo for traces, and Grafana dashboards on top. If your team already lives in Prometheus and Grafana, this is the path to keep that stack without running the storage backends yourself, and it works across any cloud that can ship metrics.
The model is usage-based across metrics series, log volume, and trace volume, with a real free tier (10k active series, 50 GB logs, 50 GB traces, three users). The trade-offs are the ones inherent to the open-source stack: you assemble and tune it yourself, and Prometheus cardinality management becomes your responsibility. Active series are the meter, and a careless label can multiply them.
Key strengths
Pricing
| Component | Price | Notes |
|---|---|---|
| Free tier | $0 | 10k series, 50 GB logs, 50 GB traces, 3 users |
| Metrics | Usage-based | Per active series beyond free |
| Logs / traces | Usage-based | Per GB ingested beyond free |
| Pro / Advanced | From ~$8/mo base + usage | Scales with volume |
Cost traps
Limitations
Best for: Teams already invested in Prometheus and Grafana who want a managed, cloud-agnostic stack without operating the storage layer.
Better Stack bundles uptime monitoring, incident management, on-call scheduling, log management, and status pages into one dashboard. For cloud monitoring, it sits closer to the external/uptime layer than to deep infra telemetry. It watches your cloud-hosted endpoints, collects logs, and gives you on-call and a status page without stitching together separate products.
The consideration is per-seat pricing. Every person who manages monitors or responds to incidents is a "responder" seat, so cost scales with team size rather than infrastructure size. Solo developers and small teams get a lot from the all-in-one bundle; larger on-call rotations should total the seat cost before committing.
Key strengths
Pricing
| Component | Price | Included |
|---|---|---|
| Free | $0/mo | 10 monitors, 3-min intervals, 1 status page |
| Responder | $29/mo/seat (annual) | Monitoring, incidents, on-call |
| Additional monitors | $21/50 monitors | Added to any paid plan |
| Logs | Separate pricing | Ingestion + retention-based |
Cost traps
Limitations
Best for: Small teams that want uptime monitoring, logs, on-call, and a status page in one place and are comfortable with per-seat pricing.
The right cloud monitoring tool depends on which half of the problem you're solving. Most teams eventually need both layers.
Start with your cloud shape. If you're all-in on one provider, the native tool (CloudWatch, Google Cloud Monitoring, or Azure Monitor) gives you the deepest, lowest-friction infrastructure telemetry with no new vendor. The moment you're genuinely multi-cloud, a native tool becomes one silo among several, and a cloud-agnostic platform (Datadog, New Relic, Dynatrace, Grafana Cloud) earns its cost.
Separate infra telemetry from external checks. Host metrics, traces, and logs answer "what is my infrastructure doing?" External checks and dependency correlation answer "is my service working for users, and is a vendor the reason it isn't?" Datadog, New Relic, and Dynatrace cover both at a price. If you already have infra metrics covered by a native tool, adding a focused external layer like DevHelm or Better Stack is often cheaper than upgrading to a full-stack suite.
Model total cost at your real scale. "Starting at $15/host" is not the bill. Usage-based tools grow with metric cardinality and log volume; per-seat tools grow with headcount; consumption tools grow with units you can't easily predict. Run the numbers for your host count, data volume, team size, and check frequency before signing anything.
| Scenario | Recommended Tool | Why |
|---|---|---|
| AWS-only, want native infra metrics | Amazon CloudWatch | Zero-setup AWS telemetry, no extra vendor |
| GCP-only, want SLOs and traces | Google Cloud Monitoring | Native metrics, logs, and error-budget tracking |
| Azure-only, want APM + logs | Azure Monitor | Application Insights + KQL log analytics |
| Multi-cloud, want one full-stack pane | Datadog or New Relic | Infra + APM + logs across all clouds |
| Enterprise, want AI root-cause | Dynatrace | OneAgent + Davis automation at scale |
| Prometheus/Grafana shop | Grafana Cloud | Managed open-source stack, cloud-agnostic |
| Have infra metrics, need external + dependency layer | DevHelm | Endpoint checks + vendor correlation at flat pricing |
| Small team, want uptime + logs + on-call | Better Stack | All-in-one dashboard, per-seat pricing |
What is cloud monitoring?
Cloud monitoring is the practice of collecting metrics, logs, traces, and availability data from applications and infrastructure running in the cloud, then alerting when something degrades. In practice it splits into two layers: infrastructure monitoring (host, container, and service telemetry) and external monitoring (endpoint checks and third-party dependency correlation). Most teams end up running tools from both layers.
Which cloud monitoring software is best for AWS, Azure, or Google Cloud?
For single-provider shops, the native tool is usually the lowest-friction choice: CloudWatch for AWS, Azure Monitor for Azure, and Google Cloud Monitoring for GCP. They collect provider telemetry with no setup. The trade-off is lock-in. None of them is a strong multi-cloud pane of glass.
What are the best multi-cloud monitoring tools?
If you run workloads across two or more providers, a cloud-agnostic platform beats stitching native tools together. Datadog and New Relic are the most complete options for infra, APM, and logs across clouds. Grafana Cloud is the open-source-friendly choice. DevHelm covers the external layer (endpoint checks plus vendor-dependency correlation) without per-host pricing.
What should I use for Kubernetes and container monitoring?
For deep Kubernetes and container telemetry, Datadog, Dynatrace, New Relic, and Grafana Cloud (via Prometheus) all ingest pod, node, and cluster metrics. If you only need to know whether the services exposed by your cluster are reachable and correct from the outside, an external checker is a cheaper complement to cluster-level telemetry.
Is there open-source cloud monitoring software?
Yes. Prometheus with Grafana is the standard open-source stack, and Grafana Cloud is its managed form (Prometheus, Loki, Tempo). It's cloud-agnostic and covers metrics, logs, and traces, at the cost of more assembly and cardinality management than a turnkey tool.
How do I keep cloud monitoring costs under control?
Watch the meters that scale silently: metric cardinality, log ingestion and retention, per-host fees, and per-seat charges. Flat-rate tools (like DevHelm) and native pay-as-you-go services are the most predictable. Usage-based full-stack suites are the least. Model total cost at your real host count, data volume, and team size, not the advertised entry price.
What's the difference between cloud monitoring and observability?
Monitoring answers known questions ("is CPU high, is the endpoint up, did the error rate spike?") with metrics, dashboards, and alerts. Observability adds the ability to ask new questions after the fact by correlating high-cardinality metrics, traces, and logs. Full-stack platforms market themselves as observability; focused uptime and dependency tools market themselves as monitoring. Most teams need some of both.
What's the best cloud monitoring tool for startups?
Priorities for a startup or small SaaS team: a usable free tier, predictable pricing, and fast setup. New Relic's free tier and DevHelm's flat plans are both strong starting points. You get real coverage without an enterprise contract, and you can add infra-metrics depth later as you scale.
If you want to go deeper on one slice of cloud monitoring, these guides pick up where this comparison leaves off.
Compare more monitoring tools
Go deeper on the concepts
Every tool on this list can draw a graph of your cloud. The ones that matter are the ones that answer the question you actually ask during an incident: what broke, and can I fix it, or is it someone else's cloud? Infrastructure-metrics platforms tell you your containers are healthy. They're less helpful when your app is degraded because a third-party API you depend on is having a bad day. That's a failure inside your service that isn't your infrastructure's fault.
If your infra telemetry is already handled by a native tool or a full-stack suite, the missing layer is usually external checks with response validation and dependency correlation. Knowing that checkout is down and that the common thread is Stripe. If you want that layer without a per-host or per-GB bill, try DevHelm's free tier: 50 monitors, response assertions, and resource-group correlation at $0, no credit card. If you need deep infrastructure metrics, APM, and log analytics across clouds, Datadog, New Relic, or a native tool will serve you better.
The worst outcome isn't picking the "wrong" tool. It's buying a full-stack observability platform to solve an uptime problem, or bolting an uptime checker onto a system that needed tracing. Match the tool to the half of the problem you actually have, and add the other half only when you feel its absence.
Originally published on DevHelm.