Dynatrace builds its AI-powered observability platform around four proprietary components: Grail (its data lakehouse), Smartscape (real-time topology mapping), OneAgent (automatic instrumentation), and Davis AI (causal and predictive analysis plus a generative layer). Gartner named it a Leader in the 2025 Magic Quadrant for Observability Platforms for the 15th consecutive time, with the highest position for Ability to Execute. The platform earns that recognition, but its cost structure and architecture push many engineering teams to evaluate alternatives.
Why look for Dynatrace alternatives
Teams leave Dynatrace when the platform’s architecture creates budget and data-control constraints. Instrumentation adds a second operational constraint: OneAgent couples visibility to the vendor. Davis AI and telemetry processing run on vendor infrastructure. These are direct consequences of how Dynatrace prices and builds the platform.
Unpredictable ingestion-based costs
Dynatrace bills through the Dynatrace Platform Subscription (DPS), a consumption model charged hourly across host-hours, memory-GiB-hours, pod-hours, and per-GiB ingest. Log analytics alone charges $0.20/GiB to ingest and process, plus retention and query costs on top. A traffic spike during an incident multiplies your bill at the exact moment you need more data, not less.
This creates a structural problem: the more you observe, the more you pay. Teams respond with aggressive sampling and low-priority log drops; some skip non-production instrumentation to control spend. Budget planning becomes guesswork when the bill depends on log volume, datapoint counts, and trace-scanned gigabytes that shift week to week. Typical enterprise commitments start at $120,000 to $180,000 per year, and scaling from a handful of environments to hundreds of nodes does not scale linearly.
Agent and instrumentation overhead
OneAgent installs at the host or container level and auto-discovers processes, which is convenient until you need to leave. Full-stack mode adds per-service instrumentation, and the default Kubernetes DaemonSet config requests 512Mi and limits to 1.5Gi of memory per node. Reddit users report real-world consumption spiking to 1.5G RAM with 500 to 700MB average.
The overhead numbers are contested. Dynatrace community posts state CPU stays below 3 to 5% under load, while one independent benchmark (single blog, undisclosed methodology) measured 22% CPU and 6% memory for OneAgent versus 1.8% and 0.4% for an eBPF solution. Dynatrace's own docs state overhead "cannot be expressed as a single constant number." The operational burden compounds beyond raw resource use: every Kubernetes upgrade triggers agent compatibility checks, and instrumenting new services can require developer coordination that turns the platform team into a bottleneck.
Data residency and vendor-hosted AI
Dynatrace SaaS stores your telemetry in AWS, Azure, or GCP data centers that Dynatrace operates, with region selection at provisioning time. EU options exist (Frankfurt, Netherlands, Zurich), but your data still transits and rests on vendor-controlled infrastructure. For fintech, insurtech, and healthcare teams, that architecture requires separate compliance review of cross-border transfer terms and Standard Contractual Clauses.
Davis AI processes your telemetry on Dynatrace infrastructure as well. Its causal AI is topology-driven and deterministic, built on Smartscape's real-time call graph, which is a real capability. OneAgent auto-instrumentation feeds that graph, so the analysis and the data it analyzes both live inside the vendor boundary. Regulated teams that cannot send production telemetry to third-party AI processing are structurally blocked.
Modular pricing and lock-in
Dynatrace bundles more than Datadog does, but charges still layer across host units, container units, RUM sessions, synthetic actions, and ingest volume. RUM runs $2.25 per 1,000 sessions; RUM with Session Replay runs $4.50 per 1,000 sessions, and browser monitors run $4.50 per 1,000 synthetic actions. Each dimension is a separate line to forecast.
Lock-in runs deeper than pricing. Historical data cannot be migrated between Dynatrace tenants, only configuration. Metrics can only be written with timestamps up to one hour in the past, which blocks bulk historical import into a replacement system. Smartscape topology retention on Grail is fixed at 35 days. Davis causal AI has no direct equivalent elsewhere, so leaving means rebuilding alert logic on explicit thresholds. These are switching costs that grow every quarter and weaken your renewal leverage.
Top Dynatrace alternatives at a glance
Five platforms consistently appear on Dynatrace shortlists, each solving a different subset of the cost, instrumentation, and residency problems above. The table summarizes where each fits:
Top Dynatrace alternatives in depth
Each platform below addresses Dynatrace's weaknesses differently. groundcover attacks cost, instrumentation, and residency through architecture. Datadog and New Relic keep the SaaS agent model and compete on breadth and consumption flexibility. Grafana Cloud leans on open source. Chronosphere focuses on cost governance for high-volume Kubernetes shops.
groundcover
groundcover replaces Dynatrace's agent-and-ingestion model with a single eBPF sensor, BYOC data residency by default, and flat per-node pricing that never scales with data volume. groundcover was built for Kubernetes and Linux workloads from inception.
The BYOC architecture is split-plane. The data plane, meaning all compute, storage, and telemetry, runs entirely inside your own cloud account. groundcover manages the control plane remotely: UI, APIs, and orchestration. Your observability data never leaves your environment.
Collection runs through the Flora eBPF sensor, deployed as a Kubernetes DaemonSet, one pod per node. Flora captures metrics, traces, logs, and Kubernetes events directly from the Linux kernel by inspecting every packet each service sends and receives. No SDKs. No language-specific agents. No application restarts. Every capability, from infrastructure monitoring to APM to LLM observability, comes included in every tier.
Quick facts:
- Founded August 2021, headquartered in Tel Aviv and California.
- Flat per-node pricing: Free ($0/host/month, BYOC), Pro ($30/host/month, BYOC), Pro+ ($35/host/month, BYOC), Enterprise on-premises ($50/host/month). Verify current rates before purchase.
- G2 rating 4.8/5 across 26 reviews (88% five-star).
- Available on AWS Marketplace, Google Cloud Marketplace, and Azure Marketplace.
- Storage on ClickHouse (logs, traces, Kubernetes events) and VictoriaMetrics (metrics, full PromQL compatibility), both deployed in your environment.
- Holds SOC 2 Type II and ISO 27001; HIPAA and GDPR listed on AWS and Google Cloud solution pages.
groundcover vs Dynatrace
The two platforms differ on every structural axis a Dynatrace evaluator cares about:
What users say
groundcover's BigBasket customer story reports a 50% observability cost reduction while expanding coverage. In the BigBasket customer story, Sushant Gulati, Senior Engineering Manager at BigBasket, said the company cut costs in half and increased observability coverage in testing environments. The mechanism is direct: flat per-node pricing means adding testing environments does not change the unit economics.
groundcover's Afida customer story says Afida expanded access from roughly 10 licenses to around 100 users without incremental cost. In the Afida customer story, Andreas Kappler, Senior Cloud Engineer at Afida, described moving from about 10 New Relic licenses to roughly 100 users in groundcover. groundcover does not charge per seat. User count is irrelevant to the bill.
Strengths: groundcover gives Kubernetes teams flat per-node cost that a 10x incident log spike does not change. Flora provides zero application instrumentation and full-cluster visibility within hours, with no instrumentation sprint. BYOC data residency is available at every tier, including free, and all capabilities are included without modular charges or per-user fees.
Tradeoffs: groundcover's scope is intentionally limited to Linux and Kubernetes workloads. It is also a newer entrant, with 26 G2 reviews and 32 GetApp reviews, fewer than incumbents.
If you want to validate the zero-instrumentation claim on your own cluster, the free tier includes the full BYOC architecture: deploy Flora on a single cluster and reach full-cluster visibility within hours, no credit card required.
Datadog
Datadog competes on breadth. It has broad integration coverage among the platforms compared here and covers nearly every signal type, at the cost of a modular pricing structure that requires careful modeling across multiple SKUs.
Datadog delivers observability as a SaaS platform with agent-based instrumentation. You install the Datadog Agent, then add product modules for infrastructure, APM, logs, and more, each priced separately. That separation is the tradeoff: you buy exactly what you use, but every capability is a distinct line item, and traffic spikes multiply costs across several SKUs at once.
Quick facts:
- Infrastructure monitoring: $15/host/month (Pro) or $23/host/month (Enterprise), billed annually.
- APM: $31 to $40/host/month depending on tier; Standard includes 1M indexed spans and 150 GB ingested spans.
- Log management: $0.10/GB ingestion, plus indexing at $1.27 to $2.50 per 1M events depending on retention.
- G2 rating 4.3/5 across 545 reviews.
- Named a Leader in the Forrester Wave: AIOps Platforms, Q2 2025, with highest scores in log management, APM, and cloud/infrastructure.
- Available on AWS, GCP, and Azure Marketplaces, with confirmed commit drawdown on Azure.
Datadog vs Dynatrace
Both publish detailed rate cards, but the billing philosophy diverges:
What users say
Practitioner comparisons consistently flag Datadog's "bill shock" risk: modular SKUs mean a traffic spike multiplies costs across infrastructure, APM, and logs at the same time. At smaller scale with selective capabilities, that same modularity makes Datadog cheaper than bundled competitors. Datadog's own G2 standing (4.3/5 across 545 reviews) reflects broad adoption across use cases.
Strengths: Datadog offers broad integration coverage among the platforms compared here. It has confirmed marketplace commit drawdown on Azure (MACC), listings on AWS and GCP marketplaces, and strong AIOps recognition as a Forrester Wave Leader in Q2 2025.
Tradeoffs: Datadog's modular pricing compounds at scale and resists forecasting. SaaS-only delivery means telemetry rests on vendor infrastructure, and per-product access controls add complexity as usage grows.
New Relic
New Relic prices on data ingest plus users rather than hosts, and it offers the most generous free tier of any platform here: 100 GB of ingest per month, one full-platform user, and unlimited basic users, perpetually.
New Relic is a full-stack APM and observability platform delivered as SaaS with agent-based instrumentation. Its consumption model charges for data you send and for the engineers who need full-platform access. Teams that keep ingest disciplined and limit full-platform access pay less; teams that roll access out broadly pay more.
Quick facts:
- Free tier (perpetual): 100 GB ingest/month, 1 free full-platform user, unlimited basic users.
- Data ingest beyond free: $0.30/GB (original) or $0.50/GB (Data Plus); EU data center add-on +$0.05/GB/month.
- New Relic charges full-platform users at $10 for the first (Standard), then $99 for up to 5; Pro runs $349 annual per user.
- G2 rating 4.4/5 across 511 reviews.
- New Pay-As-You-Go pricing structure took effect September 1, 2025.
- Azure Marketplace private offers allow 100% of New Relic spend to retire MACC.
New Relic vs Dynatrace
The pricing axis is the sharpest contrast, since New Relic charges per full-platform user while Dynatrace includes unlimited user seats:
What users say
New Relic's free tier and per-user model make it accessible for small teams and individual developers, which its 4.4/5 G2 rating across 511 reviews reflects. As organizations scale, seat pricing limits how broadly they can extend full-platform access. That is the exact pattern Afida described when it moved off New Relic to a per-node model.
Strengths: New Relic gives teams a perpetual 100 GB free tier with 1 free full platform user and unlimited basic users. It supports 100% MACC drawdown through Azure private offers, and disciplined ingest keeps consumption pricing low.
Tradeoffs: Per-user fees limit broad organizational access. Data ingest pricing scales with volume, reintroducing the spike-cost problem, and New Relic keeps the SRE Agent human-in-the-loop with no autonomous action.
Grafana Cloud
Grafana Cloud is built on open-source foundations (the LGTM stack: Loki, Grafana, Tempo, Mimir), which appeals to teams that value portability, and it requires manual assembly of multiple collection agents while hosting data on Grafana-managed infrastructure.
Grafana Cloud is a set of signal-specific backends you configure and price independently. You deploy Grafana Alloy (or a compatible OTel Collector) for collection, choose per-signal backends (Mimir for metrics, Loki for logs, Tempo for traces, Pyroscope for profiles), and integrate Grafana dashboards. Each component is priced separately at the Pro tier, so cost modeling means summing across five backends.
Quick facts:
- Free tier (always free): 14-day retention across metrics, logs, traces, profiles.
- Pro: from $19/month plus usage; 13-month metrics retention, 30 days for logs/traces/profiles.
- Enterprise: starts at $25,000/year commit.
- Metrics (Mimir): free to 10k active series; then $6.50 per 1k series (10k–100k band), tiering down at volume.
- Logs (Loki): free to 50 GB/month; Pro splits into Process, Write, and Retain charges.
- G2 rating 4.5/5 across 131 reviews.
Grafana Cloud vs Dynatrace
The contrast is assembly-versus-automatic and open-source-versus-proprietary:
What users say
Grafana's open-source lineage and dashboard flexibility drive its 4.5/5 G2 rating across 131 reviews, and its Assistant Investigations feature coordinates a swarm of specialized agents across metrics, logs, traces, and profiles. The recurring friction is assembly: teams must configure and maintain Alloy, select backends, and reconcile per-component pricing, which is operational work groundcover and Dynatrace both absorb automatically.
Strengths: Grafana Cloud's open-source foundations reduce lock-in on collection and query. It supports full PromQL and OTel compatibility, with a generous always-free tier for evaluation.
Tradeoffs: Grafana Cloud requires manual multi-agent assembly across five backends. Data is hosted on Grafana-managed infrastructure, and per-component pricing complicates forecasting.
Chronosphere
Chronosphere targets cloud-native organizations whose containerized workloads generate 10 to 100x more data than legacy VMs, and its central mechanism is a Control Plane that shapes, aggregates, and downsamples telemetry before retention to control cost.
Chronosphere is Kubernetes-native and relies on traditional agent/SDK instrumentation for collection. Its differentiator is cost governance: the Control Plane assigns utility scores to metrics, aggregates and drops non-valuable data, and applies head and tail trace sampling before retention. Chronosphere prices retained useful data rather than host count or raw ingest.
Quick facts:
- Chronosphere quotes pricing through sales; no public price list.
- Chronosphere charges for retained useful data.
- Metrics backend is M3DB, designed for tens of billions of active series.
- Control Plane claims 89% average data volume reduction and 159% average ROI (vendor-reported).
- G2 rating 4.5/5 across 20 reviews.
- Named a Leader in the 2025 Gartner Magic Quadrant for Observability Platforms, second consecutive year.
Chronosphere vs Dynatrace
Both compete on cost at scale, but through opposite mechanisms:
What users say
Chronosphere's Gartner Peer Insights rating (4.6/5) and repeat Leader placement reflect strong sentiment among high-volume Kubernetes shops, and its Control Plane is the reason: teams facing runaway cardinality use it to cut data volume before retention costs accrue. The tradeoffs are opacity and instrumentation, since pricing requires a sales conversation and collection still depends on agents and SDKs rather than kernel-level capture.
Strengths: Chronosphere's Control Plane directly attacks telemetry cost growth before retention. Its purpose-built cardinality management fits high-volume Kubernetes environments, and Gartner named it a Leader in the 2025 Magic Quadrant for Observability Platforms for the second consecutive year.
Tradeoffs: Chronosphere publishes no public pricing, so every evaluation starts with sales. Collection still depends on traditional agent/SDK instrumentation rather than zero-code capture, and data is hosted on vendor infrastructure.
How to choose the right Dynatrace alternative
Choose on cost model and deployment architecture. Cost model determines whether incidents raise your bill. Deployment architecture determines where telemetry lives and how much instrumentation work your team owns. If unpredictable spend drove your evaluation, prioritize flat or filtered pricing. If compliance drove it, prioritize BYOC. If agent overhead drove it, prioritize eBPF collection.
Budget considerations
Model your bill at your projected node count, not your current one. Flat per-node pricing (groundcover) scales linearly and stays independent of data volume, so a 10x log spike during an incident does not change the number. Ingestion and consumption models (Dynatrace, Datadog, New Relic, Grafana) scale with data, users, or both, which means forecasting from 50 to 500+ nodes requires modeling volume growth, cardinality, and seat count together.
Three cost variables deserve explicit forecasting:
- Per-user fees: New Relic charges per full-platform user; Datadog uses per-product access controls; groundcover charges nothing per seat.
- Data volume growth: Every ingestion-priced platform except Chronosphere's Control Plane-filtered model scales cost with log and metric volume.
- Marketplace commit drawdown: If you hold AWS, GCP, or Azure commit, confirm each vendor's drawdown eligibility. Datadog confirms all three clouds; New Relic confirms 100% MACC on Azure; groundcover confirms Azure MACC. groundcover is listed on AWS, Google Cloud, and Azure Marketplaces; verify AWS EDP and GCP CUD eligibility with your cloud account team.
Migration considerations
Plan the path off Dynatrace before you sign anything, because historical data cannot be migrated, dashboards require reconstruction, and Davis AI alert logic has to be rebuilt. Run the new platform in parallel rather than expecting to backfill. Dashboards need manual reconstruction due to the gap between Dynatrace's DQL and open query languages, though 30 to 50% of existing dashboards and alerts are typically retired rather than migrated.
Four factors shape migration difficulty:
- OTel coexistence: Platforms that accept OpenTelemetry as first-class input (groundcover, all majors) let already-instrumented services coexist without rework. groundcover runs OTel alongside Flora eBPF signals in the same interface.
- PromQL and Grafana compatibility: groundcover and Grafana Cloud both offer full PromQL compatibility, preserving existing queries and dashboards.
- POC scope: A groundcover proof of concept completes in a single day; agent-based re-instrumentation off OneAgent typically runs 8 to 16 weeks.
- Rollback blast radius: With groundcover, removal is deleting the Flora DaemonSet and the backend namespace. Deeper agent integrations are harder to unwind.
Final recommendations
Match the platform to the pain that drove your evaluation. The best choice depends on the team profile.
- Kubernetes-native teams needing predictable cost: groundcover. Flat per-node pricing removes the ingestion-cost incentive entirely, and groundcover's BigBasket customer story shows the mechanism through a reported 50% reduction while expanding coverage.
- Breadth-first SaaS shops: Datadog. If integration coverage matters more than cost predictability and you accept modular billing, its catalog and marketplace drawdown are the strongest here.
- Open-source-leaning teams: Grafana Cloud. If portability and LGTM foundations outweigh the operational cost of assembling and pricing five backends, it fits.
- Regulated or data-residency-sensitive orgs: groundcover. BYOC keeps all telemetry inside your VPC at every tier, and AI Agent Mode runs on Amazon Bedrock inside your AWS account, so no telemetry crosses a trust boundary for analysis.
.svg)
