Independently operated. Not affiliated with Datadog, New Relic, Grafana Labs, Dynatrace, Splunk, or Elastic. Pricing sourced from public pages and may not reflect current rates. Verify on each vendor's pricing page before purchasing.
MonitoringCost.comRun Calculator

Cost by host count

Cloud monitoring cost for 100 hosts

Verified April 2026

At 100 hosts, the structural cost differences between vendors become large enough to drive procurement decisions. Grafana Cloud and New Relic land in the low thousands. Datadog, Splunk, and Dynatrace land in the mid-thousands to low five figures. The same workload can cost 4x to 7x more on the wrong pricing model fit.

TL;DR

Grafana Cloud at $1,000 to $1,800/mo. New Relic at $1,400 to $2,800/mo. Datadog full obs at $5,500 to $9,000/mo. Splunk and Dynatrace at the higher end. The 100-host scale is where serious cost optimisation begins and where the migration economics from Datadog to a cheaper vendor become attractive.

Six vendors at 100 hosts

The realistic monthly bill

Each vendor priced for a 100-host deployment with full APM, 50 GB/day of logs, 200 custom metrics per host, and 30-day retention. Wider ranges reflect the genuine variability in workload shape. Verify on each vendor's pricing page.
VendorMonthly costNote
Grafana Cloud$1,000 to $1,800Disciplined K8s labelling lands here. Cardinality drift can push to $5,000+.
New Relic$1,400 to $2,800Single-meter ingest scales linearly. Predictable billing.
Elastic Cloud$1,500 to $3,500Resource-based deployment; medium-large tier handles 100 hosts.
Datadog (full obs)$5,500 to $9,000Per-host plus APM plus log indexing dominates. Negotiation begins at this scale.
Splunk Cloud$5,000 to $12,000Workload pricing medium pack; ingest dominant if log volume is heavy.
Dynatrace$5,800 to $9,500Full-Stack at $58/host list; some negotiation possible at 100 hosts.

What changes at this scale

The 100-host context

The 100-host deployment is the canonical mid-market observability scale. Typical organisations include the Series C startup with a real platform team, the established SaaS company with 50 to 200 engineers, the e-commerce platform serving meaningful daily traffic, and the IT operations team running production VMs across a few business units. The common feature is that observability is a substantive operational function, the team has either a dedicated platform engineer or a senior infrastructure engineer with platform responsibilities, and the bill is large enough to justify procurement attention.

At this scale, three new dynamics appear that did not matter at 50 hosts. First, Kubernetes adoption is typically standard rather than emerging, which means custom metric cardinality is a real cost driver and the per-pod billing models on some vendors compound aggressively. Second, log volume crosses 50 GB per day for most workloads, which is the threshold where Datadog log indexing becomes the largest single line item. Third, multi-region deployments become common, which adds cross-region telemetry fees on some vendors and may require multi-region observability backends.

The most consequential decision at 100 hosts is whether to negotiate enterprise pricing or accept list rates. Datadog discounts modestly at this scale (typically 10 to 20 percent off list on annual commitments). Dynatrace discounts more aggressively (20 to 35 percent off list with annual DPS pool commitments). Splunk discounts substantially (30 to 50 percent off list with workload pricing migration). New Relic and Grafana Cloud discount on annual commitments at standardised tiers (typically 15 to 25 percent below list at 100-host scale). For most teams, asking for the annual-commitment discount is the lowest-effort 15 to 25 percent saving available.

The second consequential decision is whether the existing vendor is still the right fit. A team that started on Datadog at 10 hosts and grew to 100 hosts now faces a meaningful structural cost premium versus Grafana Cloud or New Relic. The migration economics typically pay back in 7 to 14 months, with the major risk factor being the loss of Datadog-specific add-ons that competitors do not match. Teams without serious dependencies on DBM, NPM, or CI Visibility face an attractive migration choice at this scale.

Where the bill compounds

Three cost drivers at 100 hosts

Datadog log indexing

At 50 GB/day of logs at Datadog default indexing, the indexing line item runs $4,000 to $20,000 per month depending on event density. Configure index exclusion filters from day one. Drop DEBUG, INFO, and structured access logs at the agent before they hit the indexing meter.

Custom metric cardinality

Kubernetes labels routinely produce 200 to 500 custom metrics per host. At Datadog overage rates, the additional cost is $100 to $400 per month at 100-host scale; on Grafana Cloud the active series count can blow up the bill 5x to 10x without warning. Drop high-cardinality labels at the agent.

Cross-region telemetry

Multi-region deployments often incur cross-region data transfer fees on the cloud provider side and per-region telemetry charges on some observability vendors. A 100-host deployment split across three AWS regions can add $500 to $2,000 per month in cross-region observability cost.

The migration question

When 100 hosts is the right time to switch

The 100-host scale is where the cost-savings analysis for migrating from Datadog to a cheaper vendor becomes economically attractive for the first time. Below 100 hosts, the absolute dollar savings are too small to amortise the migration cost; above 200 hosts, the savings are large enough that the case is overwhelming. The 100-host range is the boundary where the case becomes worth seriously evaluating but not yet automatic.

The standard cost-savings analysis at 100 hosts looks like this. Current Datadog spend at 100 hosts with full observability is $5,500 to $9,000 per month, or $66,000 to $108,000 per year. Migrating to Grafana Cloud lands at $1,000 to $1,800 per month, or $12,000 to $22,000 per year. Annual savings are $44,000 to $96,000. One-time migration cost is typically $40,000 to $80,000 in engineering time (dashboard rebuilds, alert rule rewrites, parallel-run period of 60 to 90 days, team training on PromQL and Grafana). Payback period is 7 to 14 months. Three-year cumulative savings net of migration are $90,000 to $230,000.

The non-financial considerations are equally important. Datadog migration involves losing the integration breadth (650 plus integrations versus Grafana Cloud's broader but less curated open-source exporter ecosystem), losing specific add-ons (DBM, NPM, CI Visibility), and accepting the operational learning curve for a new platform. For teams with serious dependencies on Datadog-specific capabilities, migration may be operationally infeasible regardless of the cost-savings case. For teams using Datadog as general-purpose observability without specific add-on dependencies, migration is increasingly attractive at 100 hosts.

Migration is not the only lever. For teams that prefer to stay on Datadog, negotiation at the next renewal is the rational alternative. A 100-host team that has not yet asked for an annual-commitment discount is leaving 10 to 20 percent on the table. A team that has already negotiated once can push for further concessions at the three-year mark, particularly if a migration to Grafana Cloud or New Relic is credibly on the table as an alternative.

Cost reduction levers

Three things to do at 100 hosts

Negotiate the annual-commitment discount

The single lowest-effort 15 to 25 percent saving. Every major vendor discounts annual commitments versus monthly billing at this scale. Ask for the annual rate at next renewal; most teams do not realise this discount exists.

Configure log index exclusion

On Datadog, the single largest budget mover. Index exclusion filters drop low-value logs before they hit the indexing meter, saving 50 to 80 percent of indexed-events cost without touching ingestion or query performance.

Audit custom metric cardinality

Run a quarterly audit of custom metric series count by service. Drop the top 5 to 10 high-cardinality labels at the agent. Recovers 30 to 70 percent of custom metric overage cost on Datadog and similar share of active-series cost on Grafana Cloud.

Run the migration math

For a workload-specific cost comparison and migration economics, run the inputs through the multi-vendor cost calculator. Compare against your current vendor cost and amortise migration engineering cost against the projected savings.

Frequently asked

How much does cloud monitoring cost at 100 hosts?
Between $1,000 and $15,000 per month depending on vendor and observability scope. Grafana Cloud is typically $1,000 to $1,800 (cheapest for disciplined Kubernetes workloads). New Relic is $1,400 to $2,800 (cheapest for non-Kubernetes workloads with predictable telemetry volume). Datadog with full observability lands at $5,500 to $9,000. Splunk and Dynatrace are positioned at the higher end at $5,000 to $12,000. The 100-host scale is where the structural cost differences between vendors become large enough to drive serious procurement decisions.
Is 100 hosts the right scale to negotiate enterprise pricing?
Yes, this is where negotiation begins meaningfully. Datadog discounts modestly at this scale (typically 10 to 20 percent off list on annual commitments). Dynatrace discounts more aggressively (20 to 35 percent off list with annual DPS pool commitments). Splunk discounts substantially (30 to 50 percent off list with workload pricing migration plus annual commitment). The negotiation is real but not dramatic at 100 hosts; the bigger discount unlocks happen at 500 plus hosts.
Why do 100-host bills vary so much between vendors?
The pricing model fit to the workload is the dominant variable. A 100-host Kubernetes deployment with 200 custom metrics per host (typical for K8s with five labels per metric) costs $1,200 on Grafana Cloud (active series scale linearly with cardinality), $2,000 on New Relic (single-meter ingest absorbs the cardinality), and $7,000 on Datadog (per-host plus custom metric overage compounds). The same workload on 100 hosts of stateless web servers with low log volume is $700 on Grafana Cloud, $400 on New Relic, and $2,500 on Datadog. The 4x to 10x spread between vendors reflects pricing model collisions, not real capability differences.
What is the most common cost surprise at 100 hosts?
Datadog log indexing. A team with 50 GB/day of logs at default Datadog indexing pays $20,000 per month for indexing alone, often double the per-host plus APM line items combined. The fix is to configure index exclusion filters that drop low-value logs (DEBUG, structured access logs, repetitive health-check logs) before indexing. Most teams discover this only after the first quarterly invoice review. Configure index exclusion filters from day one, not retroactively.
Should I migrate from Datadog at 100 hosts?
The economics of migration become attractive at this scale. A 100-host team paying $7,000 per month for Datadog could move to Grafana Cloud at $1,200 per month (saving $70,000 per year) against a one-time migration cost of $40,000 to $80,000 in engineering time. Payback is 7 to 14 months. The trade-off is short-term operational risk during the parallel-run period and the loss of Datadog-specific add-ons (Database Monitoring, Network Performance Monitoring, CI Visibility) that Grafana Cloud does not match feature-for-feature. For teams without those specific Datadog dependencies, migration is increasingly the rational choice at 100 hosts.
Is custom metric cost a real problem at 100 hosts?
Yes, on Datadog. A 100-host Kubernetes deployment routinely produces 200 to 500 custom metrics per host (above the included 100 per host). At Datadog's overage rate of $0.05 per 100 metrics, the overage alone runs $100 to $400 per month. The bigger issue is that custom metric overage is invisible until the invoice arrives; there is no live dashboard showing imminent overage. Configure custom metric cardinality limits or drop high-cardinality labels at the agent before they reach Datadog. On Grafana Cloud the same workload would be billed via the active-series meter, which is more transparent but also more punishing if discipline lapses.