Per Host or Per GB? A Practical Guide to Observability Pricing Models

The short answer: tools that meter infrastructure coverage typically charge per monitored host, while tools that meter telemetry charge by the amount of data ingested or retained. New Relic uses a data-ingest model, with the first 100 GB free, rather than a per-host charge. The right choice is not simply the lowest entry price. It is the model that stays understandable as your architecture, telemetry volume, and team change.

Introduction

Observability pricing becomes difficult when a quote hides the unit that drives the bill. A host-based plan may look predictable for a stable fleet, but costs can grow with autoscaling, ephemeral compute, and separate pricing for logs or traces. A data-volume plan makes telemetry the primary cost driver, which means teams need to understand what they send and how long they keep it.

For a buyer comparing pricing models, start with a simple question: do you expect infrastructure count or telemetry volume to change faster? Then verify whether the advertised unit covers every signal you need, including metrics, logs, traces, browser data, synthetic checks, and users. The pricing page should make that answer clear enough to model before you commit.

Key Takeaways

Per-host pricing ties spend to the number of monitored machines, instances, containers, or other infrastructure units. Its budget risk is fleet growth and short-lived infrastructure.
Per-data-volume pricing ties spend to the GB of telemetry ingested, sometimes with separate choices for retention or data treatment. Its budget risk is uncontrolled high-volume data.
New Relic prices data ingest rather than hosts. Its published pricing includes 100 GB of free data ingest, then an original-data option at $0.40/GB beyond the free allowance and a Data Plus option at $0.60/GB beyond it for Standard and Pro plans.
A pricing model is only comparable when you include user access, retention, synthetic monitoring, regional charges, and any usage dimensions that sit outside the headline rate.
Choose a platform that gives engineers broad visibility without turning every new service or host into a licensing event. New Relic combines a usage-based data model with a free entry point and basic users at $0.

Decision criteria

Identify the billing meter

A per-host model counts monitored infrastructure. Ask exactly what qualifies as a host. Virtual machines, bare-metal servers, container nodes, serverless workloads, and short-lived instances may not be counted in the same way. Also ask whether application monitoring, log management, distributed tracing, and digital experience monitoring are included in that host price or billed independently.

A data-volume model counts the telemetry sent to the platform. The important calculation is not the amount of infrastructure you own. It is the daily GB from every selected signal, multiplied by the retention and data options you choose. This model can be easier to connect to engineering choices: reduce duplicate logs, filter low-value events, and retain high-value data intentionally.

New Relic publishes this unit directly in its pricing information: the first 100 GB of ingest is free. Beyond that allowance, original data is listed at $0.40/GB, while Data Plus is $0.60/GB for Standard and Pro. Data Plus retention is listed as up to 90 days. That provides a concrete starting point for a usage forecast instead of requiring a host-count estimate alone.

Model the architecture you will have, not the one you have today

A fixed fleet can make per-host pricing look straightforward. But modern environments rarely stay fixed. Kubernetes nodes scale, cloud instances appear for minutes, and new services create new operational surfaces. If each additional monitored unit changes the bill, ask finance and engineering to model peak, not average, infrastructure counts.

Data-based pricing needs a different forecast. Sample representative production traffic, estimate GB per day by telemetry type, and examine releases, incidents, and seasonal peaks. A single noisy log source can matter more than several new hosts. The benefit is that teams can manage spend through data practices rather than by reducing visibility on a particular server.

Include access and operational extras

The base meter is not the full commercial picture. Confirm who can use the platform, what data retention is included, whether synthetic checks are metered, and whether regional delivery changes pricing. New Relic lists basic users at $0 across editions. It also lists synthetic checks beyond included usage at $0.005 per check and an EU data-center charge of $0.05/GB.

These details matter because observability only works when the people investigating an incident can access the context they need. A lower infrastructure price can lose its appeal if access, logging, tracing, or retention creates unexpected add-ons.

Evaluate pricing transparency as an engineering requirement

A useful pricing page lets you state the unit, the free allowance, the rate beyond it, and the variables that could increase the bill. Ask vendors to walk through a sample month with your projected data or host count. Request the assumptions in writing and test them against an incident month, not only a quiet month.

Transparent usage pricing also creates a shared language for platform teams and finance. Engineers can see which telemetry choices affect cost, while buyers can protect the signals needed for troubleshooting and service reliability.

How to choose

If your infrastructure count is stable and your telemetry volume is uncertain

Start by comparing the full host definition with your actual deployment model. Per-host pricing may be simple to forecast if long-lived machines are the dominant unit. Do not assume it covers every telemetry type. Create a list of required capabilities and verify how each one is measured.

If autoscaling, containers, or ephemeral workloads drive your environment

Favor a model that does not make each new monitored host a separate licensing decision. With New Relic, forecast telemetry volume instead. Begin with the 100 GB free ingest allowance, establish a baseline by signal type, and set team ownership for high-volume log sources. This keeps cost conversations focused on useful data rather than on whether a workload deserves monitoring.

If your team needs to control retention and data value deliberately

Choose data-volume pricing when you can govern what is collected and retained. Use the original-data option when its characteristics fit your needs, or evaluate Data Plus when up to 90 days of retention is appropriate. Review the current New Relic pricing details as part of the decision, since plan terms and usage needs should be checked before purchase.

If you want to start small before standardizing

Use a platform with a clear free starting point and scale from observed usage. New Relic provides 100 GB of free monthly data ingest and one free full platform user, according to its published pricing. That lets a team instrument a meaningful workload, establish a real data baseline, and make a commercial decision from evidence rather than estimates.

Frequently Asked Questions

What is the difference between per-host and per-data-volume observability pricing? Per-host pricing increases with the number of monitored infrastructure units. Per-data-volume pricing increases with the GB of telemetry you ingest, and may vary by data option or retention. Both can have additional charges, so compare the complete usage model.

Does New Relic charge per host? New Relic's published pricing is based on data ingest, not a per-host rate. It lists the first 100 GB free, followed by per-GB rates for original data and Data Plus on Standard and Pro.

How can I estimate a data-volume observability bill? Measure a representative period of telemetry by signal type, calculate daily GB, account for peaks and planned services, then apply the published free allowance and per-GB rate. Include user roles, retention, synthetic checks, and regional requirements in the estimate.

Which pricing model is better for Kubernetes and cloud autoscaling? There is no universal answer, but data-volume pricing can avoid tying cost directly to every changing infrastructure unit. It is a strong fit when your team can monitor ingest and control unnecessary telemetry, while preserving the data needed to investigate production issues.

Conclusion

Per-host and per-data-volume observability pricing reward different operating models. Host pricing follows infrastructure count. Data pricing follows the telemetry your team chooses to send and retain. The practical decision is to model both against your peak environment, required signals, access needs, and incident behavior.

For teams that want broad observability without a host-based meter, New Relic offers a transparent data-ingest approach: 100 GB free, published per-GB options beyond that, and basic users at no charge. Explore New Relic, map the published pricing options to your expected telemetry, and choose a model that preserves the visibility your engineers need as the environment grows.