Grafana Fleet Management vs LinkMesh

On 8 July 2026, Grafana made remote management of upstream OpenTelemetry Collectors generally available in Grafana Fleet Management. If you run a collector fleet, that matters: a large, well-known vendor now offers real-time health monitoring, centralized configuration, and attribute-based targeted rollouts for both Grafana Alloy and upstream OTel Collectors — and it comes with a Grafana Cloud account rather than a separate line item.

We build LinkMesh, which does a lot of the same things. So take the obvious bias as read, and let us be useful anyway: for most teams evaluating this, Grafana Fleet Management is a perfectly good answer. This post is about the specific case where it isn’t, and how to tell quickly which case you’re in.

Everything below about Grafana is quoted from their public documentation and linked at the point of use, checked on 31 July 2026. Product docs move; if you find something here that’s gone stale, tell us and we’ll correct it in place.

What Grafana Fleet Management does

Credit where it’s due — the feature set is real and it covers the fundamentals of fleet management:

• Health and inventory. An at-a-glance view of fleet health, collector counts, versions and resource usage, plus registration and bulk management of collectors with custom attributes.
• Remote configuration. Configuration pipelines authored as standard OpenTelemetry YAML and assigned to collectors by matching attributes, which is a clean model for targeted rollout.
• Both collector families. Alloy and upstream OTel Collectors from one control plane, which was the gap the July GA closed.

Architecturally, upstream collectors connect over OpAMP — Fleet Management “exposes an OpAMP endpoint at /v1/opamp”, and “the OpAMP Supervisor is required to receive and apply remote configuration”. Alloy takes a different path, polling a GetConfig endpoint on the interval set in its remotecfg block. If OpAMP itself is new to you, we wrote up how the protocol works and the control-plane gap it deliberately leaves open.

None of that is a criticism. It’s a competent implementation of centralized collector management, and as of July 2026 “manage your collectors centrally” is no longer a differentiating feature. It’s table stakes.

The question that actually decides this

Not what the control plane does. Where it runs, and who holds your fleet metadata.

Grafana’s own documentation is admirably direct about this. From the self-managed setups page:

“Grafana Fleet Management is a Grafana Cloud feature, but you can use it with self-managed enterprise or on-premises setups where a connection to Grafana Cloud is possible.”

That last clause is the whole boundary. “Self-managed” here means your collectors and your data sinks are self-managed. The control plane itself stays in Grafana Cloud. And the architecture docs confirm what that implies for your fleet’s metadata:

“All data used to manage your fleet, including collector attributes and configuration pipelines, is durably stored in the Fleet Management database.”

Plus: “All Fleet Management API and OpAMP traffic passes through this gateway” — the Grafana Cloud gateway that handles authentication and routing.

So the honest summary is: your collectors can live anywhere you like, but the inventory of what you run, how it’s configured, and the channel used to configure it, all terminate in the vendor’s cloud.

For most teams that is entirely fine, and worth being clear about rather than scaremongering over. It’s the same trade every SaaS control plane asks you to make, and Grafana is a serious operator.

Where it runs out of road

There is one case the model can’t reach, and again their docs say so plainly:

“Fleet Management doesn’t currently have a native solution for air-gapped environments, but you can establish secure connections to the Fleet Management service even from restricted networks.”

The documented workarounds are cloud-provider private connectivity (AWS PrivateLink, Azure Private Link, GCP Private Service Connect), a proxy configured through Alloy remotecfg proxy arguments, or an internal reverse proxy forwarding Fleet Management API requests.

Read those carefully and they share one property: each is a different way of getting a connection to Grafana Cloud, not a way of doing without one. That’s not a flaw — it’s a coherent product decision. But if your constraint is “this network reaches nothing external”, or “fleet metadata does not leave our perimeter, as a matter of policy rather than preference”, then a better tunnel isn’t the answer to the question you’re being asked in the audit.

That constraint is real for a narrower set of teams than vendor marketing usually implies: regulated finance and insurance, critical infrastructure, public sector, defence, and anyone operating under a data-residency mandate that names the control plane and not just the telemetry.

If that isn’t you, stop reading and go use Grafana Fleet Management. Genuinely.

How LinkMesh is built differently

LinkMesh is a self-hosted control plane for OpenTelemetry collector fleets. The design choices follow from that one constraint:

• The whole control plane is one binary with embedded storage. No external database to provision, no worker tier to size and hand-configure, no SaaS dependency. It runs on a VM, in Docker, on Kubernetes, or fully airgapped, and makes no CDN calls at runtime.
• Your perimeter, your credentials. Every collector authenticates to the control plane with a durable Bearer token over TLS you terminate at your own ingress, and enrollment is an explicit onboarding step rather than an open registration endpoint. The trust boundary — control plane, fleet metadata, collector credentials — stays inside infrastructure you run.
• Fleet management with depth, not just config templating. Central config push with version history, a visual pipeline builder backed by configuration-as-code, live topology with per-edge throughput, and health and lifecycle management per collector. Keeping a fleet patched is its own problem, which we covered separately in how to keep an OpenTelemetry collector fleet up to date.
• Routing you can read. A guided field/operator/value route builder for attribute-based routing, with raw OTTL available as the Advanced fallback for expressions the builder doesn’t cover, plus label-based destination selection so tagging a destination auto-joins it to matching routes.

Worth naming the edges of that last one, because a comparison post that only lists strengths isn’t worth reading: the destination selector is equality-only — no In, NotIn or existence operators — custom selectors are authored through the API or YAML rather than drawn on the canvas, and routing evaluates as a priority-ordered, first-match-wins cascade rather than a declarative N-way routing table. Those are design choices with trade-offs, not gaps we’re hiding.

Side by side

	Grafana Fleet Management	LinkMesh
Control plane location	Grafana Cloud	Your infrastructure
Fleet metadata & config storage	”Durably stored in the Fleet Management database” (Grafana Cloud)	Embedded storage in your own deployment
Air-gapped operation	”No native solution” today; connectivity workarounds reach Grafana Cloud	Runs with no external connectivity
Manages upstream OTel Collectors	Yes, GA July 2026	Yes
Manages Grafana Alloy	Yes, via remotecfg	Yes
External dependencies	Grafana Cloud account	None — single binary, embedded storage
Price model	Included with a Grafana Cloud account; metering happens on telemetry ingest	Per managed collector; free to 25

What we’re not claiming

We are not claiming to be cheaper. Fleet Management comes with a Grafana Cloud account whose free tier costs nothing, and competing on headline price against free is a losing and slightly dishonest game. The difference worth understanding is where the meter sits: with a cloud observability platform, the bill grows on the ingest line as your telemetry volume grows. LinkMesh is priced per managed collector — CHF 12 per collector per month billed annually, with the bill capped above 200 collectors, and a Community tier that covers 25 collectors with the full feature set, forever, without a credit card. Push 100 GB or 100 TB through a collector and the price is identical. Which model is cheaper depends entirely on your volume-to-collector ratio, and you should do that arithmetic yourself.

We are not claiming feature parity across the board. Grafana Fleet Management is part of a mature observability platform with an ecosystem LinkMesh doesn’t have. If you’re already invested in Grafana Cloud, that gravity is real and rational.

We are not claiming their architecture is wrong. A cloud control plane buys you an operational burden you never carry. That’s a genuine benefit, and for most teams it outweighs everything in this post.

Choosing between them

Answer one question: can your collectors reach Grafana Cloud, and are you content for your fleet inventory and configuration to live there?

If yes — and for most teams the answer is yes — Grafana Fleet Management is the lower-friction choice, particularly if you already use Grafana Cloud. Take it.

If no, because the network is genuinely isolated or because a policy names where control-plane metadata may reside, then the connectivity workarounds are solving a different problem than the one you have, and you want a control plane that runs inside your perimeter. That’s the case LinkMesh was built for.

If you’re in that second group, you can run it yourself today — the Community tier covers 25 collectors with the full feature set, and installation is a single binary, a Docker image, or a Kubernetes deployment. The features page has the detail on what’s in the box.

If you’d rather not do it alone, we’re looking for a handful of design partners: we work through the install on your infrastructure with you, you get an unlimited license, and in exchange you tell us honestly what’s missing or broken. It’s not a sales call, and we’ll say so if we think another tool fits you better.

And if you’re weighing this against a broader set of telemetry pipeline tools rather than Grafana specifically, our Cribl alternatives guide covers the wider field with the same intent: tell you honestly where each one fits.