Skip to content
Logmetry
What you run
SplunkMicrosoft SentinelDatadogDynatraceSolarWindsCriblOpenTelemetryA Data Lake
The Logmetry Blueprint
The whole methodPhase 01 · FoundationPhase 02 · FootprintPhase 03 · AI triageEverything as codeAI native
CompareJournalAboutBook the review

What you run · OpenTelemetry

You made the right collection bet

OpenTelemetry is the open collection standard every vendor has to accept: the Collector runs on every host as pipelines of receivers, processors, and exporters, reads metrics, logs, and traces with nothing changed in any application, and ships them anywhere over OTLP. It carries no licence, and it has no alert engine of its own, so alerting lives as rules in the store you send metrics to. An estate already on it owns its collection, and the work is the fleet: OpAMP control from a plane in your account, a console in your repository, and then the Lake the collectors feed.

01Today

How an OpenTelemetry estate looks today

An estate already on OpenTelemetry owns its collection, but rollouts run by script, drift is found by accident, and every destination still receives everything.

How an estate already on OpenTelemetry looks todayThe collectors are on the hosts and the collection bet was right, though vendor agents still linger beside OTel on many of them. Configuration rolls out by script and SSH, host by host, so drift is found by accident. And every destination still receives everything, so the SIEM bills per GB and the APM per host on data that never earned those meters.YOUR ESTATEServers, VMs, containersCOLLECTORS, PLUS LEFTOVERSVendor agents still beside OTelNetwork gear, SNMP and flowCONFIG BY HANDScripts and SSH,host by hostBy handTHE DESTINATIONSEach billing on what arrivesEverything, to every destinationPER GBTHE SIEMBilled on everything inPER HOSTTHE APMBilled on every hostA collector is not the hard part.Running five thousand of them,and deciding what each of thesereceives, is.WHAT IS MISSINGCollection is solved. Thefleet around it is not.Rollouts are scripts. Driftis found by accident.Every destination stillreceives everything.Vendors will happily rentthe pane back to you.
An estate already on OpenTelemetry owns its collection, which is the position every other page is trying to reach. What is missing is the fleet: rollouts are scripts, drift is found by accident, and every destination still receives everything.

What OpenTelemetry does best

OpenTelemetry won. It is the one collection standard every vendor has to accept, the collector reads practically everything a host produces with nothing changed in any application, and it carries no per-host or per-agent licence. An estate that standardized on OTel owns its telemetry at the source, which is the position every other page on this site is trying to reach.

What OTel does not give you out of the box is a fleet. A collector is not the hard part. Running five thousand of them is: configuration rollout, health, version drift, and knowing what is actually running where. That is the layer most OTel estates are missing, and the one the platform vendors will happily rent back to you.

02The five verbs

What we do to an OpenTelemetry estate

Keep it and govern what flows in, shrink its footprint, cut what it costs, extend it with an agent, and replace only where replacement is honest.

The Logmetry Blueprint applied to an estate already on OpenTelemetryThe collectors stay and the leftover vendor agents retire. Each host dials out to a control plane in your account over OpAMP and is handed its fleet configuration, rolled out in rings with gates between, from a repository where fleets are files and changes are pull requests. The control layer decides what each destination receives, so the SIEM and the APM bill on signal, and everything lands at full fidelity in a Lake you own. Pins: keep on the collectors, shrink on the leftover agents, extend on the fleet plane, replace on the rented console, cut on the meters.YOUR ESTATEServers, VMs, containersONE COLLECTOR PER HOSTThe leftover agents retireNetwork gear, SNMP and flowKSYOUR GROUNDTHE CONTROL LAYERDecides what eachdestination receivesAs code, in your reposAll of itYOUR CONTROL PLANEEach host dials out.Config down, health upOPAMPEYOUR REPOSITORYFleets and rings as files,changes as pull requests,the console reads itREverything, full fidelityTHE LAKE, YOURSPER GB, PER HOSTOnly what each one needsCTHE DESTINATIONS YOU KEEPBilled on signal, not on everythingTHE SIEMA governed fractionof the volumeTHE APMCrown-jewel apps,deliberately smallWHAT CHANGEDOne collector, one fleet,rings from a repository.Drift is visible, rollbackis built in.Each platform bills onwhat it is good at.
The Logmetry Blueprint applied. The fleet comes under OpAMP control from a repository you own, the control layer decides what each destination receives, and the Lake accumulates full-fidelity history. The pins mark where each of the five verbs acts.
KKeep

Your collectors and your instrumentation. Nothing about the collection bet changes, because it was the right one.

SShrink

The per-vendor agents still lingering beside OTel on many hosts retire, and the estate converges on one collector.

CCut

The control layer decides what each destination receives, so the expensive platforms bill on signal instead of on everything the collectors happen to produce across the estate.

EExtend

The fleet comes under OpAMP control from a plane in your account, with a console in your repository, and the Lake lands enriched history for the agent phases.

RReplace

The rented fleet consoles. The pane of glass over your collectors belongs in your repository, not in a vendor subscription, and it is the one thing on this page that was only ever rented.

The letters mark where each verb acts in the drawing above

03The approach

The fleet, controlled as code

Each collector runs under a small supervisor that dials out to a control plane in your account over OpAMP, so the fleet is managed from your repository with no inbound credential to your network.

Configuration changes roll out in rings from a repository your team reviews, a few test hosts first, the estate after. The console shows every host, its health, what it is running and whether that matched what was sent. It holds no state of its own to lose, because the repository is the state.

From there the Blueprint continues naturally: the control layer routes, the Lake accumulates enriched full-fidelity history, and the agents read it.

04Questions

Asked about OpenTelemetry estates

We run OpenTelemetry already. Why involve anyone?

Because collection and fleet are different problems. If your collectors are versioned, ring-deployed, health-visible, and centrally controlled from a repository, you have the fleet too and need nobody. If rollouts are scripts and drift is discovered by accident, the fleet layer is the missing piece.

Is the fleet plane another product we rent?

No. The supervisor and control plane speak OpAMP, the OpenTelemetry standard for managing collector fleets, and the plane runs in your account. The console is built for you and reads your repository. There is no per-node or per-agent licence anywhere in it.

What about the destinations we send to?

They stay your choice, and that is the point. The control layer in front of them decides what each receives, so each platform bills on what it is actually good at, and changing a destination is a routing change rather than a re-instrumentation project.

Start with the review

You share your diagrams, we review them with you, and you leave with your version of the Logmetry Blueprint drawn on your OpenTelemetry estate. No system access, no obligation.

Book the reviewOr take the Logmetry Blueprint as a PDF →