News Support Contact

Plumbline platform

Four stages built on one model. Collection is read-only and agentless. Everything after it, the standard build, consistency checking, change automation and reporting, works on the same normalised view of the estate.

Collection

Plumbline talks to network elements the same way your existing management systems do. There is no agent to install, no image to certify and no change to the element configuration in order to onboard it.

MethodUsed for
NETCONF / YANGNative model retrieval where the element supports it.
gNMIOn-demand or subscription-based collection on modern platforms.
SSH / CLIParsed CLI for the long tail of legacy elements.
EMS northboundREST and SOAP where direct element access is not permitted.
Backup importIngests what your current backup tooling already collects, as a zero-touch start.

Credentials are held in the operator's vault and referenced at run time. A collection pass over 1,200 elements typically completes in under twenty minutes.

Standard build

Raw configuration is parsed into a vendor neutral model organised by element role. Every element carrying a given role is measured against the expected shape for that role, so a router in the aggregation layer is compared with its peers rather than with the whole estate.

Standard builds start from what you already run. Plumbline clusters elements by role, shows where they agree, and presents the disagreements for a one time decision: is this a legitimate exception, or is it drift that has been in place long enough to look deliberate? Each standard build carries a version, an owner and an approval record, and changing one is itself a reviewed change.

Exceptions are recorded against an element or a role with an expiry date, so a temporary difference for a migration or a market specific requirement does not quietly become permanent.

Consistency checking

A raw diff is not useful on its own. Plumbline groups each deviation before it reaches an engineer, so the team sees the handful that matter rather than the several hundred that do not.

GroupMeaningDefault handling
known changeMatches an open change record, applied inside its window.Logged. Standard build update proposed at change closure.
exceptionMatches a recorded, unexpired exception.Suppressed. Listed in the exception expiry report.
driftA difference from peers with no change record, low operational impact.Queued for review. Optional automatic return to standard.
off standardA difference in a parameter class the operator has marked as important, such as software version or interface settings.Raised to the domain owner.

Which parameters count as important is configuration, not code. Software versions, interface and MTU settings, timers, routing policy and naming conventions are defined by the operator and versioned alongside the standard build.

Change automation

The model that detects drift also runs change. A workflow is a parameterised method of procedure with machine checkable assertions before and after the apply. It exists to make routine, repetitive change faster and less error prone, not to replace the engineer who designs it.

  • Pre flight assertions confirm that the starting state on every target matches what the workflow was written against, before anything is written.
  • Execution carries a change reference from the operator's change management system, so every automated change is ticketed.
  • Apply is staged. Canary first, then batches, with configurable blast radius per domain.
  • Post flight assertions check end state and service indicators, with automatic rollback on failure.
  • Every run writes a record of what changed, where and when, so the team can see the history without reading device logs.

Integrations

CategorySupported
Change managementServiceNow, Jira Service Management, BMC Remedy, generic REST
SecretsHashiCorp Vault, CyberArk, Azure Key Vault
IdentitySAML 2.0, OIDC, LDAP
MonitoringPrometheus, syslog, Kafka event stream
InventoryNetBox, vendor EMS inventory, CMDB over REST or CSV
Source controlGitLab, GitHub Enterprise, Bitbucket Data Center

Deployment and commercials

Self-hosted only, inside the operator's OSS boundary, on bare metal, VMs or Kubernetes. There is no SaaS option and no outbound connectivity requirement from the management network. That rules us out of some deals. It is also usually what makes the platform straightforward to approve internally.

Licensing is an annual subscription banded by managed element count. Domains are licensed independently so you can start narrow. Support is 24x7 for production deployments.

Evaluations run against a configuration export you provide, so you can see a consistency report on your own estate before anything is installed.

Request an evaluation