Start with the failed application transaction
Use this workflow when users report failed or intermittent access but infrastructure alerts do not explain the service outcome. Record the source location, destination, protocol, time window and expected transaction. Distinguish a failed login, slow response and unreachable service before collecting evidence.
NetPrompt’s network-device and server connectivity capabilities let you organize checks around the affected flow. Define the devices and server platforms in scope, validate their access, and use your troubleshooting context to select the relevant evidence.
Follow the flow across operational boundaries
Collect evidence for the actual source and destination rather than stopping at a successful management ping. Align timestamps across tools and keep expected design separate from observed behavior.
| Stage | Evidence to gather | Decision it supports |
|---|---|---|
| Name and destination | Resolution result and expected destination | Is the request reaching the intended service address? |
| Network path | Relevant routing context, interfaces and path observations | Where does the evidence suggest a connectivity break? |
| Security boundary | Applicable policy and relevant permitted or denied flow evidence | Does policy explain the reported behavior? |
| Server and service | Host state and approved service checks | Is the destination available to serve the request? |
| Recovery | Repeat the representative application transaction | Has the user-facing service recovered? |
Illustrative example: a router is reachable but the service is not
A branch can reach the management address of its router, while transactions to an internal application fail. The supplied Network Context explains that application traffic uses a separate routing context. The investigation therefore follows the service path and its return traffic instead of treating management reachability as proof of application health.
If current evidence points to a server-side issue, the workflow records that evidence and the next responsible team. If a policy change is proposed, the engineer reviews its scope and approval requirements. This example describes an evaluation scenario, not a customer result.
What to prepare for a NetPrompt evaluation
Provide the service map, target device identities, applicable platform versions and approved checks. Server access and command templates must be verified for each platform. Where a load balancer, DNS or application system is accessed through an API, validate the required connector operation and permissions.
Keep the first trial focused on one service. Test a known fault and a healthy transaction so that the team can review whether the workflow distinguishes meaningful evidence from unrelated alarms.
- Source and destination details with a reproducible symptom.
- Network Context for paths, zones and dependencies.
- Troubleshooting Context for interpretation and escalation.
- Working device or connector access for the chosen scope.
- A service owner and explicit recovery test.
Measure a useful investigation outcome
Review whether the workflow narrows the fault domain, identifies missing evidence and creates a coherent handoff. Measure investigation time and operator corrections against your own baseline without assuming a universal improvement percentage.
A supported conclusion may still be “insufficient evidence.” Keep that limitation visible and assign the next check. Configuration changes and service restarts require the applicable review process, followed by service verification.
Common questions
- Does this automatically diagnose every application problem?
- No. Coverage depends on the connected systems, supported checks and available evidence. Application internals may require additional tooling and the application owner.
- Can this span multiple network vendors?
- Yes, evaluate the relevant device profiles, software versions and commands for your service path. Multi-vendor support does not imply identical coverage on every device.
- What is the first demonstration to request?
- Use a known connectivity incident with verified evidence and ask NetPrompt to explain the fault domain, missing inputs and final service verification.
Continue exploring
Further technical reading
External references explain general practices; they are not endorsements of NetPrompt AI.