Evidence
The wire, the kernel, or the KRO intent — whichever plane the fault actually lived on.
How an engagement runs
Impact first. Then the wire. Then the fix.
Lesley Murfin sits at the executive table and with the engineers who own the blast radius. We understand impact to the business, to your services, and to your customers before — and while — the technical work. Commercial terms stay off this site.
01
Impact first
Business — revenue, burn, risk, diligence. Services — launcher, serve, Cloud, SLOs, MTTR. Customers — conversion, trust, and the ghost pane that aborts the session. We name that impact in the room before the capture.
02
Executive altitude
Leadership gets risk, blast radius, options, and a decision in plain language. Same command posture as a national major-incident bridge. Not a status meeting that restates the dashboard.
03
Integrate with your teams
We sit with platform, SRE, network, and product. Captures, dashboards, and CI gates are shared. Your people stay the owners. We do not drop a black box in the corner.
04
Drive to resolution
Symptom → evidence → root cause → config, code, or architecture → prove-it retest. We stay through verification. There is no recommendations PDF and a wave goodbye.
True SRE
Service Reliability Engineering.
The ladder is the order of the work after impact is named. Packet Analysis is how the path gets proved. The migration console is how a Kubernetes fleet then gets moved, with KRO as the source of truth.
- Service
What the customer feels — the session, the API, the agent.
- Network
The path, the socket, the fabric.
- Compute
The host, the cgroup, the process tree.
- Then
Config, code, or architecture — and a prove-it retest.
What you leave with
A cause, a change, a retest.
The change
Config, code, or architecture. For a fleet, a RileyApplication and a runway, not another hand-written sync.
The proof
Expected-Red before the patch, Expected-Green after, and 0→1→0 where processes and sockets must return.
Your team
Still in the room. Complex architectural design — networks, data centers, cloud, applications — is done with them, not around them.