RU Contact us
ESC / 09 of 09 · Hard problems

The problems others have given up on

We take apart problems that sit between network, systems and applications. We work from evidence: packet captures, logs, reproduction.

When people call us
  • 01The problem “cannot be reproduced”
  • 02Every team has proven it is not their fault
  • 03A reboot helps for a few days
  • 04The vendor closed the case as “works as designed”
+Hard problems: ink illustration
T-0 / the failure repeats every 47 minutes
ROOT / the cause is in a neighbouring system
ESC / 09Findings are illustrative
+
Day one

What we check first

  • Build a timeline: what changed and when
  • Look for a pattern: time, load, specific nodes
  • Capture traffic and logs at several points at once
  • State a hypothesis and a way to disprove it
Tool

Describe the incident in a minute

The text is assembled in your browser. Copy it and send it to us.

+
Failure patterns

Where it usually breaks

A

Treating where it hurts

The symptom shows in the application, the cause is a load balancer timeout or DNS. The wrong layer gets fixed.

B

The unrecorded change

“Nothing was changed” almost always means the change was not recorded. Most often it is the cause.

C

Coincidence instead of cause

The failure coincided with an update, which was rolled back. The problem stayed, because the cause was a scheduled job.

+
Outcome

What you keep

  • A cause confirmed by reproduction
  • A fix and proof that the failure does not return
  • A report: what happened, why, and how to prevent a repeat
← VND

Cisco and other vendors

NET →

Networks

+
Contact

Describe your task

Tell us what is happening and what has been tried. The engineer who will handle it replies.