Response times and severity
Severity is not set by how big the customer is, but by how much the business is affected. The definitions sit in a contract annex, so no incident starts with an argument about whether it counts as a P1.
| Severity | Response time | Applies to |
|---|---|---|
| P1 | 30 minutes | Core service down or the whole network unusable — emergency response starts at once and management is notified |
| P2 | 8 hours | A critical link or device degraded, with partial business impact |
| P3 | 24 hours | A single failure or performance issue with no impact on the main service |
| P4 | 48 hours | Advisory, improvement requests, routine changes |
The table shows the standard tier. Actual times follow the signed contract, shorter ones can be agreed for key accounts, and different systems at one customer may carry different tiers.
Four things written into the contract
Shift cover
Someone is always on, 7×24, in Chinese, English and Cantonese; cross-timezone projects are rostered to the customer's clock.
Regular reporting
Incident reports follow each event, quarterly inspections give conclusions, and the annual review gives trends and budget advice.
Spares and managed support
The spares pool is bound to the equipment register: core units in stock, swap before repair.
Traceable change
Every change goes through the customer's CAB, configuration snapshots are kept, and afterwards you can trace which change caused the problem.
Where spares and support end
Spares are not whatever happens to be on the shelf. We line the equipment register up against stock and decide what to hold by failure probability × business impact:
Core layer
Core switches, firewalls and controllers are held 1:1 or 1:N — replaced on failure, without waiting for a vendor RMA.
Access layer
3–5% by model mix, with district rotation stock on multi-site rollouts.
Consumables and cabling
Optics, patch cords and power supplies are stocked once for the project cycle, so a second purchase never delays the schedule.
How advisory and architecture are delivered
Usually two to three weeks, with three outputs: a current-state and risk register, a three-year roadmap, and a budget and option comparison. The assessment starts from your requirements, and the report states plainly what existing equipment still has life and what need not be replaced.
Typical triggers: two networks to merge after an acquisition, a data room to relocate, a head office mandating one standard — or simply fixing something every year without being able to say where the problem is.