Database down, store offline. You call "urgent support" number: hold music, then "please open priority ticket." Ticket vs phone in critical outage: phone promises synchronicity; ticket promises trace — both fail if contract guarantees neither skill nor deadline.
Shared entry: ticket only, 24–48h response. Premium VPS: priority ticket + sometimes phone. Enterprise/bare metal: named on-call. Useful comparison reads SLA contract, not footer channel icon.
Critical outage comparison
| Dimension | Ticket | Phone |
|---|---|---|
| Traceability | Excellent (timestamps) | Weak without email summary |
| Real speed | Depends written SLA | Depends real on-call |
| Diagnostic complexity | Attachable logs | Verbal, omission risk |
| 24/7 availability | Often async | Rarely direct L3 engineer |
| Included cost | Standard | Premium plan |
| Post-mortem | Ticket history | Subjective notes |
Green number without infra on-call is IVR creating ticket — with extra frustration.
Ticket: when enough (or better)
Skilled internal ops — need targeted host action (hardware reset, network check).
Premium SLA first response <30 min 24/7 documented.
Non-urgent incidents: certificate, billing, planned migration.
Need proof for SLA credit or insurance.
Critical ticket good practices:
- Subject
[CRIT]+ revenue impact - Logs, graphs, start time UTC
- Explicit escalation request if SLA missed
Phone: when it changes outcome
Enterprise plan with named on-call (bare metal, finance).
Non-technical decision maker alone at night — phone for triage (high error cost).
Datacenter network outage — status history + hotline sometimes before public ticket.
Verify: is number included or premium rate? French spoken? Engineer escalation in how many written minutes?
Surveys: night support, SLA exclusions, status page.
Contract mistakes
- Confusing "chat 9–6" and 24/7 on-call
- First response SLA without restoration
- Shared "unlimited support" exclusions
- No status page — useless calls
The peak: channel does not compensate empty SLA
Decide and move forward without blind spots
- Read SLA + exclusions before signing.
- Test ticket + call pre-prod.
- Require status page + incident webhook.
- Document internal runbook — host support complements, does not replace.
Directory support filters · compare tool.
Frequently asked questions
Phone faster than ticket?
If skilled 24/7 with infra access. Else same queue — without trace.
What SLA for critical outage?
First response + written escalation. "24/7" without deadline = marketing.
Ticket enough?
Yes with premium SLA and internal ops. Better post-mortem trace.
Test before signing?
Off-hours ticket + call + status page + incident example.
Before next outage, ask: maximum delay before engineer can act on infra — and where is it written? If nowhere, phone number is decorative.