Prepare the deflection before the peak
Publish the five articles that answer the questions the peak will generate, and put them in the chat widget and the ticket form. During the peak itself there is no time to write anything.
Announce before you are asked
A planned maintenance notice posted in advance prevents dozens of identical tickets. During an incident, a status update every hour — even with no news — reduces inbound volume dramatically.
Triage ruthlessly
Create a temporary category and route everything about the peak event into one queue with one canned reply. Do not let it contaminate normal work, and do not let normal work starve.
Use temporary staff correctly
Seasonal accounts with restricted permissions, working only the categories your knowledge base covers well. Escalation to permanent staff for everything else. Deactivate the accounts afterwards.
Protect the team
Watch the workload view daily during a peak. The person carrying double will not tell you, and the cost of that arrives four weeks later as a resignation.