How DHCP timers work
A lease grants an IP for a time T. Clients typically renew around T/2 (renewal, unicast to server) and rebind near 7T/8 (broadcast if the server is unreachable). Short leases recycle addresses quickly for transient populations; long leases reduce DHCP chatter for stable devices.
DHCP message flow (DHCPv4)
Initial acquisition is DISCOVER (client broadcast/unicast helper) → OFFER (server proposes address and options) → REQUEST (client asks for that lease) → ACK (server commits). Renewals skip DISCOVER when the client still holds a valid lease and the server responds to unicast renew. Understanding this flow explains why rogue DHCP servers answering first can hijack pools and why helper-address / relay configuration is critical on routed VLANs (DHCP relay, option 82).
Capacity planning
Model worst-case RPS as clients / (T/2) for renewals plus burst DISCOVER during outages. If lease time is 30 minutes and 6,000 clients renew, expect roughly 200 renew attempts per minute baseline before Wi-Fi reconnect storms.
| Scenario | Typical lease | Rationale |
|---|---|---|
| Guest / hotspot | 30–120 minutes | Rapid reuse of limited pools |
| Corporate office | 24h–8 days | Stable desktops; aligns with many RIR-style designs |
| IoT VLAN | Short + reservations | Pair short leases with reservations for printers/cameras |
Enterprise context
Microsoft DHCP and ISC Kea often scope leases per VLAN. IPAM ties DNS names to leases; MAC randomization on phones can inflate unique client counts—size pools using observed peak plus headroom, not naive device counts.
False positives
Periodic drops can look like RF issues but are sometimes failed renewals against an overloaded DHCP service or a secondary rogue server answering first.
See changing lease time, DHCP scopes, and IP troubleshooting for related fixes.