While the 90-minute boot time is a reality for some high-end core switches (like the Cisco Nexus 9500 series in full-chassis configurations), there are several engineering strategies beyond a UPS to reduce downtime and manage recovery.
1. Can You Shorten the Boot Time?
Yes, but only by a certain margin. These switches are essentially high-performance servers, and their boot time is dominated by hardware self-checks (POST) and FPGA loading.
- Arista "Fast Boot" / SSU: Arista offers Smart System Upgrade (SSU) and Fast Boot capability. While designed for upgrades, it optimizes the reload process to minimize downtime. In some cases, reload fast-boot can skip certain non-critical hardware initializations.
- Warm Reload / ISSU: Both Cisco and Arista support In-Service Software Upgrades (ISSU). This allows the control plane to reboot while the data plane continues to forward packets. This doesn't help with a "cold" power loss, but it prevents the 90-minute wait during maintenance.
- Configuration Trimming: Disabling unused features and complex ZTP (Zero Touch Provisioning) scripts can shave minutes off the software initialization phase. If a switch is searching for a DHCP server that doesn't exist on a remote site, it will hang until the timeout is reached.
2. What Else Can Be Done (Besides UPS)?
- Dual-Supervisor Redundancy: Use switches with dual "brains" (Supervisor engines). If one fails or reboots, the other stays active. In a 2110 environment, this is standard practice for core switches.
- Redundant Power Feeds (A/B Power): Ensure the truck is plugged into two completely separate power sources (e.g., House Power and a Generator). The switch’s dual power supplies (PSUs) should be split across these feeds so that one source dying doesn't trigger a reboot.
- PTP Management: One of the longest parts of a 2110 recovery isn't just the switch booting; it’s the network re-syncing to PTP (Precision Time Protocol). Using Event Handlers to disable ports until the switch is fully synced can prevent "signal flapping" that disrupts the rest of the truck.
3. What Is the Fallback If a Reboot Happens?
In a properly designed 2110 system, the fallback is not "waiting 90 minutes" — it is ST 2022-7 (Seamless Protection Switching).
- The "Red/Blue" Network: 2110 trucks are built with two completely independent networks. Every camera and every server sends two identical copies of its data — one to the "Red" switch and one to the "Blue" switch.
- The "Hitless" Failover: If the "Red" switch reboots, the "Blue" switch is still carrying 100% of the traffic. The end device (like a multiviewer or a Kayenne switcher) simply ignores the missing packets from the Red side and uses the Blue side.
- The Ultimate Fallback (Analog/SDI Monitoring): Many high-end trucks still keep a small "emergency bypass" system. This is a tiny SDI router or even a few copper cables that bypass the IP fabric entirely to get a "Main Program" and "Host Feed" to the transmission (TX) booth so the show can stay on the air (even in "limp mode") while the big switches are still counting down their 90-minute boot cycle.
Summary:
The industry's answer to the 90-minute boot time isn't to make the switch faster — it's to build the system so that you never care if one switch dies.