Router Backup and Sparing

High reliability without the cost

The problem:

10/100BaseT and GigE copper routers are key elements for many networks. In these critical networks, 100% redundancy (i.e. for every main router there is a physical backup) is required to ensure the highest level of uptime. Achieving this high uptime necessitates the ability to transfer operations from the main router to the backup in the event of main router failure. Recently this transfer is accomplished by using the router’s existing control software and virtual switching. Typically it takes between one and two minutes to complete the reconfiguration of all connections and for the connections to be transferred.

This is an unacceptably long network interruption for customers with critical networks. Additionally, the router’s own virtual switching redundancy uses two expensive, high-speed circuits for each router spared, which can cost up to several thousands of pounds per router spared, per month.


What these customers require is:
• A process that automatically detects network problems and switches between main and backup paths even at remote dark sites
• The ability to switch between routers in seconds or less, not minutes
• An SNMP controlled switching system that creates SNMP traps, alerting them of any fault detection and automatic switching
• Elimination of the need for expensive backup, high-speed circuits per router
• A better way of performing annual major router firmware upgrades that can take up to 7 minutes per router. Typically, users upgrade the backup router first before switching to this router and upgrading the main router, ensuring no service interruption.

 

 

The solution:
The better solution to this router sparing requirement is to physically switch between the routers. At the physical level, the proper operation of the Ethernet Routers is continuously monitored via SNMP traps from the routers themselves, or at the physical layer, by detection of the link pulse for every 10/100 BaseT connection. The failure of a link pulse is then the basis for automatic switching to the backup path. With this solution, actual switching takes 10mS or less and about seven seconds for the Ethernet connections to establish themselves end to end. Thus recovery time is counted in seconds rather than minutes. A physical switch can optionally be operated via SNMP command, or manually, or automatically through link pulse

It is possible to implement controlled switching between the routers, when required, in manual mode for maintenance or software upgrade purposes.

 

RPS24

 

The product:

All of the above requirements are met by Cornet Technology’s RPS-24 router backup switch. A 24-channel A/B switch that is capable of autoswitching from A to B or B to A upon detection of a link pulse failure in a 10/100BaseT Ethernet connection.

Switching can also be performed as a result of SNMP command, for example, derived from alarm trap events from the router itself. Or switching can be initiated by the operator by manual command through Web browser, Telnet, or Craft port menu.

The switch’s 24 channel capacity is designed to mimic router port number modularity that goes up to 48 ports. For applications needing 48 ports, two 1U high, RPS-24’s can be used powered by a single redundant, PSC-4N, 1U high power supply that can be mounted anywhere on the rack including at the rear of the rack if required. A power failure of either power module is reported to the RPS- 24 as an alarm or SNMP trap.

The RPS-24 is designed with RJ-45 connectors on the front of the rack to handle A and B inputs. The connectors are in the same configuration as the ports on the routers. To accommodate front entry cables within racks with closed doors, the RPS-24 mounting ears adjust so to recess the chassis by several inches.

Benefits derived:
1. Switching time < 10mS.
2. Compact 1U design saves precious rack space. A 48 port system only takes up 2U to space.
3. Locating the RJ-45 inputs on the front panel corresponds with the outputs on most routers.
4. Modularity based on 24 matches the router modularity i.e.12, 24, 48.
5. Automatic switching ensures that circuit connections are automatically maintained even after router failure.
6. Manual switching minimizes downtime during periodic router upgrades.
7. Dual redundant power supply ensures backup power in the event of power supply failure. Use of latching relays for switching maintains connections even in the absence of power to the RPS-24. Failure of either power module is reported to the PRS-24 as an alarm or SNMP trap.
8. A considerable saving is achieved in not having to pay monthly rental for high speed lines required for virtual backup switching. This rental varies between countries, but a saving of several thousand pounds per router spared per month is possible. The potential return on investment of the RPS-24 is therefore very significant.