Sunday
Sep232007
HA for switches
Sunday, September 23, 2007 at 6:31PM
Hi,
Can someone teach me how you implement network switch fail over since we are paranoid for single point of failure.
For example, you have:
a dozen web servers -> switch -> DB cluster
that switch is a SPOF. How does one implement dual switch in a fail over fashion?
Can someone teach me how you implement network switch fail over since we are paranoid for single point of failure.
For example, you have:
a dozen web servers -> switch -> DB cluster
that switch is a SPOF. How does one implement dual switch in a fail over fashion?
Reader Comments (2)
Does your service provider have a switch you can use? That might be cheaper depending on your number of racks and your bandwidth. And usually the equipment vendor has doc on how to setup redundancy, but I think these articles do a good job explaining some options: http://www.commsdesign.com/showArticle.jhtml?articleID=25600515">Redundancy: Choosing the Right Option for Net Designs, http://en.wikipedia.org/wiki/Virtual_Router_Redundancy_Protocol">Virtual Router Redundancy Protocol, and http://articles.techrepublic.com.com/5100-10879_11-6149096.html">Build a fast, reliable, and affordable switch infrastructure for your datacenter.
It's really easy to have multiple nics in the same vlan on multiple switches run a HA config.
It's even as easy as using the linux "bonding" module (or a windows paid for driver). The host arpings a target (the DB cluster, or a gateway, or ...) on all it's interfaces.
If an interface does not receive an arp reply, it is considered dead and removed from the virtual interface. This allows for network failover in the order of tens of ms.