[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: DATACENTER: DataCenter Failover

Thanks for the feedback. It sounds like using BGP as a failover solution
isn't too common or popular. I'm wondering if anyone can give me some
references on users of Cisco's Distributed Director and/or F5's 3DNS
product. Any opinions?

Jason Emery
Network Engineer

Avenue A Media
voice:	206-971-8211
fax:	206-521-8808
 <mailto:[email protected]> mailto:[email protected]
 <http://www.avenuea.com> http://www.avenuea.com



		-----Original Message-----
		From:	[email protected] [mailto:[email protected]]
		Sent:	Wednesday, April 14, 1999 4:18 PM
		To:	Jason Emery
		Cc:	[email protected]
		Subject:	Re: DATACENTER: DataCenter Failover

		> Does anyone have any recommendations or details regarding
internet
		> datacenter network redundancy/failover. We have been
tossing around the idea
		> of using a single network (say a /20) in two geographic
locations. Is it
		> possible to advertise this space from two locations to the
same and/or
		> different upstream providers - like Sprint and UUNet? If a
datacenter is
		> down/demolished etc... is this a viable way to implement
failover and/or
		> load sharing? Other solutions we are looking at are F5
BigDNS and Cisco
		> Distributed Director. Has anyone used BGP this way?
Thanks,

		You can do it with with BGP, by making the AS_PATH coming
from the
		failover location substantially longer by prepending your
ASN a few times
		in the case of 2 different upstreams.  In the case of a
single upstream
		you can do it with BGP's MED (Multi Exit Discriminator)
metric.

		Of course, gets you no load balancing between the two like
the F5's
		DNS-based solution does.  You will always be using the
primary datacenter
		until the failover happens.

		-nikw