Follow-on to the prior blog entry titled “Understanding Network Redundancy”
The second theme in network redundancy design is the desire to make the most efficient use of the network. Some organizations are so focused on maximizing network utilization that they begin using redundant links to the extent that redundant operation is compromised.
Management sees network utilization reports that show links are running less than 50% utilization and want to increase the utilization as a way to hold costs down while the business grows. Let’s assume that you’ve designed primary and backup paths. If the total volume of data on both paths is greater than what will fit on one of the two paths and still provide acceptable performance, you don’t have a fully redundant configuration. You’ll need 95th percentile performance figures (see 95th Percentile Tech Tip) so that you know the network utilization over periods of time. If the sum of the 95th percentile on both primary and backup links is greater than 100%, then when a failure occurs, all the traffic will be using one path and that path will be over 100% subscribed and will be nearly 100% utilized (I say ‘nearly’ because there will always be some level of inefficiency on a link or path).
For true redundancy, the primary and backup paths should have enough reserve capacity to handle the full load when one of the two paths fails. A viable alternative is to allow operation at reduced capacity. For example, if the sum of the load on both paths is 120% of one path, a failure of one path will cause some traffic to be lost, causing the applications using that path to run at reduced capacity.
Adding additional paths would seem to answer the problem of insufficient backup bandwidth, but it increases network complexity, makes it difficult to monitor, and increases difficulty in troubleshooting. The result is seldom acceptable (see the prior blog entry for an example).
How do you learn about good network redundancy design? Cisco has a number of whitepapers, including the Solution Reference Network Design guides (SRNDs) that describe best practices for redundancy. Some of these documents include ways to configure fast failover so that time sensitive applications often don’t notice the failure of one path and the switch to the backup path. There is also a Cisco Press book that is a good resource, Building Resilient IP Networks.
Re-posted with Permission
NetCraftsmen would like to acknowledge Infoblox for their permission to re-post this article which originally appeared in the Applied Infrastructure blog under http://www.infoblox.com/en/communities/blogs.html