Desktop

Ask something
Discussion Locked

Local and remote failover systems

0 helpful
3 replies

by editor's response

Closed to new repliesThis discussion is closed. It stays readable, but it takes no new replies.

Tell us what you think about Mike Talon's advice concerning local and remote failover systems, as featured in the April 22 Disaster Recovery e-newsletter. Does your organization favor one over the other in an effort to cut costs, or does it recognize the long-term value of duplicate systems? Let us know if this information is useful to you.

If you haven't subscribed to our free DR e-newsletter, be sure to do so today!

https://www.techrepublic.com/techmails.jhtml?repID=r001

3 replies

by dbradley

Inter Data Center synchronization can cause massive problems when WAN links fail between data centers. An outage that degrades performance can lead to cluster segregation and high retransmission rates. Both of these issues ruin the backup data center's recovery capability. Mike's model makes sense; Primary site fail to primary site backup and then finally to real a disaster recovery site.

by havenmoth

Every morning the first thing our VP does is check to see if there was any downtime since he last checked. In our organizations we have many types of failover scenarios, as we are a 24/7, 365 business. Currently we utilize the failover from primaryto primary backup, then to the recovery site path for the majority of our critical platforms, [which is every kind of box ever made....:) ]however I find that increasingly technical support is more concerned with downtime than disaster recovery, andsince we are 24/7 the only test time we get is if there is a failover to implement some new hw/sw. Any suggestions on how to emphasize that "disaster recovery" should be placed prioritized as important as downtime?

by miketalonnyc

Two things:

1 - any plan that doesn't put scheduled maintenance into effect is a plan doomed to failure. You can quote me on that, as I've seen it applied to reality more times than I can count.

2 - "five nines," the mythical goal of DR planning, is 99.999% *percieved* uptime. Scheduled downtime is not considered as part of the downtime in calculating your "total nines."

In short, there are no datasystems that can be up 24/7 without proper maintenance windows. They simply don't exist. While there are many ways to eliminate a tremendous amount of uneccessary downtime (with the proper budget), nothing will ever remove the need for proper maintenance on the server level, even mainframes occasionally need maintenance.

Always remember that a single failed disk drive can bring even a redundant cluster crashing to its knees. Reality is a harsh mistress, an no less so in the datacenter than anywhere else.

Mike Talon

This discussion is closed to new replies.