We're making some pretty big changes in how we deliver high availability and disaster recovery, and to do that, we have to change how we look at HA & DR. But in order to do that, we first need to debunk a few myths about clustering that seem to have crept into a lot of people's heads over the years.
This is the first in a series of posts I'll be putting up here at Symantec Connect over the coming week or so in which I'll lay out some common misconceptions of HA clustering and explain why they're wrong. Here's an example of what some people believe about clustering and which I'll refute over the coming days:
- Clustering is unreliable
- Clustering is too complex
- Clustering is expensive
Sure, like any other myth these have their origins in someone's actual experience, but just like the mythology of ancient Greece, there's a yawning chasm between these myths and reality. Sorry, Zeus.
How did these myths come to be? Pretty simple, really. Combine equal parts unrealistic expectations, inadequate training, inferior cluster products, and poor internal processes, blend with a dash of short-sightedness, and you've got the perfect Confusion Souffle right there. OK, maybe I'm taking the cooking analogy a little too far, but you get the idea.
The problem began with how we viewed HA clustering early on. We always focused on clustering the big, backend database server, which more often than not also hosted the application that was using that database. But application architectures and other data center technologies evolved while our view of HA clustering didn't. Well, actually, our view of clustering evolved, but we weren't always very good at communicating that evolution to our customers.
So, starting today, we're hitting the big, red "Reset" button on how we look at high availability, disaster recovery, and - most importantly - application management in the modern data center. Stay tuned.