Revisiting “Active-Active” vs. “Disaster Recovery”
I can’t count the number of times we’ve heard it over the years. When in discussions with an organization looking to do DR or Continuity, the initial answer to the “How much downtime can you endure?” question was very often, “Oh well ZERO, of course. We need to be hot/hot with zero downtime.”
That remark, of course, was then followed up with a detailed explanation of RTO/RPO, bandwidth and distance/latency limitations, and related costs, typically leading to an “Ok well, it’s possible…but you probably can’t afford the solution components required to achieve it.” In most cases, it was just a completely unjustifiable insurance policy. Thankfully, the majority of the applications didn’t REALLY need that level of availability and the more realistic RPO/RTO targets made the associated replication/failover solution more financially and logistically viable.
That perception of active/active (hot/hot) as being the impossible “holy grail” has been ingrained so deeply that many people, even today, tend to disregard it as an option completely. It became somewhat of an industry joke that only people who didn’t really understand DR even asked for it (outside of some very specific high-end environments). New York trading floors replicating to Jersey? Sure. For a typical enterprise where money actually ISN’T unlimited, probably not.
But, things have changed and the conversation is different now. Fiber is more plentiful and available at a much lower cost. In addition, the infrastructure hardware and software required to implement such solutions has gotten more functional, easier to implement, and perhaps most importantly, much less expensive. In fact, this has changed so much recently that not only is an active-active configuration viable for more customers, it may actually be LESS EXPENSIVE to own and operate than a more traditionally-architected Replication-Failover type of DR/Continuity solution.
Active-Active? Less expensive than Replication/Failover? Easier? Depending on your environment and requirements, it very well could be.
Granted, it still isn’t for everyone… and certain limitations remain (particularly around distance and bandwidth requirements). But it‘s definitely much more realistic now for more organizations.
So as a result our implementations of split-metro configurations are up dramatically over the last couple years. Instead of having to talk people out of wanting to do it, we now find ourselves having to raise it with customers as a very realistic (and potentially less expensive) option to the traditional cold/warm-site failover-based DR (with the associated failover complexity and time windows) they think might still be their only option. And this not just because it has gotten easier and less expensive to deploy. It’s also being driven by the fact that more applications really do now demand 24×7 zero-downtime access (not just in the event of unplanned downtime, but for system/site maintenance, upgrades and moves). IT environments are also now growing and changing so fast that more traditional Failover/DR architectures are increasingly difficult to maintain.
So if you are looking for an effective site-to-site HA/DR solution, are ok with the sites residing within the same metro area (you can still shoot a copy out of geo as well if that is a requirement), have reasonable access to metro fiber, and hate the idea of a bunch of idle resources sitting on the other side, you probably want to at least re-consider the active-active option.
Even if you have considered it before and thought (or were told) that it wasn’t viable or affordable, it’s worth another conversation. Keep in mind, that not all platforms support it as well or as easily and inexpensively as others. So it’s worth talking to more than one vendor as well as to people who really understand all the pieces.
VeriStor has extensive experience with these deployments and a variety of supporting platforms. Based upon our teams’ knowledge of the real-world pros and cons of various vendor implementations, we can quickly demonstrate the viability (in the right cases), as well as architect a solution that achieves your availability goals this in the easiest, most cost-efficient manner possible.
Just because it was “impossible” (or unaffordable) in the past doesn’t mean that it’s not possible now.
And if active-active HA sites aren’t quite a fit for your organization yet, but you still want a state-of-the-art, proven and cost-effective DR solution, then VeriStor’s fully-managed Cloud DR as a Service (DRaaS) would be another great conversation.