First of course you call the National Enquirer because, after all, they are the only publication with the depth of reporting good enough to get this story the attention it deserves.
Joking aside, let’s say a year ago your business was doing what it should where its cloud upkeep is concerned. Your company made High Availability a priority in its architecture. However, your cloud wasn’t designed for multi-zone tolerance in the case of a datacenter going down.
And then the data center goes down. As we saw with Sandy, that fear is more real than many cloud providers want to let on.
Fast forward a year, you’re fully HA, running in Amazon in multiple East and West Coast availability zones. But what if a major storm (or asteroid or falling folk hero in a sleigh) knocks out the East Coast data centers?
If your business’ load balancer is distributed, you’re ok. But for this sake of this example, let’s say the business only located its load balancer on the East Coast. In this case, even with HA extending across the US, without the parallel distribution for load balancing, the business is still down, losing untold amounts of revenue for every hour it can’t connect with its customers.
Cloud is a steep learning curve. You won’t get everything 100% right on the first try. There will be downtime, which results from experimentation more often than outright catastrophe.
Partnering with a good cloud provider is key to ensuring that your downtime isn’t disruptive to your business continuity. A good cloud partner can help evolve your business from a site that’s doing reasonably well in a private data center to be a multi-AZ, fully bulletproofed, redundant, scalable cloud your business needs to make a market impact.
Scalability and high availability are of course related. You always want to ensure that your functional workload can be maintained while accounting for access to the resources necessary to scale up to meet increased demand. To maintain kind of cloud that won’t fall victim to a Santa crash, a Sandy, a Mayan Armageddon, or any other insane event, the key is to have the extra workload distributed across each availability zone.
To achieve this requires many different parts. You need geo IP load balancers, so your customers can be automatically directed, based on their actual location, to the nearest load balancers. Beyond this, having database that can handle the latency between the zones is central to success. You also will need programming schemas that account for being in different sites.
In the end, doing all this yourself is completely possible. But having a cloud partner that can guide you and your business through this process, smoothing the path so time from planning to implementation to utilization is as easy as possible, is a real boon to any organization seeking to make cloud computing systems a strategic requirement for their growth goals.
By Jake Gardner