Disaster Recovery

“No one ever understood disaster until it came.”  Controversial writer Josephine Herbst penned that phrase decades ago when the words “information technology” were still tucked away in Bill Hewlett and Dave Packard’s little garage. But as our dependence on IT and the mass amounts of data it generates became the center of our business and personal universe, it wasn’t long before Josephine’s words were ringing in the ears of every network engineer and architect on the planet.

Disaster Recovery (DR) has always been looked upon as the “God forbid….” of IT design. But it’s often confused with the goals of High Availability (HA). 

 

  • DR retains data, HA retains service.
  • A good DR plan minimizes downtime. HA shoots for 100% uptime.
  • DR looks to minimize loss in the event of catastrophe. HA aims for zero data loss when operating properly.

But today, the lines between disaster recovery and high availability are being blurred thanks to hyperconvergence.

Maintaining two or more distinct data centers has long been the solution to avoiding substantial data loss if a primary data center fails or is lost to some sort of disaster. But a hyperconverged infrastructure, like those being adopted at hospitals and in major datacenters like those supporting Amazon Web Services (AWS) and Microsoft’s Azure platform, are making DR and HA two peas-in-a-pod.

What is Hyperconvergence?

Hyperconverged infrastructure (HCI) is all about the software that tightly integrates compute, storage, and networking in a commodity hardware box. ActualTech Media’s 2016 State of Hyperconverged Infrastructure report found that it’s one of the fastest growing trends in IT networking, with adoption growing by 54 percent between 2015 and 2016. And, while VDI was the most common use for hyperconvergence previously, more organizations are making use of it for things like ROBO (Remote Office/Branch Office). The main drivers of hyperconvergence adoption are a desire to simplify IT and reduce costs by no longer having to rely on one manufacturer, making one specific box, for one particular purpose.

Where HA and DR Meet

Traditionally, an organization generates data and stores it on hardware in a data center. That data is “backed up” to another datacenter in batches to ensure recoverability in the event of a catastrophe. But if disaster strikes in between back-ups (or the DCs fall out of sync), some data will inevitably be lost.

Enter HCI, a system in which there are two (or more) of everything – two firewalls, two servers, etc. – in constant communication. There’s no replication of data necessary; instead, data is written to multiple locations at once, automatically, and everything is stretched simultaneously over those data centers, connected by dark fiber. The instant one fails, the other takes over and is functioning without interruption. No flipping of switches or hours of downtime required. Data remains intact. High availability and disaster recovery become one and the same.

Hyperconverged.org outlines several additional benefits:

  • Elasticity – the ability to scale resources based on current business demands
  • VM-centricity – a focus on the virtual machine as the cornerstone of enterprise IT with all other supports revolving around VMs
  • Data protection – thanks to HA, data is easily restored when lost or corrupted
  • Mobility – greater mobility for applications and workloads
  • Data efficiency – a reduction in storage, bandwidth, and IOPS requirements
  • Cost efficiency – reducing waste through a sustainable, step-based economic model

Taken together, HCI makes for an increasingly attractive option for enterprises – such as healthcare organizations – needing to update their infrastructure while controlling costs and ensuring high availability.

Implementing HCI in the Healthcare Setting

We’ve successfully implemented hyper-converged networks using software-defined data center (SDDC) methodology at a number of large institutions.  With continued downward pressure to provide more services with less resources, hyperconvergence is a perfect fit.

The typical healthcare configuration is for three nodes on an active-active-active network (with each location connected by fiber within a ten-mile radius).  Complex failover processes are eliminated and uptime increases as a result. Scaling is nearly linear by simply snapping on additional, commodity drives, processors, and software elements to control them. One VertitechIT client reduced its new data center construction budget by 40%, saving over $3.5 million dollars. Healthcare Informatics called it the Healthcare IT infrastructure of the future.

It’s true. The impact of a disaster can only truly be felt after it occurs.  But once it does, you might thank your lucky stars in the HCI sky.