Who Turned Out The Lights? - Disaster Recovery Strategies


nyc-image-picjumboThe post 9/11 years in New York were a breeding ground for DR solutions that few were even imagining in the years prior. There was a time when the worst case scenario was losing a drive in a RAID0, or two in a mirror. After 9/11 there were greater things to worry about, as we were reminded in August 2003 when the lights went out in NYC. We considered what would happen to the shows that we're delivering. Would we be able to continue production on those in the near future if the power didn't return? What if this was an attack, and we were vulnerable to this happening again? Luckily that event was not terrorist-related, and power was back in a couple days.

But the seeds were sown - I started to get requests for off-site storage, and started developing new multi-dup policies of all masters (all tape-based back then of course) and keeping them in a secure location outside the city. Once footage was digitized, standalone spinning disks, DVD-R and new LTO technology became a hot commodity.

A few manufacturers had been offering accelerated data transfer over the internet, mostly for delivery and R&A, but then they also started packaging backup and archive services. The problem ten years ago was the infrastructure. Even selected media transfer (project metadata, renders, mixdowns etc.) through standard transfer protocols would take hours.

Now in 2014 we have options. Still the most popular is spinning disk. If you're one of those facilities with a couple hundred Lacie-type 2TB USB/ESATA drives up on the shelves of the media vault, you know what I'm talking about. Nothing beats it for ease of use and portability. You can plug it in anywhere and get the data. Not the best for reliability, as you will find out when 10% of the drives on your shelf refuse to spin up next year.

Enter the cloud drive.

Someone else has the responsibility be sure the data is safe. They present a location to you, and measure the data being added, bandwidth being used, and charge you accordingly. This is the hope and dream of cloud archive.

  • Problem #1: It's long-term archive, not backup.

Backup requires replication of all data in a state that can be readily retrieved when needed. The trickle of data to and from the cloud drive, although fast by 2003 standards, is still going to hold you hostage for days when you have several projects to recover.

  • Problem #2: Cost.

Like any service, the more you use it, the more it costs. Your SAN system will likely top 40TB right out of the gate, and expand to over 100 easily. What's a sustainable cost/GB for a cloud service? Can you charge for redundancy?

The answer seems to be somewhere in the middle - cloud or tape for long-term archive, spinning disk or managed tape for backup. Full replication for backup, selected consolidated media sets and project metadata for long-term archive. For TerraBlock, we have qualified software that runs on the server itself, and makes local disk, local tape, remote network drives and cloud storage available from the same interface. Scheduled daily backup, deep archive or DR copy can be run directly from the storage server, and automated based on the location, type and date of the source files. We call this SyncBlock, and like the rest of our product line, focused on simplicity and functionality.