Tape as a storage medium has fought off all attacks from disk-based backup and archive vendors and is enjoying a resurgence in the data center. In fact, Storage Switzerland recently polled about 70 large enterprise IT professionals and not one of them indicated that they had displaced tape in their environments. Furthermore, only three of them had plans to reduce their commitments to tape based storage. So much for tape being dead.
The resurgence of tape was on display at the Fujifilm annual Global IT Summit in Houston, featuring CIO and IT executives from around the world. Storage Switzerland and other Industry Influencers like Jon Toigo and Mark Peters, were in attendance.
At this summit an interesting idea was proposed by Toigo and others; is it time for tape to be used in production? Ironically, tape has been used as a production storage device for decades, starting with IBM mainframes. But with the advent of high speed disk drives and now flash media, tape, in non-backup implementations, seems to be relegated to an archive role.
I have written several times that tape should continue to play a key role in backup and not just focus on archive, but have rarely discussed its viability for production data. Until recently there has not been a strong case to do so. But three key factors are making tape an attractive, viable alternative for production use cases today.
First, long term operational cost is weighing on IT planners, as they ponder the hard reality of maintaining disk farms that spin forever. Second, tape’s portability is a massive advantage over disk. If you only have a few GBs to deal with disk-to-disk over the internet will serve you well, with a few PBs tape and a FedEx truck is going to serve you far better. Finally, tape access has improved substantially over the last two to three years. It’s no longer a complicated process to get data on and off tape. It can now be done without a dedicated application.
What Makes Tape Viable for Production Data Today
Cost and transportability are the obvious reasons. And really there is nothing new here, except that tape keeps getting more cost effective and more reliable to transfer. A key point that I like to make is that even if you got your disk for free the cost to build out data center space, to power and cool your “free” storage and then retain information for decades would break your budget. Buying the TB of storage is cheap, maintaining it year after year, for decades, is really expensive.
For tape to truly be considered a production data solution it has to be “clickable”. This means the difficulty of user access must be lowered to a mouse click or an application programming call. As Jon covered in his presentation, LTFS starts this conversation. And as Molly Rector, Chief Marketing Officer at Spectra Logic made clear in hers, DS3 puts an exclamation point on it.
As we describe in our article “What is LTFS?“, LTFS turns tape into a mountable file system. In loose terms you can think of it as a USB hard drive, just with a lot more capacity and in a form factor that’s safe to ship around the world. Solutions like those from CrossRoads Systems, IBM, Quantum, etc. all bring LTFS to libraries and add a NAS front end. Spectra Logic’s DS3 as we covered in a recent webinar, ”Driving Down the Cost of Forever”, allows tape to be accessed via a RESTful API. This means the cloud providers with experience in REST programming can access tape from within their applications and not have to integrate complicated SCSI commands.
Being able to access tape as easily as you access any file system means that it could be viable as a repository for a lot of production data that does not need instant accessibility. Or it could be used in workflow driven environments where data access requirements are predictable.
If you know how to use a NAS share or how to program with a RESTful API bringing tape into production is now a reality. The cost savings over disk-based long term storage can be considerable and the ability to retain information forever an invaluable asset. The next step is to select and design a tape production architecture, which will be the subject of my next column.