It happened again, another industry veteran has dismissed tape as an obsolete technology. This time it was an EMC executive responding to the tired question “is tape dead yet?”, posed in a recent TechTarget interview. His response was “There is not a lot of new innovation on tape, but our customers need it and we are there for them”. Our friends at EMC get many things right, but this one is wrong. Tape is actually experiencing a renaissance of innovation that makes it arguably more innovative than disk based systems.
Disk drive technology is almost as old as tape and certainly can’t crow about its roadmap, since data density and performance for magnetic disk haven’t really improved for well over 10 years. While there have been some recent innovations in disk drives, like Seagate’s Cloud Ready hard drive, one could make the case (and several have) that if it wasn’t for flash providing a performance crutch, disk drives would be even farther behind the technology curve. To be fair, this executive probably meant tape in a backup role, but even for backup, there’s still a lot of innovation in the tape industry.
One example is LTFS, the open source file system for LTO tape that came out a couple years ago. This technology has been a big hit in the Media and Entertainment industry where users want to share data between applications and need a platform-agnostic solution that makes tape ‘file aware’. But non-proprietary solutions are not something that a lot of backup companies really want to promote since it makes backed up data completely portable. Backups made with Netbackup, for example, could be restored using any other backup application – or directly from tape as files, in many cases. It’s interesting to note that none of EMC’s backup products currently supports LTFS.
Much of tape’s history has been in backup but, as my colleague George Crump pointed out in “Tape is More Than The Backup of Last Resort”, its role shouldn’t be minimized in our disk-focused present. That said tape’s future is clearly in archive. Content repositories in the broadcast industry and the cloud-based consumer storage sites are all well into the petabytes. Keeping all that data spinning isn’t only expensive, it’s not needed. Most digital content is stored once and seldom recalled. So tape is a natural, especially when you take time to calculate the costs to store decades worth of data.
All the cloud-based “-as-a-Service” businesses, Infrastructure, Compute, Storage, Backup, etc, are facing the same issues, and tape is starting to provide an answer. While they won’t admit it, even companies like Amazon are using tape as Glacier is widely believed (and reported by several sources) to be a tape-based archive.
Tape is also now accessible via a REST (REpresentational State Transfer) interface, the same way data is accessed by cloud-based infrastructures – and by object storage systems. Spectra Logic has developed a new protocol called DS3 (Deep Simple Storage Services), which, ironically, is a modification to Amazon’s S3 interface. What this means is tape can now be directly controlled by applications that are REST-enabled, bypassing the need for complex archive software solutions or even traditional file systems in many use cases.
This is especially interesting, since most of these ‘hyper-scale’ environments are using REST-ful interfaces to manage the enormous data sets needed to support their applications. By teaching tape the ‘language of the internet’ DS3 is helping to break down the primary barrier this technology has historically faced, access time and integration complexity. This innovation can help tape maintain its role as the IT industry struggles with the Big Data Archive problem.
Disk drives have been the primary storage medium for the past two decades. And even though magnetic disk technology has lagged in performance and data density, it won’t be replaced in the near future. Ironically, it’s the other storage technologies that are keeping disk drives relevant. Instead of being a competitive threat to replace it, flash storage has helped disk address its latency problem.
Tape has done the same thing with disk’s storage density and long term cost issues. Far from being dead, tape’s future is brighter than ever. This is thanks to a steady roadmap of real improvement from the LTO consortium, ongoing innovation like the DS3 REST interface and, of course, the relentless growth in storage demand pushing web-based industries.