The Death of Tape?
The Death of Tape?
Tape has been with IT professionals for a very long time, but over the past five years the calls for its demise have accelerated. So is tape really dead? In my opinion the answer today is no. Clearly the technology to replace tape exists; disk to disk backup, disk archiving, and electronic vaulting of data all punch holes in tapes value. The problem is that few environments today, especially in larger data centers, have all three of these solutions. To eliminate tape you would need to have all three of these technologies in place.
Disk answers much of the challenges created by tape, but it is only recently that those disk products began addressing and improving on what tape did well; store data quickly, store data cheaply, impressive scale by simply inserting more blank media, offer portability for shipping data off site. Tape has performed all of the functions well but disk is catching up in most of those categories.
Until the last few years tape was the primary target for backup, but now disk is slowly eating away at that. Tape became a good secondary target because disk was too expensive to hold more than a few weeks worth of backup. Disk was essentially a cache, then data deduplication became a mainstream capability and is now a must have offering. By eliminating duplicate data (the similarities between this weeks full and last week’s) it made disk significantly more cost effective. Because of data deduplication, tape’s role was limited to long-term storage (archiving) or archiving and, because of its continued portability, for DR.
Disk Archive Technology came to market like those offered by Permabit and Copan. The emerging cloud storage solutions like Nirvanix and Parascale, are starting to have an impact. The ability to build disk pools with massive scale that rivals tape now exists. Data deduplication devices have also solved portability of data to a large extent because of their ability to only replicate net new or changed blocks of the backup. Getting tonight’s backup off-site, electronically, is very doable today with disk to disk and deduplication.
What is tape’s remaining role? It essentially fills the gap between these solutions. Most data centers cannot afford to implement all of these solutions at once. Additionally the tape manufacturers are not going down without a fight, and in fact last quarter tape drive shipments were actually significantly higher. The capacity on each tape continues to expand so it continues to have a cost advantage. Tapes index and positioning on modern drives is improving substantially so search is getting better.
What we find is that in the very large enterprise tape is not going anywhere, they simply have too much data to eliminate tape. In the large to medium size enterprise, while it may be more feasible to eliminate tape, you have to commit completely to disk. This is done by implementing disk to disk backup with deduplication so it can deliver a solid, cost effective backup target that can be replicated for disaster recovery and a disk archive for long-term retention of data. All these components must be in place. Without a 100% commitment to disk it is difficult to eliminate tape.
Even in the medium size data center tape lives on and has a valuable role as a cost effective storage platform to bridge the gap and given enough time, tape may surprise us all and become once again too compelling to pass up. Tape’s challenge is that disk solutions will continue to get more robust, drive densities will increase and the cost of those disks will continue to be driven down. Will tape eventually die? Only time will tell but clearly there is a lot of technology investment going on to make that happen.
In a future entry we will discuss how to improve your tape media’s reliability and recoverability.
Thursday, June 26, 2008