As we discussed in our recent article "Comparing LTO-6 to Scale-Out Storage", tape has several meaningful advantages over disk. The single biggest advantage is that tape is continuing to get faster. Disk has been stuck at 15,000 RPM for over a decade. If we look at LTO as an example in the tape industry, it has gotten progressively faster with each iteration. When LTO-6 ships, it will be able to transfer about 525 Mbps with compression, faster than many servers will be able to deliver that data. Remember these numbers are per drive, without some sort of scale-out storage infrastructure.
Now clearly this advantage is when the data on tape has been found and just needs to be transferred back. You need to add a minute or so of seek time to find the data. On large transfers, though, tape should outpace most disk systems. From an ingest perspective, LTO-6 and other enterprise tape formats may be unrivaled when compared on a single unit basis.
The delay in getting to first read is the flaw in my prediction of tape replacing the hard drive. Disk will have a role so that restores, especially of smaller data sets, don't have to wait for tape systems to find the position of the data on the tape. That is certainly going to be the case for the next several years, and the strategy we articulate in the above article of leveraging both scale-out storage and tape is the way to go. That said, if you project a little further in the future a tape-only or tape-mostly strategy could be feasible.
Tape replacing the hard drive is as much dependent on the success of solid state storage as it is on tape's own advances. There are two use cases where solid state can help the tape industry. It may be a long time before we see SSD eliminate hard disk drive, but as the use of SSD increases the HDD has to find a new role.
Two of the trends we are seeing in flash is the move to server-side caches built from PCIe SSD and the move to 100% flash-based storage systems in the SAN. While these two compete with each other and most customers will likely pick one strategy or the other, both strategies lead to a changed role for disk--one that is capacity centric and less performance centric. Sounds like tape.
>As an example, in the server-side cache use case, some vendors can provide PCIe SSD cards with over 1 TB of flash capacity and they can support multiple cards in the server. What if every physical server had 3 cards RAIDed together for about 2 TB of protected, local cache? The chances of a cache miss plummet. If you are 99.99% accurate requesting data from cache, then you might be able to push that non-active data set to a tape-based library, with a small hard disk area. The concept is a little out there, but considering how we have progressed thus far it is not out of the realm of possibility.
Follow Storage Switzerland on Twitter
Security concerns give many companies pause as they consider migrating portions of their IT operations to cloud-based services. But you can stay safe in the cloud. In our Cloud Security report, we explain the risks and guide you in setting appropriate cloud security policies, processes, and controls. (Free registration required.)