Download the authoritative guide: Enterprise Data Storage 2018: Optimizing Your Storage Infrastructure
What's on the Long-Term Horizon (36-48 Months)
I believe anyone who tells you they know what's going to happen with storage beyond 48 months, unless they claim more of the same but faster and greater densities, spent way too much time partying in the 1960s. There are far too many technology hurdles out there for any prediction to come true with any degree of certainty.
Even if the technology works, getting it from the laboratory to the manufacturing floor is a big issue. TeraStor had a few working laboratory units, but they were never able to mass-produce products. Seymour Cray had a working Cray-4 in 1995 and Supercomputer Systems Inc (SSI) had the world’s fastest computer in 1993, but mass production proved insurmountable for both. The list goes on and on — it's been happening since computing began. Remember the Babbage Calculator that could never be produced?
It would be nice if optical from companies like Inphase and others became a reality — proven technology that was:
- Highly reliable
- At least an order of magnitude higher in terms of density
- Faster than tape for load, read, write, and unload
The big question for my role as pundit predicting the future is consumer technology. If someone develops a consumer technology that is, say 100x DVD density, then I retract everything I've predicted. What could change the high end of the market is a viable consumer technology that can be morphed into a technology for the enterprise.
ATA disks might be a good example. IDE drives have been used in consumer and business PCs around the world for years; now given the cost per MB, every RAID vendor has or is looking at an ATA option as part of their RAID offerings for cost-effective storage. I see this as the morphing of high-end technology from consumer technology, and if another consumer technology becomes available during this period, we could see it happen again. A few important facts:
- SATA drives in RAIDs are not fully accepted in the enterprise
- The time to get the SATA drive integrated into the RAID storage has not been
- If SATA is successful, the SCSI and FC disk market will be hurt badly given the cost differences
Sites are just now feeling the pain of data migration given the decline in transfer rate-to-density ratio over the last 30 years, and the situation will not be changing anytime soon.
If transfer rates had kept up with density increases, we would be reading and writing tapes at well over 16 GB/sec. Moving to a new technology that is reliable, has good performance, and has great density will not solve all of the problems — you'll still have the age-old interface problem. What if someone builds an optical platter that can last 100 years, but does so with a 2-Gig Fibre Channel interface? Try finding an HBA, driver, software, or anything that can read it in 10 years, much less 20 or even 50 years.
The bottom line is that it is not the storage that counts, but rather the data. Someone needs to solve the data problem, and the storage solution will come with it.