In the InPhase release announcing its 515Gbit/square inch areal density achievement Wolfgang Schlichting, Research Director, Removable Storage, IDC, said: "InPhase Technologies' announcement is an important milestone in storage density, demonstrating impressive capacity increases enabled by holographic storage. The technology represents a potential alternative to incumbent technologies for archival storage requirements."
Potential, yes; likely, no.
LTO3 tapes have a 400GB native capacity and an 80MB/sec I/O rate. InPhase holo disks will have a 300GB capacity and a 20MB/sec I/O rate. Holo disks will rise, InPhase says, to a 1.6TB capacity. Tapes are heading towards a 1TB capacity. What interests me here are the I/O rates. I've touched on this before. It bears loking at again with the rise in holo's I/O rate.
InPhase is aiming its holographic technology at the media market. Let's take a 5GB file. How long will it take to read it from an LTO3 tape and a holo disk?
Currently the LTO3 tape will deliver it to a server in 62.5 seconds. The holo drive will deliver it in 250 seconds; say four minutes versus one minute. The tape is four times faster. There is an obvious implication here for backup. We backup file collections, not individual files at a time. A back-of-the-envelope calculation says that backing up 200GB to the tape will take 41.66 minutes. The holo drive will take 166.6 minutes, over two and a half hours.
No tape user is going to swap out their tape drives for a holographic disk backup system. The disks hold less than the tape and deal with data four times slower.
The holo market proposition could be parodied thus: we're offering you a medium that is four times slower than LTO3 tape and holds 100GB less. It's so bad that I'm wondering if I've missed something.
What does holo have over tape?
A holo disk is smaller than a tape cartridge. It should have a longer life; fifty years perhaps versus ten. The time to locate a file is probably faster because the holo device is random-access whereas the tape is sequential. The access time advantage gets less the longer the file to be read is. For our 5GB file restore above the tape can afford to take two minutes to locate the file and still deliver it a minute faster than the holo disk.
So the professional archive appeal looks to be long life. It's a kind of super UDO disk in a way.
Over time holo recording density will go up but, for me, the I/O rate is the big limiting factor. For holo to really take on tape in the general backup and archive market then it needs a four to five times increase in I/O rate.
For now holographic disk I/O rates are seemingly too slow to make the technology a generally viable backup and archive medium. Consumers could use it for personal media file storage as consumers are not going to buy an LTO3 tape drive. The InPhase consumer future is probably memory-card type size disks with many more times the capacity of today's flash-based devices. But the drives have got to be miniaturised and made affordable.
(Incidentally, the first version of our InPhase news story carried the wrong information about the areal density number and storage numbers. These have been corrected. I'm very sorry about that.)