Comment A disk drive array can't cut it: Oracle says its 31 January game-changing announcement will provide a "dramatic leap in storage technology" for enterprise data centres. A disk drive array alone can't provide that. What can the technology be?

Whatever it is, it will offer, Oracle says:-

Much lower costs—per terabyte of capacity and floor space consumption
Dramatically higher performance — let’s just say the old benchmarks just got older
Greater capacity — in a much smaller footprint
Greater capability — optimized for archive and data protection

These are the four boxes the announcement will tick. How can a traditional-design storage array, using hard disk drives (HDD) and possible/probably solid state drives (SSD), tick all four? Maybe a MAID (Massive Array of Idle Drives) implementation could provide the implied increase in TB/sq ft but it wouldn't provide the dramatically higher performance.

An all-SSD array would provide screaming performance but fail the first box, the "much lower cost" one. How about a new tape format?

We can see incremental storage array increases coming but not game-changing ones, not with Dell, EMC, HP, HDS, IBM, NetApp and Oracle all trying their damnedest to push the storage array technology envelope as far and as fast as they can. It beggars belief, I'd say, to think that Oracle has found a way to deliver game-changing drive array technology that has escaped the others.

Which leaves me thinking there could be a tape element to this announcement. Does this resonate with you at all?

There are a couple of pointers to this notion; for example, the Oracle webcast lead is Jim Cates who was running Oracle's tape HW products in the summer. The Oracle text talks of a dramatic leap in "storage technology" and StorageTek, Oracle's tape unit, is a condensed way of writing storage technology.

Two possibilities come to mind. First, suppose we think of a hybrid product technology combining a drive array with an enterprise tape library? I'm thinking of a combination of an Open Storage 7000, the ZFS-using storage array, and a StreamLine 8500 tape library. Is that completely off the wall?

The tape library would then have a thumping great, server-heavy, caching drive array in front of it that could act as a virtual tape library (VTL), and the hybrid device offer the benefits of a combined VTL and PTL (physical tape library) in one product. Add in new 5600 Intel processors to give the 7000 a speed jump, and we have ourselves technology that could start to tick the four features above.

Secondly, we could envisage the third iteration of Oracle's T10000 tape technology with the current T10K B (1TB raw capacity, 120MB/sec throughput) giving way to, say, a T10000 C, with 3 - 5TB raw capacity and 180 - 270MB/sec throughput. That has been shown on an Oracle tape roadmap.

Several industry sources say tape is going to feature in, or be the main focus of, Oracle's announcement.

One industry insider said: "I think Oracle’s announcement timing is very likely aligned with the large growth the tape industry is seeing in the unstructured data being archived off to tape.  This is a huge area of growth in the storage industry right now.

"The four value proposition bullets [above] sound spot on for tape library demands from the archive market...  I think it is quite likely Oracle will have a tape drive or media component to its announcement."   ®