Quantum has nearly finished integrating ADIC into its own operations, with rationalised facilities and R&D, and it has emphasised that its focus is the backup, recovery and archive (BRA) market. It wants to be the largest independent, meaning server systems-independent, vendor of data protection products.

CEO Rick Belluzo says it is in that position currently, now that StorageTek has been sucked into Sun. (He’s quoting Gartner figures.) The target market is general data protection, including both enterprises and small, medium business (SME) customers.

There are three SME products generally: tape autoloaders; the newish GoVault removable disk; and stand-alone tape drives. (IBM OEMs the GoVault product by the way.)

Is a removable disk autoloader likely?

What may be coming here is a removable disk autoloader, one using rwemovable hard drive (RHD) cartridges. How else can the capacity of a tape autoloader be combined with the ease of use and speed of disk-based backup? This would follow the lead set by Iomega with its REV autoloader. However, using removable hard drives for long-term archival-type storage would involve buying lots and lots of RHDs, They would need to be little more expensive than equivalent capacity tape catridges. That pricing remains to be seen.

Imation is also selling RDX product as well as its own Odyssey removable hard drive product. RDX docking units are open, taking different manufacturer's RDX cartridges, whereas GoVault can only use Quantum cartridges. This may limit its future growth, as might not having an autoloader capability coming, if RHD-based archiving takes off.

For now Tandberg is selling bucket loads of its RHDs for immediate backup needs in SMEs, not for archiving. It has no plans to introduce an RHD autoloader as it sees no SME need.

Enterprise products

In the enterprise market Quantum's products include the traditional tape library products, DX virtual tape libraries, PathLight disk+tape backup products, and the newer DXi disk-to-disk backup products with deduplication and replication between sites for disaster recovery. The PathLight product is a combination of virtual tape library (VTL) and tape library with good movement of data from the virtual tape to the physical tape.

Quantum resell's Netapp's Decru division DataFort appliances to answer data-at-rest security needs.

Quantum's direction

Prior to the ADIC acquisition Quantum was decidedly tape-centric but that dependency is weakening.

Belluzzo says the tape market is a low growth one, with some flat, even negative areas, such as the SME tape market. The general data protection space is growing around 13 percent a year whereas the tape market on its own is only growing at 4 percent. So Quantum is going where the money is - into general data protection - and that means embracing disk and software.

Quantum's analysis of the market is that customers want a data protection offering that is comprehensive, meaning embracing the entire backup to archive spectrum, and encompasses remote offices. It embraces disk-to-disk backup and therein lie the problems and the opportunities.

Opportunities and problems

Opportunties are things like continuous data protection to greatly increase the granularity of data protection and data de-duplication to greatly decrease the disk space needed to store data.

One problem is that disk backup is embracing data de-duplication and tape cannot take advantage of it. You can only regenerate de-duped content on disk. In theory you could backup a de-duped data set to tape but you would need to restore it to disk, then regenerate the files, before you could restore them to end users. It's faster to restore a backed up file on tape direct to the end user.

Changing archive functions

An archive used to be defined pretty much by the media it was stored on, and that was tape. Its only use was to hold old files that had to be kept as a matter of record or to provide a disaster recovery facility for a tape library.

But now, in the age of compliance and legal discovery, an archive is a store for non-active data that must be considered to be facing access in the future, a store of historical record. That data is potentially of massive amounts and potentially facing access from a number of directions, meaning not just file name.

It's also, arguably, not a data protection technology. It is a historical record of mostly inactive data. For many suppliers data protection comes under business continutity and disaster protection headings. The technologies include disk ones such as RAID, snapshots, replication, disk-to-disk backup up and continuous data protection. Also, of course, the tape-based ones of backup and recovery.

A tape library was called an archive because it could hold immense amounts of off-line data. It was used as a data protection technology when disk-based offerings were too poor to support the gamut of restoration facilities needed. As the disk-based options have flowered so has a view that a tape-based archive is too slow to act as a data restore resource, tending to turn it into a lonh-term data store. But a tape library is a poor place from which to start building the ability to respond to compliance and legal recovery requests.

This on-going tension between tape libraries and archival needs has yet to be resolved. How Quantum defines an archive's functions will determine its archive product strategy.

Archive technology futures

Quantum's StorNext data management product is a file-sharing product aimed at workflow situations where data is dealt with in turn by various users on various systems. It is a data mover in an information life cycle management sense and, in effect, makes each accessing server a NAS head looking at block-based data stored in a SAN.

It can move data from primary disk, to secondary disk, to VTL and on to a tape library. In the StorNext world an archive is a tape library or a (not Quantum) optical storage facility. HOw does data get to a Quantum tape library?

With a non-de-duping VTL then PathLight makes perfect sense and there is a neat one line direction of travel from primary disk to secondary disk to VTL to tape library.

In a de-duping world that simple direct line is broken. Files cannot simply be transferred from de-duped disk backup to tape. They need to be regenerated first.

This is going to encourage storage engineers to look at random access (spinning disk) archive stores and away from sequential (tape) access archives.

For example they might be trying out combinations of technology such as a de-duping disk drive array, a de-duping MAID array or a de-duping optical disk array.

That's an archive storage media technology problem. Of more concern to Quantum is the archive software.

The restore-based file access of the past was targeted on a particular file. Today's compliance/discovery-based archive access is targeted at sets of files that meet some criteria, such as everything dealing with product X between dates Y and Z.

StorNext and DXi developments

The data in such archives needs classifying and indexing. StorNext has no such capabilities.

Belluzzo says that StorNext will be developed to expand its market focus and build out new solutions to better meet today's data protection needs. He's said that a de-duping capability will be added to it.

That signals StorNext has a strong role to play in Quantum's integrated product range plans but it also suggests that some form of disk-based archive may be coming, built on DXi technology, a vastly more capacious DXi system - a DXi MAID system perhaps?

If Quantum is going to supply products to provide archiving in the compliance and legal discovery sense then it needs indexing and data classification technology. If it had such technology then StorNext would be the natural 'container' for it.

None of this is fixed. But the backup, recovery and archive market has started out on a journey with disk-based storage and the ultimate destination involves a significant weakening of tape-based systems. The speculation is that Quantum is likely to introduce several new products to take increasing advantage of disk-based data protection, and fill out its line of products; some missing pieces being:-

- data classification and indexing extensions to StorNext,
- a DXi archive with a capacity vastly greater than the current half-petabyte or so DX15500.