Quantum DLT-S4 review

Competition in the mid-range backup market is a lot more focused than in the SMB sector, as there are realistically only two main contenders. In the blue corner there’s the lightning-fast Ultrium LTO, while over in the red corner there’s the steadfast and solid SuperDLT (SDLT).

The DLT-S4 represents the fourth generation of SDLT and also a departure from its predecessors, as it’s taken well over two years to be released – the previous generations were each out with little over a year between each one. Interestingly, we don’t see the expected doubling in performance, although the DLT-S4 isn’t as bad as the DLT-V4, which offered only a meagre 25% increase over VS-160. The DLT-S4 delivers a more healthy 67% speed boost to an impressive 60MB/sec but, more importantly, it offers a huge leap in capacity to a native 800GB – twice that of the top-selling Ultrium LTO-3. Backward compatibility isn’t so good, as DLT-S4 can only read tapes produced by the previous two generations, whereas LTO-3 can read and write to LTO-2 tapes.

Two features that make the DLT-S4 stand out are Quantum’s DLTSage xTalk utility and its DLTice technology. The first is a diagnostics toolbox that can run a barrage of tests on the drive, while the second is Quantum’s response to the demands for standards-based data archiving facilities. Introduced with the SDLT600, DLTice writes a unique tamper-proof electronic key to tapes selected for WORM duties. This is carried out from the xTalk utility and ensures that the tape is now non-erasable and non-rewriteable – data can be appended but the tape contents can’t be altered. A big advantage of the Quantum method is that standard DLT-S4 cartridges can be used as WORM media.

Although not as fast as Ultrium LTO-3, a native transfer rate of 60MB/sec meant we’d still need a decent test platform to avoid introducing a bottleneck. To this end, the tape drive was installed on a dedicated Ultra320 SCSI adapter in a Supermicro dual 2.4GHz Xeon server running Windows Server 2003. Using Iometer, we found that local storage wasn’t going to be fast enough, so we called in Adaptec’s Snap Server 520 with the aim of utilising its high-speed iSCSI targets as a backup source. With Microsoft’s iSCSI initiator 2.1 logged on, we saw Iometer report an average raw read performance between the two systems of 105MB/sec – more than enough for the tape drive.

We copied a 9GB mixture of data typically found on a workgroup or SMB server to the iSCSI target and called in Computer Associates Brightstor ARCserve 11.5 to handle backup, verification and restoration tasks. The DLT-S4 impressed for backup speeds, with ARCserve reporting an average of 67.2MB/sec. Disk-to-tape verification saw a noticeable drop to 50.2MB/sec, with a full restoration of the test data to the iSCSI target returning 51.4MB/sec – not as fast as HP’s StorageWorks Ultrium 960e but still impressive nonetheless. To put these results in perspective, when we tested the HP LTO-3 drive using the same test data and backup software we saw it deliver an amazing 80MB/sec for backup. Verification and restoration were also faster, with the LTO-3 drive returning 72MB/sec and 69MB/sec respectively.

At this level of the backup market, the only choice is between SDLT and Utrium LTO. If the highest performance is a top priority, then the latter is the best option. But the massive storage capacity and superior WORM capabilities make the DLT-S4 a great alternative.