Ecrix VXA-1 In The Spotlight.
Ecrix's VXA drive is an interesting entry on the market. The manufacturers claim that the drive has a native transfer rate of 3MB/sec. The V17 tape promises a native capacity of 33GB. With compression, these numbers can be expected to be about double the native numbers.
Although the drive is, at its heart, an 8mm helical scan tape drive, using technologies similar to Exabyte's 8mm drives and Sony's AIT drives, it also has many differences. Perhaps the most prominent is its sub $900 street price--the nearest 8mm competitor on price (and not capacity) alone is Exabyte's Mammoth LT with a street price of about $1,250 (a full 1/3 higher). By contrast, the Mammoth LT writes to tape with a 14GB native capacity and has a transfer rate of 2MB/sec. On price alone, the VXA-1 looks like a bargain when compared to its 8mm cousins.
Other than the fact that all 8mm drives use cartridges that appear to be roughly the same size and all have a rotating 8mm drum and a somewhat similar tape path, many of the similarities end there. The VXA-1 has four heads, all working simultaneously, implementing what Ecrix calls Discrete Packet Format (DPF). Data is written to the tape in small packets. Each packet features error correction code. Each packet is individually and uniquely numbered. The packets are assembled into a two dimensional matrix-error correction code can rebuild lost packets based on vertical, horizontal, or diagonal error correction data. The result is purported to be extremely error tolerant and highly accurate. Specially developed ASIC accomplish the near miracle of packet writing and packet reassembly.
Additionally, Ecrix drives include what is referred to as Overscan--each packet passes at least two heads. If one head doesn't correctly read the packet, a second one gets another crack at it. Further, motor speed control allows the drive to adjust to the speed that the computer can support--if a data stream from the computer slows down or speeds up, the VXA can accommodate to the change by changing the speed of its motor. Similarly, if data going from the drive to the computer drops in speed (for example, when the computer's data buffer is full and it can't accept data as quickly), the VXA can slow the tape to accommodate it. The speed control reduces the amount of rewinds and restarts required to position to the tape to the head, in order to read the next bit of data or to write to the next open spot. Overall, the technologies designed into VXA-1 are said to require lower tolerances than are necessary in other 8mm technologies, in addition to reducing wear on the drive and media.
The packet writing format is also said to render damaged media more easily recoverable. Torture tests performed by Ecrix, including immersion in hot coffee, immersion in boiling water, and freezing yielded media that, once dried, was recoverable in VXA drives. In theory, a cartridge written by any VXA drive can be read by any other drive.
Can the VXA-1 drive live up to its claimed performance? Computer Technology Review puts it to the test.
We reviewed a VXA-l external drive. The drive is an attractive unit with a diamond-shaped eject button at the lower right of the front panel. To its left is a multi-function LED that is designed to blink when the tape is being read or written and to blink various diagnostic codes. The LED also changes color to indicate errors.
A three light array indicates the motion of the tape--a left arrow, right arrow, and middle block communicate the motion (or lack of motion) of the drive. The external unit has a blue power on led. At the rear of the enclosure, an ID selector enables selection of SCSI ID. Two SCSI connectors, a power connector, power switch, and DB9 jack for connecting a serial tester are also located in the back of the drive. Oddly enough, the design of the drive makes it easy to get confused and mount it upside down. VXA upside down is still VXA and the external enclosure has similar feet on top and bottom. A bit of care can determine the right orientation. The drive we received came with VXA 17 tape cartridges, a tape cleaning cartridge, a power cord, SCSI terminator, quick start guide, packing list, and CD-ROM with utilities and documentation.
Installation and setup were easy and straightforward. The default SCSI ID setting of three was a good guess and the drive was recognized by the Adaptec 2940U2W controller to which it was attached as an Ecrix drive.
It should be no surprise that software that was shipped before the drive was announced might have problems working with the drive because the drive is new. In fact, NovaBack 6.0 was not able to recognize the drive.
A Drive Preparation utility, included on the CD-ROM shipped with the drive, solved the drive identification problem. The utility allows you to select the drive identification number, the SCSI adapter, and the application that will be used to perform backup. Once selected, the utility changes the drive's ID string, if necessary, so that the software can recognize it. When the system is restarted, the drive may no longer appear as an Ecrix drive, if the target application won't support it. In the current example, when I restarted the system, my drive was no longer an Ecrix drive-it had become an Exabyte 8900 drive.
Once the change in drive identification was made, NovaBack had no problem recognizing or communicating with the VXA drive. During testing with other backup applications, the utility changed the drive's identification string between Ecrix and Exabyte with no apparent changes in performance.
For the purposes of testing, multiple data sets, discussed in more detail in this issue's Notes from the Lab column were used. These sets ranged in size from a few hundred megabytes to more than 50GB and included as few as one drive and as many as 13 logical drives. Data sets with few compressed files were used, as well as sets that included many previously compressed file types. Software compression was disabled, allowing the VXA-1 drive to assume full responsibility for data compression.
Dantz Replica Backup and two versions of Veritas Software Exec (3.0 and 4.2) were also used for testing the drive. All three software packages were able to identify the drive as an Ecrix VXA drive and were able to work correctly with the drive.
The VXA drive was put through multiple backup tests. Tests were performed using multiple data sets and four different backup software products, as noted above.
Initial testing yielded results inconsistent with quoted specs. However, after replacing many system components, as detailed further in Notes from the Lab, data backup and restore performance closely matched Ecrix's quoted specifications for native performance.
Dantz Retrospect 5.0 reports backup performance for each drive being backed up. A data set that excluded most compressed file types yielded results ranging from 132.7MB/mi (slightly over 2 MB/sec) on 7.3GB of data on an IDE drive running in ATA-33 mode to 209.2MB/mm (about 3.5MB/sec) on a SCSI LVD drive with 3.4GB. A small data set of 64.6MB was recorded at 215.1MB/mm, although this should not be considered representative of overall performance.
More typically, performance ranging between 150MB/min and 180MB/min was recorded by Retrospect when data was being read from SCSI LVD drives. Smaller partitions on the IDE drive also yielded performance figures ranging from 150MB/min-200MB/min.
These results are consistent with those found when backups were run with Backup Exec 4.2. Backup speeds ranging from 9GB/hr to 11GB/hr were observed during backup.
NovaBack+ 6.0 similarly yielded overall performance figures ranging between 100MB/min and 150MB/min (2.5MB/sec) on multiple data sets. Excluding certain compressed files from backup seemed to improve performance, although higher performance was noted when similar backups were performed using Backup Exec or Retrospect.
Compression was enabled on the VXA drive. During testing, software compression was disabled on the backup software. Possibly as a result of a large number of already compressed files on the drives being backed up and a smaller percentage of highly compressible data files on the drives, I was unable to get the high storage capacities claimed by Ecrix for its V17 cartridge. When most of the compressed files were excluded, the maximum amount of data that could be put onto one tape cartridge was about 39GB. When all files were backed up, data cartridges stored approximately the stated capacity of 33GB per cartridge. Presumably, a different mix of data types, including a larger proportion of highly compressible data, would result in significantly increased storage capacity on the V17 media.
Ecrix emphasizes recoverability of data from damaged cartridges and across multiple drives. Promotional materials have noted that cartridges have passed multiple torture tests, including freezing, immersion in hot coffee, and dropping into hot coffee. These tests were not replicated for the purposes of this review. Restore tests run on the VXA drive yielded no failures or error messages, as should be expected on a practically new drive.
The VXA drive included a QuickStart guide, a cleaning cartridge, a tape cartridge, and a CD-ROM with configuration software, utility software, and the contents of the Ecrix web site. The manual was stored on the CD-ROM in PDF format.
The software was essential for configuring the drive for use with backup software that did not recognize the drive. This software worked flawlessly, changing the drive identifier to make it appear like an Exabyte 8900 drive.
The Ecrix tool software enabled drive diagnostics, logging of drive performance and BIOS update. During testing, the drive BIOS was updated, using code available from the Ecrix website (www.vxatape.com).
Technical support was friendly and generally on target with its recommendations. During the course of this review, the tech support department was contacted numerous times. Although there was a technical support person dedicated to support reviewers, I contacted both the special dedicated person, as well as the general tech support department. Wait times for tech support were generally short.
The technical support department offered general suggestions for improving the performance. Ultimately, replacement of the host system's motherboard and SCSI controller solved the throughput problems. Moving the drive to a second SCSI adapter, for exclusive use by the drive, provided minimal performance benefit, although it is fairly obvious how dedicating a device to its own HBA can theoretically improve performance. The Ecrix technical support department clearly demonstrated their concern about a customer's drive performing properly.
The documentation provided was basic. Specifications, more detailed installation instructions, troubleshooting, and a glossary were primary portions of the documentation. For most users, the information is probably all that's necessary for installing and maintaining such a low maintenance device. A section on how to remove a cartridge that doesn't eject may be useful for users who must get the cartridge out. The 66 page electronic manual is adequate for its purpose and intended audience.
The Ecrix VXA-1 tape drive was a solid performer, flawlessly backing up data at a rate of between 8GB/hr and 12GB/hr in our testing. Speed varied based on a number of factors, including types of files being copied, source drive performance, and backup application being used.
Capacities of at least the rated 33GB on a 170m (V17) cartridge were achieved with the highest capacity of nearly 39GB. Presumably, with more compressible data than that used in our test data sets, much closer to the 66GB compressed capacity will be achieved and recorded in other environments.
With pricing beginning below $900, the VXA tape drive has been favorably compared to higher end 8mm and DLT drives. In its price range, its price target is closer to DDS-4 than it is to either Mammoth, AIT, or DLT drives. Benchmark's DLT1 drive at about $1,200 may be its nearest capacity and performance competitor.
Qualstar and, possibly, other companies are incorporating both Benchmark's drive and Ecrix's VXA tape drives into automated systems. Benchmark offers an automated system based or its DLT1 drive. Although these drives may not meet the performance levels of their AIT, Mammoth, and DLT cousins, they shave thousands of dollars off the price of automated products. With automation intact, eliminating the need for an operator to switch media, lower cost automated tape systems based on the VXA-1 may suit the needs of companies with moderate daily backup requirements.
Editors Note: If you like shirtsleeve reviews and want more, e-mail Katherine Diaz at firstname.lastname@example.org If you want more information on the methodology, go to our website--www.wwpi.com.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||Product Information|
|Publication:||Computer Technology Review|
|Date:||Feb 1, 2000|
|Previous Article:||The Fundamentals Of AIT Technology.|
|Next Article:||Notes From The Lab.|