| The HP LTO Ultrium media cartridge is designed | | | | error rate. |
| and manufactured to optimize both media and | | | | The results for each of the two HP LTO3 Ultrium |
| drive performance. HP conducted an experiment | | | | cartridges, in each of the drives in which they |
| among different brands of LTO 3 ultrium backup | | | | were tested, showed consistently low and stable |
| data cartridges. The tests prove that users of HP | | | | error rates throughout the 200 continual backup |
| LTO3 Ultrium data cartridges could obtain twice | | | | and 200 continual restore cycles. The amount of |
| the data capacity per backup and significantly | | | | the data backed up each cycle was also |
| greater reliability. HP branded LTO3 Ultrium data | | | | measured to ensure that the backup cartridge |
| cartridges provided upto three times as many | | | | was continually providing the full maximum 400 |
| continuously successful backup and restore cycles. | | | | GB of data available. After writing 80,000 GB of |
| The price of HP LTO-3 Ultrium cartridge is less | | | | data, the HP cartridges were still providing |
| than the other competing brands. Testing in | | | | optimum performance and maximum capacity |
| extreme climatic conditions have demonstrated | | | | without risk of data loss |
| the reliability of HP LTO Ultrium media, where | | | | The results for each of the two Other Leading |
| data has been fully recovered from tapes that | | | | Brand cartridges, in each of the drives in which |
| have withstood the world's most extreme | | | | they were tested, were equally as consistent but |
| environments. | | | | showed a very different picture. After just 30 full |
| The test engineers took four LTO3 Ultrium RW | | | | backup and 30 restore cycles, less than a third of |
| cartridges. The two tapes were from HP and two | | | | the way through the first stage of the test, |
| from another leading brand and tested them in | | | | performance began to degrade. After no more |
| four HP StorageWorks Ultrium 960 drives. They | | | | than 60 continual backup and 60 restore cycles, |
| test was conducted in a controlled environment of | | | | the error rates were so high, on both cartridges, |
| 74ºF and relative humidity of 15%. | | | | that it was no longer possible to complete a full |
| In the first stage of the test procedure, the | | | | 400-GB backup operation. |
| cartridges were put through 100 continual full | | | | Two different types of failures occurred in the |
| volumes, 400-GB native capacity, backup and | | | | tests. These failures were directly related to the |
| restore cycles. This means that the total was 200 | | | | poor error rate performance. Both failure types |
| full-volume operations. The bit error rate BER was | | | | meant that it was not possible to back up the |
| measured every 2 GB throughout the process. | | | | maximum 400 GB of data. The first type of |
| Data error rates provide the most informative | | | | failure occurred as a result of reaching the end of |
| measure of performance. The data rate also | | | | the tape before completing the full capacity 400 |
| relates to how the tape is manufactured, the | | | | GB backup operation. |
| signal characteristics of the tape, its quality of | | | | Numerous repeated attempts to write some of |
| coating, and the surface roughness. If the error | | | | the original data were necessary, which resulted in |
| rates are high, the drive has to work hard to | | | | using more tape with the other leading brand |
| verify that data is being saved or restored | | | | cartridges, than was required when using the HP |
| properly. | | | | brand backup cartridges. The second type of |
| High error rate occur when the signal becomes | | | | failure was also related to the poor error rate |
| noisy and distorted. In the worst case, error rate | | | | performance of the other Brand cartridges. |
| levels ultimately reach a point where the drive | | | | However, this type of failure was due to it no |
| cannot cope with correcting mistakes and data | | | | longer being possible to verify the accuracy of |
| may be lost altogether. | | | | the written data even after repeated attempts, |
| HP has developed proprietary software for the | | | | causing the drive to fail and abort the backup |
| HP drives, this software acts like a powerful tape | | | | operation completely. Both types of failures |
| microscope and allows the user to analyze the | | | | meant a cleaning cartridge needed to be inserted |
| performance of the backup and restore cycles in | | | | before the test procedure could continue. |
| far more detail than would otherwise be possible | | | | The available capacity also deteriorated with |
| using off the shelf backup applications. In order to | | | | sometimes as little as 50% of the total capacity |
| ensure that the results of the testing procedure | | | | being utilized before reaching the end of the tape. |
| were not overly influenced by the performance | | | | Other leading brand cartridges did not complete |
| of individual drives, the HP LTO3 Ultrium cartridges | | | | the full 100 backup and 100 restore cycles in |
| in the Drives were exchanged with the other | | | | either of the drives in which they were tested. |
| leading brand tape cartridges for the second | | | | The high error rates can lead to: |
| stage of the tests. | | | | - Reduced cartridge capacity, so you have to buy |
| In the second stage of the tests, the backup | | | | more cartridges to back up the |
| tape cartridges were once again tested | | | | - same amount of data. |
| throughout 100 continual full volumes, 400-GB | | | | - Failed backup and restore operations because |
| native capacity. The backup, restore cycles and | | | | there are too many errors to correct. You lose |
| the bit error rate was again measured every 2 | | | | your data. |
| GB throughout the process. This brought the total | | | | - More frequent use of a cleaning cartridge. |
| number of backup and restore cycles throughout | | | | - Lower transfer rates (that is, poorer |
| the entire test to 400 (200 full-volume backup | | | | performance) because the drive will repeatedly |
| cycles and 200 full-volume restore cycles). The | | | | attempt to re-write or re-read the same user |
| average error rates were then calculated for | | | | data until it has been able to successfully verify |
| each full volume and any variation in performance | | | | that the written or restored data is 100% |
| during the course of the tests could then be | | | | accurate. |
| easily determined by observing any change in | | | | |