Magnetic Tape Deterioration and Data Recovery 4. DATA AUDIT In a data audit we select a sample of tapes and cartridges from the total data archive and run a series of tests to determine their condition. Clearly the number of tapes selected needs to be large enough such that it is statistically representative of the entire dataset, but small enough so that the cost of the audit is not excessive. The method that we use for the sample selection is to sort the tapes into batches, where the tapes in each batch would be expected to exhibit similar characteristics. For seismic field tapes, it is likely that all tapes from a survey would have been recorded by the same contractor at the same time on the same brand of tape, and that they would have been stored in the same environmental conditions since. A seismic survey therefore constitutes a batch of tapes. For seismic processed tapes normally stack and migration data are stored. These may have been produced by different contractors at different times, and must therefore constitute separate batches from the field tapes. For well log tapes the sorting into batches is less straightforward. Ideally a batch should be defined as a visit to a well site by a contractor, but in practice this would mean a large number of batches each comprising very few tapes. We therefore define a batch as tapes recorded in the same year on the same brand of tape. From each batch of tapes a random sample is selected for auditing. Normally the sample would be 1% of the batch, with a minimum of four tapes from each batch. The end result of this selection process is that typically 1-2% of the total number of tapes are audited. During the audit process a number of tests are done to determine the media condition. These include: physical tape errors - any gaps due to missing oxide, edge damage, etc. signs of stiction - this can be seen when the tape is run through a tissue tape cleaner, number of corrected errors - errors detected by the tape drive which it was able to correct in reading the tape number of uncorrected (parity) errors - errors detected by the tape drive which it was not able to correct. It is worth pointing out that a series of other tests are made during the audit to check the validity of the data on the tape, to check the contents of the tape compared with the information on the tape label and the database index, and to determine the tape characteristics (for example, which are in non-standard formats). The results of the audit are examined carefully to determine the batches of tapes where problems exist. A report is presented showing which batches of tapes/cartridges need urgent recovery, which are showing early signs of deterioration but are not yet in too bad condition, and which batches are still in good condition. We strongly recommend a data audit to all companies who want to preserve the integrity of their database. Table of Contents INTRODUCTION TAPE MANUFACTURING PROCESS TAPE DETERIORATION DATA AUDIT DATA RECOVERY SUMMARY See other technical papers.