Introduction

Most of the physical tapes used today for backup purposes are LTO tapes. The vendors are specifying a "native" and a "compressed" tape capacty. The "native" capacity is basically the amount of data the tape can hold without any compression. But as these devices are compressing the data, it is expected more than the native capacity will fit on the tape. That is why a specific compression ratio is assumed by the vendors to be able to specify the "compressed" capacity.

Question

In a real-life situation, also with Data Protector, it's often seen that less than the compressed capacity is fitting on the tape. In some cases, it's even less than the native capacity. Why?

Answer

The first situation (less than compressed capacity) is rather easy to understand.

• The most obvious reason will be that the assumed compression ratio is not reached. As the compression ratio is depending on the data type (how compressible the data is), it will be impossible to predict the ratio. If the data being backed up is less compressible than what the tape vendor has assumed as an average then less than the specified data will fit on the tape.
• It is also needed to understand that more data will have been written to the tape than the backed up user data only. The DP tape format does not only consist of data segments, but also catalog segments which can take a fair percentage of the total written data on top of the amount of data being backed up.

But if it is assumed the compression ratio will never be less than 1 and the meta-data written to the tape is also taken into account already then how is it possible to see less than the native capacity fit on the tape?

• Most likely the data transfer speed will have been too low which resulted in "padding". As the drop of the tape speed (in a case where not enough data is streamed to the device) is limitted, the device will use this padding mechanism when the minimum tape speed has been reached. It basically means that the uncomplete buffer will be filled with padded data, so the buffer can be flushed to the tape as a full one. But that means that less "real" data is fitting on the tape and this can even lead to less than the native capacity fitting on the tape.
• Another influencing factor may be the number of recoverable write errors. Basically when a write error occured, the device may be retrying the same write, obviously at the expense of tape space. So excessive write errors may result in a lower "real" capacity.

Solutions

If a certain number of files and directories need to be backed up then it is clear there is no way of enhancing the compression ratio or decreasing the amount of meta-data. The backup speed, however, is something that may possibly be fine tuned by increasing the concurrency (multiple data streams to the same device; see Advanced device options) or by changing the backup path and/or the infrastructure (network, SAN, interfaces, different media agent client, ....). Write errors, if they occur, obviously need to get resolved on the hardware side by replacing a faulty device or tape (contacting harware support may be required).

You may be interested to some of my other Support Tips too: Support Tips listed per category