In connection with Exchange Back Pressure you can read that the number of version buckets also has an influence on back pressure. But honestly, who knows what version buckets are? Here is a short explanation:
To understand what version buckets are, you need to take a look at how Exchange works. An incoming message is not written directly to the queue database, but first ends up in the working memory of the server, then the message is written to the transaction log of the queue database, from there it then goes to the queue database. The outstanding messages in the working memory that have not yet been written to the transaction logs are referred to as version buckets.
Why do the version buckets affect Back Pressure? Quite simply, the version buckets are only available in the server's working memory. If the server crashes (e.g. due to overload), data is lost as these messages have not been written to the transaction logs or the queue database on the hard disk.
What are batch points? Batches come before version buckets. Incoming messages with large attachments are broken down into several batches and then processed into version buckets before being written to the transaction logs. The number of batches in the working memory is measured in batch points.
The value of batch points and version buckets are therefore important indicators for determining the load at Exchange Transport level. A SPAM wave can quickly cause the number of version buckets or batch points to skyrocket. At what point Back Pressure intervenes can be determined in this Contribution can be read here.
