'Log I/O-Queue' too small: total <pages>, max. used <pages>

The queue for log entries may be too small. When change transactions write log entries, the log entries are first stored temporarily in a queue, before the logwriter writes them to the log devspace. This queue usually consists of a single page, but for mass statements (mass DELETES, array INSERTS, and so on), there may be more log entries to write than the system can simultaneously physically write to the disk. If the log queue overflows, it cannot accept any more log entries. This results in a large number of internal database wait situations (Vsuspend) within a short amount of time. Because transactions that write log entries hold SQL locks, other transactions will be affected.

Although the log queue has not overflowed, you should increase the XPARAM parameter LOG_IO_QUEUE (maximum: 2000) for the medium-term. Also, check whether the log devspaces can be moved to a faster disk, to accelerate the physical log I/O.