onbar backup with strange writes

[schu]

We are using 7.31.UC6 with Legato Networker 6.x and experiencing slow backups suddenly. Before (mid Feb) we were always backing up the databse in about 1hr, then suddenly it when to 4-5hrs and has been like that ever since.

Some investigation we found that the backups were producing a lot of writes from page flushers.
We performed the following before the onbar
onmode -l
onmode -c
onstat -z

Here is the output of onstat -u midway through the backup
Userthreads
address flags sessid user tty wait tout locks nreads nwrites
1871c018 ---P--D 1 informix - 0 0 0 94 58
1871c504 ---P--F 0 informix - 0 0 0 0 7349
1871c9f0 ---P--F 0 informix - 0 0 0 0 138662
1871cedc ---P--F 0 informix - 0 0 0 0 2
1871d3c8 ---P--F 0 informix - 0 0 0 0 3
1871d8b4 ---P--F 0 informix - 0 0 0 0 0
1871dda0 ---P--F 0 informix - 0 0 0 0 27553
1871e28c ---P--F 0 informix - 0 0 0 0 132191
1871e778 ---P--F 0 informix - 0 0 0 0 158751
1871ec64 ---P--F 0 informix - 0 0 0 0 130929
1871f150 ---P--F 0 informix - 0 0 0 0 5705
1871f63c ---P--F 0 informix - 0 0 0 0 166551
1871fb28 ---P--F 0 informix - 0 0 0 0 161080
18720014 ---P--- 19 informix - 0 0 0 0 0
18720500 ---P--B 20 informix - 0 0 0 0 0
187209ec --AP--M 25 informix 47 0 0 0 2057 0
18720ed8 ----X-- 25 informix 47 0 0 0 2415747 0
187213c4 ---P--D 23 informix - 0 0 0 0 0
187218b0 ------- 25 informix 47 0 0 0 0 0
19 active, 128 total, 29 maximum concurrent

There are no other users, just informix and the backup thread at tty:47

Anyone have similar problems or know what is happening?