Thanks for all valuable comments I did as follows : 1. Set DEFAULT_SDU_SIZE=32767 in sqlnet.ora both client and server and check this is negotiated properly via net8 trace and it is . 2. Set on both linux machines: net.core.rmem_default = 1048576 net.core.rmem_max = 1048576 net.core.wmem_default = 262144 net.core.wmem_max = 1048576 3. set in sqlnet.ora on both server and client machine: RECV_BUF_SIZE=65536 SEND_BUF_SIZE=65536 and guess what ... It's not better at all :) my signle session to database is not able to do more than about 2000 kB/s through 100Mbit ethernet network . The funny thing is I can run parallel sessions from that client and saturate that bandwith but not with 1 net8 session :). There is something realy wrong here and I'm about to blame network Admins because of QoS or something so single session is limited on network layer and I can do nothing about that . (still not confirmed) . About fetch size , I can do nothing seems like there is hardcoded 136 arraysize and its in pmdtm binary (part of Informatica package) . I'm kind of desperate and when provided with root access try to do some tcpdump and check raw network related stuff maybe I find a clue there . Regards GregG -- //www.freelists.org/webpage/oracle-l