Wired Intelligent Edge

Is anyone able to explain why LACP PDUs Tx/Rx values reported on different LACP trunks (Switch-to-Server versus Switch-to-Switch) can be so dissimilar (especially considering the received LACP PDU from remote peer...in this case a Server)?

Is below difference reasonable or not (the LACP is working normally AFAIK)? 

Switch-to-Server LACP Trunk case (fast):

VSF# show lacp counters ethernet 1/B15,2/B15

LACP Port Counters.

               LACP      LACP      Marker   Marker   Marker   Marker
  Port  Trunk  PDUs Tx   PDUs Rx   Req. Tx  Req. Rx  Resp. Tx Resp. Rx Error
  ----- ------ --------- --------- -------- -------- -------- -------- --------
  1/B15 Trk110 3554552   118098    0        0        0        0        0
  2/B15 Trk110 3554439   118099    0        0        0        0        0

Note: the LACP Trunk on remote Server is done on IBM Power8 VIOS.

Switch-to-Switch LACP Trunk case (fast):

VSF# show lacp counters ethernet 1/C1,2/C1

LACP Port Counters.

               LACP      LACP      Marker   Marker   Marker   Marker
  Port  Trunk  PDUs Tx   PDUs Rx   Req. Tx  Req. Rx  Resp. Tx Resp. Rx Error
  ----- ------ --------- --------- -------- -------- -------- -------- --------
  1/C1  Trk1   174902    174899    0        0        0        0        0
  2/C1  Trk1   174898    174897    0        0        0        0        0

Shouldn't be that - in an ideal situation - LACP PDUs Tx and Rx counters should increment in unison (like the case of Switch-to-Switch LACP)?