Although there are a ton of SLP tuning options, there are zero relating to bandwidth. None. Zip. Nada. You can play with how many kick off for a particular SLP, how large each is, how long between checks, the # of images in each, etc. in an attempt to mitigate the problem. But you cannot control the max bandwidth used between the Masters from inside NBU.
Your best bet is to talk to your network group and have them limit the bandwidth utilization on their side if NBU hogging the entire pipe is causing problems in production. Note that this has a pretty good likelihood of making you fall behind in your replications and hence increasing your RPOs. I've seen good results in the past from showing network-side bandwidth reports showing the connection is pegged at 100% for days & the corresponding replication backlog to customers and asking if they're okay with this; generally they decide to increase the cap between sites. =)
Best of luck.
