RemoteApp 2012 - Optimize connectivity

dear gents,

here setup:
session-based deployment
rdcb01.domain.local - connection broker
rdgw01.domain.local - rdweb + rdgw roles located in dmz.
dc01.domain.local - ts license server
file01.domain.local - central home drive , profile (folder redirection) store.

2 remoteapp collections:
rdsh01-04-.domain.local (rd_farm_01.domain.local)
rdsh10-16.domain.local (rd_farm_02.domain.local)

i`m using tls certificate remote.domain.com public pki for:
rdcb - publishing
rdweb
rdgw

, self-signed tls certificate *.domain.local (customer has no internal pki) for:
rdcb - sso
rdsh - remote desktop connections

there client-side gpos in place, per documentation, allow credential delegation rdcb , rdgw servers , self-signed certificate enrollment.
on rdgw i`m allowing http on default port, limit network firewall requirements.
full folder redirection in place network storage on file01.domain.local

questions:

#1 active remoteapp intermittently disconnects , reconnects (x minutes of inactivity), causing discomfort users. there no active session timeouts , have gpo setting keep-alive set 10 minutes.

#2 thing puzzles me long (~15 seconds) wait till can see rdsh server initiates user logon process. can`t find errors on either rdcb or rdgw eventlog, of being wrong. can see successful handshakes. reason why wondering if *ok*, `cause colleague has built analog , takes around 5 seconds start load. in both cases talking of decent infrastructure, connectivity on national scale , access via web-interface.

#3 need in order allow udp support on rdgw network-wise?
udp_3391 from:internet_any to: remote.domain.com
udp_3391 from:remote.domain.com to: rdgw01.domain.local
udp_3391 from: rdgw01.domain.local to: ???

found fault - printer redirection. it`s 30 seconds max, mapped drives across globe.

Windows Server  >  Remote Desktop Services (Terminal Services)