I agree with Phoneboy that there is a difference in overhead incurred by the two different VPN clients, but that yawning chasm between performance levels seems way too large to account for all of the VPN client implementation differences. A few things to check:
1) Are the Windows system and the Ubuntu system on the same VLAN with the exact same network path between them and the firewall where the RA VPN is terminating? If not you need to do this for your testing to eliminate any intervening network differences due to MTU sizes, packet latency/loss etc. which can dramatically impact performance.
2) Next step is to ensure the Ubuntu box itself is not experiencing any localized NIC issues. Run netstat -ni and make sure that ERR/OVR/DRP counters are zero or very close to zero, and especially that they not actively incrementing during your VPN performance test.
3) During a VPN performance test run top on the Ubuntu box and hit 1. While traffic is transferring are any CPUs running at 100%? I'm not sure if the Linux SNX client is multithreaded, but if a CPU is saturated that indicates a CPU bottleneck. (this situation is pretty unlikely but needs to be checked anyway)
4) Next step is prior to a VPN test, run netstat -s on the Ubuntu box and note ICMP & TCP counters reported (also try netstat -svv and netstat -see to determine if you can get verbose/enhanced protocol statistics on that version of Linux). Run your VPN test for about 5 minutes then stop it. Run netstat -s again and compare the ICMP/TCP counters. Retransmissions? Bad segments? Any suspicious ICMP counters increment significantly like Source Quench, Dest Unreachable?
5) Final step if the problem still persists is to take a packet capture on the Ubuntu box with tcpdump -w (filename) while a VPN transfer is running, then pull that capture into Wireshark for analysis. Wireshark does a great job of highlighting retransmissions, TCP zero window events, and other network conditions that can kill performance and help get you pointed in the right direction for remediation. One thing to be on the lookout for is long inter-packet delays, and which side then finally speaks up to break the logjam and get things moving again. Especially look for long delays where your VPN client is waiting for the firewall to do something, or perhaps vice-versa, and at least you will know which side is holding things up and you can focus your investigation there.
Good luck!
Gateway Performance Optimization R81.20 Course
now available at maxpowerfirewalls.com