UTM ownCloud very slow download

Question

I&rsquo;ve a UTM virtual appliance running at home for almost 5 years. Never got a problem. Just a few weeks ago I installed ownCloud so I can share files on my smartphone, etc. ownCloud syncs the files. My internet connection is 40 down and 4 up. The Problem When I use the ownCloud client outside my LAN, example at work, I got really really slow download speed of 15 &ndash; 30 KB/s. This problem also excists with any webbrowser. When I upload from my work to home I got my full upload speed of my provider, 4Mb. When ownCloud starts syncing I see that it starts on the maxium download speed and then quickly drops to almost nothing. On my LAN I got no problems, there is it the maxium up and down. So ownCloud must be working good. Because of that I think it&rsquo;s something in the UTM. 
 What I did so far I made 2 rules DNAT and SNAT. SNAT : For traffic from &ldquo; IP ownCloud server &rdquo; --> service &ldquo; 443 &rdquo; --> going to &ldquo; Any &rdquo; --> service &ldquo; MyPortNumber &rdquo; DNAT : For traffic from &ldquo; Any &rdquo; --> service &ldquo; MyPortNumber &rdquo; --> going to &ldquo; ExternalWAN-IP &rdquo; --> destination &ldquo; ownCloud IP &rdquo; &agrave; service &ldquo; 443 &rdquo; 
 I also tested true a VPN connection where I connect the ownCloud client with the IP of my ownCloud server and the download speed is there also bad. But the upload is the maxium. Also when I download a file from a share on one of my servers I got the full download speed. I see also no performance issues on my server or on all the clients that are running on it. I tried to turn off logging of the rules, turn off Intrusion Prevetion. 
 Anybody a clue on whats going on?

RW1 · Accepted Answer

I've found the solution [<:o)][<:o)][<:o)] LSO (Large Segment Offload) 
 Its explained on the website below 
 http://blog.cyberexplorer.me/2013/03/improving-vm-to-vm-network-throughput.html 
 "To learn more about LSO (Large Segment Offload) read http://en.wikipedia.org/wiki/Large_segment_offload . Essentially, the kernel is sending large packets (16K in the capture) and the NIC (or virtual NIC) is supposed to fragment the packet and transmit valid-size packets. On a real hardware NIC, at high speeds, this saves considerable amounts of CPU. in a virtualized environment I don't see the benefit. And it seems to be badly broken." 
 On the Ubuntu servers I used this command 
 Ethtool -K eth0 tso off (after reboot its gone, so you need to make a change in your config) 
 After this change I got a good result with iperf.