The servers themselves could/can handle the process load, the nic or the switches can't, they appear to bottlenecking and dropping packets. Those "lag" spikes are dropped packets.
So really the cause of this lag/freezing is the dropping of tcp packets. They use tcp not udp which is another nerdy topic to get into.
Thanks for the laugh... now back to reality...
The servers themselves could/can handle the process load, the nic or the switches can't, they appear to bottlenecking and dropping packets. Those "lag" spikes are dropped packets.
So really the cause of this lag/freezing is the dropping of tcp packets. They use tcp not udp which is another nerdy topic to get into.
Thanks for the laugh... now back to reality...
The servers themselves could/can handle the process load, the nic or the switches can't, they appear to bottlenecking and dropping packets. Those "lag" spikes are dropped packets.
So really the cause of this lag/freezing is the dropping of tcp packets. They use tcp not udp which is another nerdy topic to get into.
Thanks for the laugh... now back to reality...
FYI...
Look at how many retrasmission you get inside PVP compared to the rest of game. Then go into a group instance and see those drop and see the RTO time go down.
The servers themselves could/can handle the process load, the nic or the switches can't, they appear to bottlenecking and dropping packets. Those "lag" spikes are dropped packets.
So really the cause of this lag/freezing is the dropping of tcp packets. They use tcp not udp which is another nerdy topic to get into.
Thanks for the laugh... now back to reality...
FYI...
Look at how many retrasmission you get inside PVP compared to the rest of game. Then go into a group instance and see those drop and see the RTO time go down.
Well just to humor you...
Info about 79.141.174.3
First/Last Seen Tue Feb 2 19:56:15 2016 - Tue Feb 2 21:29:33 2016
Total Data Sent 222.2 MBytes/167,219 Pkts/0 Retran. Pkts [0%]
Total Data Rcvd 6.5 MBytes/113,914 Pkts/0 Retran. Pkts [0%]
Current Avg Peak Current Avg Peak
342.8 Kbit/s 342.5 Kbit/s 3.9 Mbit/s 51.4 Pkt/s 49.9 Pkt/s 597.3 Pkt/s
In super laggy Azura EU..
Since it's server lag, as in the server not able to keep up with processing game objects.. the network doesn't matter at all. It would be just as laggy with UDP or with a dedicated 100Gbit network connection directly attached to the server..
The servers themselves could/can handle the process load, the nic or the switches can't, they appear to bottlenecking and dropping packets. Those "lag" spikes are dropped packets.
So really the cause of this lag/freezing is the dropping of tcp packets. They use tcp not udp which is another nerdy topic to get into.
Thanks for the laugh... now back to reality...
FYI...
Look at how many retrasmission you get inside PVP compared to the rest of game. Then go into a group instance and see those drop and see the RTO time go down.
Well just to humor you...
Info about 79.141.174.3
First/Last Seen Tue Feb 2 19:56:15 2016 - Tue Feb 2 21:29:33 2016
Total Data Sent 222.2 MBytes/167,219 Pkts/0 Retran. Pkts [0%]
Total Data Rcvd 6.5 MBytes/113,914 Pkts/0 Retran. Pkts [0%]
Current Avg Peak Current Avg Peak
342.8 Kbit/s 342.5 Kbit/s 3.9 Mbit/s 51.4 Pkt/s 49.9 Pkt/s 597.3 Pkt/s
In super laggy Azura EU..
Since it's server lag, as in the server not able to keep up with processing game objects.. the network doesn't matter at all. It would be just as laggy with UDP or with a dedicated 100Gbit network connection directly attached to the server..
He saw the ball Zerg coming.
Theory is they moved a lot of the client calculations to the server side to prevent botting in the past (crazy coldharbor bot trains). The servers themselves could/can handle the process load, the nic or the switches can't, they appear to bottlenecking and dropping packets. Those "lag" spikes are dropped packets.
So really the cause of this lag/freezing is the dropping of tcp packets. They use tcp not udp which is another nerdy topic to get into.
The cause of these dropped packets though, oh man that is a huge mess to dig through and I feel bad for that group of IT guys. Because normally you want to point to the straw that broke the camels back but in practice it won't fix it. It's a slew of things that snowballed.
A Zerg of players though is just not the cause. Depending on setup/gear/colors/abilities/cp so on and so on a group of 8 players can create as many calculations as a group of 40 players and flood the NIC/Switches/firewall/router.
Until they change client and server relationship, this won't go away. Client requires too many check ins.
WalkingLegacy wrote: »The servers themselves could/can handle the process load, the nic or the switches can't, they appear to bottlenecking and dropping packets. Those "lag" spikes are dropped packets.
So really the cause of this lag/freezing is the dropping of tcp packets. They use tcp not udp which is another nerdy topic to get into.
Thanks for the laugh... now back to reality...
FYI...
Look at how many retrasmission you get inside PVP compared to the rest of game. Then go into a group instance and see those drop and see the RTO time go down.
Well just to humor you...
Info about 79.141.174.3
First/Last Seen Tue Feb 2 19:56:15 2016 - Tue Feb 2 21:29:33 2016
Total Data Sent 222.2 MBytes/167,219 Pkts/0 Retran. Pkts [0%]
Total Data Rcvd 6.5 MBytes/113,914 Pkts/0 Retran. Pkts [0%]
Current Avg Peak Current Avg Peak
342.8 Kbit/s 342.5 Kbit/s 3.9 Mbit/s 51.4 Pkt/s 49.9 Pkt/s 597.3 Pkt/s
In super laggy Azura EU..
Since it's server lag, as in the server not able to keep up with processing game objects.. the network doesn't matter at all. It would be just as laggy with UDP or with a dedicated 100Gbit network connection directly attached to the server..
UDP might clear a lot of it up actually.