I'm doing tests with Firefly Perimeter version 12.1-x46 and 12.1-x47 and experiencing very high transient latency of packets through the Firefly perimeter.
The lab setup is server 1 <-> ge-0/0/1.0 vSRX ge-0/0/2.0 <-> server 2
When pinging between the servers the latency is above 10msec. I know that the firefly will treat the ICMP packages destined for itself with lower priority so I expect high latency there, the high latency I'm experiencing is for packets traversing the firefly perimeter. The virtualistation platform is ESXi 5.1 Update 2. When I connect the two servers on the same vSwitch the ping times drops below 0.1 msec.
Does anyone else experience this problem?
P.S I get the same results for 12.1-x46 and 12.1-x47.
did you find a solution to this anomaly. We are seeing the same issue during evaluation on the product in our Virtual enviroments (multiple enviroments). We have raised a question with our Juniper account team, but haven’t had any feedback so far.
Pls use the vty command "set i386 ft_sleep disable" and have a try, hopefully you can get a big improvement.
But be aware that this vty cmd is only for demo but not persistent, it will disappeared once system reboot.
This feature is just an optimization to avoid occupy too much hypervisor CPU resource once system with light traffic.
Another workaround is just sending a continuous traffic bigger than 100pps(or bigger if no improvement), such as "ping XX -i 0.01", this will keep the flowd thread busy and handle traffic without sleep.