I’ve a two MK router connected with a radio bridge with PTP550E.
If I check bandwidth from the two router is very high. PTP550E Great… (4.6.0.1)
Next if I make a eoip tunnell between two router and check the bandwidth test inside the tunnell… now we have very poor troughput…
If I chenge PTP550E with two netmetal eoip work great!.
I check all the MTU and I’ve try to increase to 1700 in router and inside 550.
router is 4011 on each side, 10km link distance… TCP test is 230mbps between two routers… but 60Mbps only inside the eoip.
same condition but with F300 CSM work better… we have 10% of bandwidth difference from L2 to eoip…
We have checked in the lab and can confirm EoIP streams are not accelerated by the hardware accelerator on PTP550.
The internal ticket has been opened.
I would suggest avoiding EoIP configuration over PTP550 for now if you’d like to reach maximal throughput.
This seems like an issue with EoIP creating a situation where MAC addresses on both ends for the traffic are the same and therefore if 802.3ad bonding is used a classic problem where only one of the channels is used. Bonding has several problems. They include running at only N times the slowest overall bonded channel where N equals the numbers of bonded channels. In a situation with say 2 separate 40mhz channels you could have 4 bonded channels. Interference in any polarity would drop overall best performance to 4 x the worst 40mhz channel be it vertical or horizontal of +/- 45. It seems that this is the cause. A way to potentially handle this would be to tag each bonded channel with a VLAN then have a device intelligent enough to isolate the channels outside the 550E then put them back together maybe not using the 550E bonding if there are issues such as not enough CPU processing in the 550 series to handle the bonding method that may work best for any given situation. When cheap over the shelf WiFi hack job radios are what you use you have to be creative because there is only so much processing power that can be practically achieved and depending on goals such as EoIP on both ends of all the traffic the bonding of channels may need to take place outside the radios. Those are just ideas. Maybe with a firmware update this can be addressed outside of ideas I just brought up.
Thank you, but poor performance is the same in a 550E non bonding channell…
we waiting a firmware correction… (like Ceragon or SIAE Micr. or NEC Link, where with bonding radio all works fine!!)
we still waiting…
If Cambium can’t solve the problem, or it’s a chipset problem please consider to infomr your clients and/or product replace.
I’m waiting a fix from ONE Year, no reply from Cambium… no patch date…