IPsec Forward Error Correction

[HT_JDC]

Hello Experts,

At SD-WAN fortigate site-to-site VPN, we use IPsec aggregation with 2 IPsec tunnels (WAN),
which are connected to LTE. At each LAN side, SIP phone is connected.
As voice quality is not good (maybe due to instablity of LTE),
we added Forward Error Correction (FEC) for both direction (Egress and Ingress).
As the result, the quality is improved a little.

We saw the configuration at CLI and the following is seen.

set fec-egress enable
set fec-codec rs
set fec-ingress enable

Q1 We want to know the overhead of FEC in this case. How much is added?

Also, we found the other FEC configurations, as below.

SDWAN2-1 (vpn1_1) # set fec
fec-egress Enable/disable Forward Error Correction for egress IPsec traffic.
fec-send-timeout Timeout in milliseconds before sending Forward Error Correction packets (1 - 1000).
fec-base Number of base Forward Error Correction packets (1 - 20).
fec-codec Forward Error Correction encoding/decoding algorithm.
fec-redundant Number of redundant Forward Error Correction packets (1 - 5 for reed-solomon, 1 for xor).
fec-ingress Enable/disable Forward Error Correction for ingress IPsec traffic.
fec-receive-timeout Timeout in milliseconds before dropping Forward Error Correction packets (1 - 1000).
fec-health-check SD-WAN health check.
fec-mapping-profile Forward Error Correction (FEC) mapping profile.

SDWAN2-1 (vpn1_1) # set fec-base
fec-base Enter an integer value from <1> to <20> (default = <10>).

Maybe, some parameters are related to the FEC packets.

Q2 What parameters should we care in order to improve situation,
especially for voice/video application? What parameters affect the overhead?
How much is added? (Excuse me for vague questions)

Q3 When we use SD-WAN, should we enable fec-health-check for the corresponding SLA?

SDWAN2-1 (vpn1_1) # set fec-health-check
<string> please input string value
Default_DNS sdwan
Default_FortiGuard sdwan
Default_Gmail sdwan
Default_Google Search sdwan
Default_Office_365 sdwan
remotePing sdwan

Any comments would be appreciated.

Best regards,

[princes]

Dear HT_JDC,

The FEC is not measured by number of bytes, it is the ratio which measures the overhead which is dependent on the base payload.

It may result in overall Bandwidth increase by the applied ratio.

Best Regards,

[HT_JDC]

Dear Princes,

Thanks for your comments.

If the payload size is 1000byte, how much is added as FEC?

Best regards,

[princes]

Hi HT_JDC,

It is difficult to tell the exact number since it is dependent on multiple factors like the medium and the scheme chosen. 

Thank you.

[HT_JDC]

Dear Princes,

Here is my configuration, I think.

fec-egress : enable
fec-send-timeout : 5
fec-base : 10
fec-codec : rs
fec-redundant : 1
fec-ingress : enable
fec-receive-timeout : 50
fec-health-check : remotePing
fec-mapping-profile :

Thanks in advance,

[HT_JDC]

Dear Princes and Experts,

Any estimate overhead for payload 1000 byte with the following configuration would be appreciated.

fec-egress : enable
fec-send-timeout : 5
fec-base : 10
fec-codec : rs
fec-redundant : 1
fec-ingress : enable
fec-receive-timeout : 50
fec-health-check : remotePing
fec-mapping-profile :

Best regards,