Hello everyone,
I have a scenario where a Spark 2350 running R82.00.10 (998002203) is managed by an SMS R82 JHF Take 107.
I need to implement a QoS policy and I have some questions regarding the behavior of the different QoS policy types. I have been testing both the Recommended Policy and the Express Policy.
With the Express Policy, everything works as expected regarding DiffServ marking. The packets are correctly marked, which is a requirement in my environment. However, I cannot find a way to implement a true Low Latency Queue (LLQ) or strict priority queue. Since I have SCADA/IEC-104 traffic, having low-latency prioritization is an important requirement.
With the Recommended Policy, I have:
Custom QoS classes created.
Rules matching the appropriate traffic and assigning it to the correct class.
Guaranteed bandwidth configured per class.
The classes attached directly to the WAN interface.
Inbound and outbound bandwidth correctly configured on the interface.
The logs show that the traffic is matching the expected QoS rule and class. For example, I can see entries such as:
Best_Effort -> af31 -> Rule_af31
which indicates that the traffic is being classified correctly.
However, the packets are not being marked with the expected DSCP value. When I capture the traffic with tcpdump, the packets still leave with:
tos 0x0 (Best Effort).
My questions are:
Is DSCP/DiffServ remarking supported only with Express Policy on Spark appliances?
Is it possible to configure LLQ or strict-priority queuing for SCADA/IEC-104 traffic on a Spark 2350?
In Recommended Policy, should the DiffServ code configured in the QoS class automatically remark the packet DSCP field, or is additional configuration required?
Is there any known limitation of SMB/Spark QoS compared to Gaia gateways regarding DiffServ marking and LLQ functionality?
Any guidance or experience with similar SCADA deployments would be greatly appreciated.
Thank you.