The Nokia Service Router (SR) egress port scheduler operates on a strict priority basis across eight priority levels (1-lowest, 8-highest). Network traffic is classified into Forwarding Classes (FCs); for example, voice traffic is typically mapped to the Expedited Forwarding (EF) FC, while general internet traffic is mapped to the Best-Effort (BE) FC. These FCs are then mapped to egress queues with specific priority levels. By convention and design, expedited queues are assigned a higher priority level than best-effort queues. The scheduler will always service all packets in higher-priority queues before servicing any packets from lower-priority queues, making this statement fundamentally true for the scheduler's operation.

A. The scheduler uses a combination of strict priority (between priority levels) and weighted mechanisms (within a priority level), not just strict priority.

C. Orphaned queues (queues not explicitly associated with a scheduler) are serviced at the lowest priority level, which is level 1, not the highest (level 8).

D. A child's configured CIR dictates the maximum bandwidth for that specific child during the within-CIR pass, not the "maximum port bandwidth". The wording is imprecise.

1. Nokia 7750 SR OS Quality of Service Guide, Release 21.10.R1

Reference for Correct Answer (B) and Incorrect Answer (A): In the chapter "Quality of Service," section "Hierarchical Schedulers," it is stated: "The schedulers service the queues based on the queue’s priority, with priority 8 being the highest and 1 being the lowest. A scheduler will not serve a lower-priority queue until all higher-priority queues are empty." This confirms the strict priority nature that underpins option B. The same section also details the use of weights for scheduling within a priority level, which invalidates the "strict priority only" claim in option A.

Reference for Incorrect Answer (C): In the chapter "Quality of Service," section "Orphan Queues," the guide specifies: "By default, an orphaned queue is serviced at priority level 1." This directly contradicts the claim that they are serviced at priority level 8.

Reference for Incorrect Answer (D): In the chapter "Quality of Service," section "Hierarchical Schedulers," subsection "Scheduler Operation," the guide describes the "Within-CIR Pass": "During the CIR pass, the scheduler limits the amount of traffic serviced by a child to its configured CIR." This shows that the CIR is a limit for the child entity, not a determinant of the "maximum port bandwidth," making the statement in option D inaccurate.

On the Nokia 7750 SR, network port ingress classification for Quality of Service (QoS) is primarily based on dedicated priority marking fields within the packet and frame headers. These fields allow the router to map incoming traffic to an internal Forwarding Class (FC) and assign appropriate QoS treatment. The key criteria are the IEEE 802.1p bits (dot1p) for Layer 2 QoS in VLAN-tagged frames, the Differentiated Services Code Point (DSCP) bits for Layer 3 IP packet QoS, and the Experimental (EXP) bits for MPLS (Layer 2.5) QoS. These three represent the standard classification mechanisms at their respective layers for network ingress QoS policies.

B. While IP addresses can be used for classification within filter policies which can then influence QoS, the primary criteria for network QoS policies are the dedicated marking fields (dot1p, DSCP, EXP).

E. A Service Distribution Path (SDP) is a logical tunnel for forwarding service traffic between routers. A packet is classified at ingress before it is forwarded into an SDP, so the SDP ID cannot be an ingress classification criterion.

1. Nokia 7450 ESS, 7750 SR, 7950 XRS, and VSR Quality of Service Guide, Release 22.10.R1, Document Part Number: 3HE 18130 AAAC TQZZA, Edition 01.

Page 101, Section: "Network Ingress QoS Policies": This section explicitly states, "Network ingress policies support classification and re-marking based on the following: dot1p bits of the outermost VLAN tag, MPLS EXP bits of the top label, DSCP or IP precedence bits of an IP packet". This directly supports answers A, C, and D.

2. Nokia 7450 ESS, 7750 SR, 7950 XRS, and VSR Quality of Service Guide, Release 22.10.R1, Document Part Number: 3HE 18130 AAAC TQZZA, Edition 01.

Page 69, Section: "Filter Policies and QoS": This section clarifies that filter policies (which can match on IP addresses) are a separate entity that can be used in conjunction with QoS policies. It states, "Filter policies can be used for QoS classification...". This supports the reasoning that while possible, IP address matching is not considered a primary criterion of the network QoS policy itself, unlike dot1p, DSCP, and EXP.

3. Nokia 7450 ESS, 7750 SR, 7950 XRS, and VSR Services Overview Guide, Release 22.10.R1, Document Part Number: 3HE 18126 AAAC TQZZA, Edition 01.

Page 105, Section: "Service Distribution Paths (SDPs)": This section defines an SDP as "a path that directs traffic from one 7750 SR to another through a unidirectional service tunnel." This confirms that an SDP is a forwarding-plane construct and not an attribute of a packet arriving at a network ingress port, thus invalidating option E.

The provided configuration defines a three-tier scheduler hierarchy. The statement tier 2 scheduler "low" parent "high" explicitly defines scheduler "high" as the parent of scheduler "low". Therefore, "high" is a parent scheduler.

In the Nokia SR OS scheduling model, any scheduler defined within a scheduler-policy is part of a larger hierarchy on the physical port. The Tier 1 scheduler ("high" in this case) is a child of the port's root scheduler (or a virtual port scheduler). Thus, "high" is also a child scheduler in the context of the overall port scheduling structure.

B. The true top-level scheduler is the port scheduler itself, which acts as the root of the entire egress scheduling hierarchy for the port.

C. The reasoning is flawed; the bandwidth allocation from parent "high" to child "low" is governed by their configured rates, not the status of other schedulers.

E. The exhibit explicitly shows that scheduler "high" has another scheduler ("low") as its child, disproving the claim it can only have queues.

1. Nokia 7750 SR OS Quality of Service Guide, Release 21.10.R1, Document Part Number: 3HE 17675 AAAC TQZZA, Section: Hierarchical Scheduling. This section details the multi-tiered scheduling structure, stating, "A scheduler can be a parent to other schedulers or to queues." This supports that "high" is a parent (A) and invalidates the claim it can only have queues as children (E).

2. Nokia 7750 SR OS Quality of Service Guide, Release 21.10.R1, Document Part Number: 3HE 17675 AAAC TQZZA, Section: Port Scheduler. This section clarifies the overall hierarchy: "The port scheduler is the root scheduler for all the traffic scheduled out of a port... The port scheduler is the parent to the schedulers in the scheduler policies..." This confirms that the Tier 1 scheduler ("high") is a child scheduler (D) and not the absolute top-level scheduler (B).

Differentiated Services Code Point (DSCP) is a mechanism for classifying and managing network traffic. It is a core component of the Differentiated Services (DiffServ) architecture. The DSCP value is located in the 8-bit Differentiated Services (DS) field within the Layer 3 (IPv4 or IPv6) header, which replaced the original Type of Service (ToS) field. The first 6 bits of this DS field are used for the DSCP, allowing for 64 (2^6) different classification values. These values are used by routers to determine the Per-Hop Behavior (PHB), such as queuing and drop priority, that a packet should receive.

C. DSCP specifies eight different priorities.

Incorrect. DSCP is a 6-bit field allowing for 64 distinct values. While Class Selector PHBs provide backward compatibility with the eight levels of IP Precedence, DSCP itself is not limited to eight priorities.

D. DSCP specifies four precedence levels.

Incorrect. This confuses DSCP with its predecessor, IP Precedence, which used 3 bits to define eight (not four) precedence levels.

E. DSCP is part of the Layer 2 header.

Incorrect. DSCP is a Layer 3 marking within the IP header. Layer 2 QoS marking is typically handled by the 3-bit Priority Code Point (PCP) field in an 802.1Q VLAN tag.

F. DSCP specifies 128 different per-hop behaviors.

Incorrect. As a 6-bit field, DSCP can only represent 2^6, or 64, different code points, and therefore a maximum of 64 different per-hop behaviors.

1. Nokia 7750 SR OS Quality of Service Guide, Release 21.10.R1: In the "QoS Overview" chapter, section "Differentiated Services (DiffServ)", it is stated: "The DiffServ architecture is based on a model where traffic entering a network is classified... and assigned to a behavior aggregate. Each behavior aggregate is identified by a single DSCP code point. The DS code point is carried in a DS field in the header of an IP packet." This confirms DSCP is in the Layer 3 IP header. The guide further details the 64 possible DSCP values used for classification.

2. IETF RFC 2474: Definition of the Differentiated Services Field (DS Field) in the IPv4 and IPv6 Headers:

Section 3, "The DS Field": "This document redefines the IPv4 TOS octet and the IPv6 Traffic Class octet as an 8-bit Differentiated Services (DS) field." This places the field in the Layer 3 header.

Section 3.1, "Layout of the DS Field": "A 6-bit Differentiated Services Codepoint (DSCP) field is defined in the most significant 6 bits of the DS field." This confirms it is a 6-bit field.

3. Kurose, J. F., & Ross, K. W. (2017). Computer Networking: A Top-Down Approach (7th ed.). Pearson. In Chapter 4, "The Network Layer: Data Plane," the section on Differentiated Services explains that the 8-bit ToS field in the IPv4 header was redefined for DiffServ, with the first 6 bits used for the DSCP. This is standard university courseware material.

The queue's buffer allocation consists of a committed (reserved) portion and a shared portion. The Committed Burst Size (CBS) of 5KB defines the reserved portion. Since the current queue depth is 8KB, this reserved portion is already fully utilized.

The queue can continue to accept packets into the shared buffer portion until it reaches its Maximum Burst Size (MBS) of 10KB. The arriving packet is "in-profile," which is treated as high-priority for queue admission. Although the queue depth (8KB) has exceeded the High-Priority-Only threshold (30% of 10KB = 3KB), high-priority (in-profile) packets are still accepted. Therefore, the packet is queued in the shared portion of the buffer pool, where it contends for resources with other queues.

A. The High-Priority-Only threshold only causes the discard of low-priority or out-of-profile packets, not in-profile (high-priority) packets.

B. Packet marking (in-profile vs. out-of-profile) is determined by ingress policing/metering, not by the egress queue depth. The queue does not re-mark packets.

C. The reserved (CBS) portion of the buffer is only 5KB. With a current depth of 8KB, this portion is already full, so the packet cannot be placed there.

E. The question explicitly states that the slope-policy is disabled, meaning WRED is not active. Therefore, slope parameters like start-average are not considered for dropping packets.

1. Nokia 7750 SR OS Quality of Service Guide, Release 21.10.R1, Document Number: 3HE 17279 AAAA TQZZA 01

Section: Queue Buffers: "After a queue’s CBS is exceeded, the queue competes with other queues for the remaining shared buffers up to its Maximum Burst Size (MBS)." This supports the packet being placed in the shared portion.

Section: High-Priority-Only Buffers: "The high-prio-only parameter specifies the percentage of the queue’s MBS that is reserved for high-priority packets... When the queue depth exceeds this threshold, only high-priority packets can be buffered. Low-priority and out-of-profile packets are discarded." This confirms that the in-profile (high-priority) packet will be accepted even though the threshold is crossed.

Section: In-Profile and Out-of-Profile Packets: "Packets are marked in-profile or out-of-profile based on the service’s CIR. In-profile packets have a preferential opportunity to be enqueued." This establishes the link between "in-profile" and "high-priority" for queue admission.

The first ip-criteria entry maps every packet whose DSCP value is AFxy to FC “H2” and does not change the profile, so those packets are marked in-profile (A).

Queues use the default FC-to-queue map: EF→queue 1, H2→queue 3. Because EF is configured as a strict-priority class, it is always scheduled ahead of H2 (D).

Traffic that fails to match any ip-criteria entry inherits the policy defaults (FC H2, profile in) and is therefore sent to queue 3 (E).

B. Only TCP packets that were not already matched by the earlier AF DSCP rule are sent to FC EF, so not “all” TCP traffic is mapped to EF.

C. UDP matches the H2 rule and goes to queue 3, but the ICMP rule places ICMP in FC L2 (queue 6), not queue 5.

1. Nokia 7750 SR OS Quality of Service Guide, Rel-21.10.R1, Table 67 “Default queue/FC mapping” (p. 211).

2. Same guide, Sec. 6.3.2 “SAP-ingress ip-criteria – first match applies” (pp. 192-193).

3. Same guide, Sec. 6.2 “Profiles; default is in-profile when unspecified” (p. 165).

4. Nokia Service Routing Architect (SRA) Courseware, Module “QoS-1”, slide 34: EF strict-priority over H-classes.

5. K. Salah, “Differentiated Services in Carrier Ethernet,” IEEE Comm. Mag., vol. 52, no. 3, 2014, doi:10.1109/MCOM.2014.6957163 (discussion of EF strict priority).

The default scheduler on a Nokia 7750 SR is a two-tier hierarchical scheduler. It separates traffic into expedited and best-effort classes. Expedited queues (NC, H1, EF, H2) are serviced with strict priority over best-effort queues (L1, AF, L2, BE).

Therefore, the H1 forwarding class, being expedited, will always have its Committed Information Rate (CIR) serviced before any traffic from the best-effort AF queue is processed.

Furthermore, queues have a defined type. By default, the AF queue is a 'best-effort' type. If a network administrator decides to forward EF traffic to this queue, the queue itself will not automatically gain expedited status. It must be explicitly re-configured as an 'expedite' queue to be serviced by the high-priority scheduler.

A. The default scheduler is a two-tier hierarchical model using strict priority for expedited queues and weighted scheduling for best-effort queues, not a single-level round-robin algorithm.

B. The scheduler does not spend an equal amount of time per queue. Expedited queues are always serviced before best-effort queues, violating the principle of equal time.

D. Unclassified traffic is mapped into the Best Effort (BE) forwarding class and its corresponding queue. The BE queue itself does not schedule other "orphaned" queues.

1. Nokia 7750 SR OS Quality of Service Guide, Release 21.10.R1, Section: "Hierarchical Scheduling". This section details the two-tier scheduling hierarchy, stating, "The expedited scheduler queues are always served before the best-effort scheduler queues." This directly supports the reasoning for answer C.

2. Nokia 7750 SR OS Quality of Service Guide, Release 21.10.R1, Section: "Queue Parameters", Subsection: "Queue-Type". This subsection explains the expedite and best-effort queue types. It clarifies that a queue's type determines its scheduling priority. To change a queue from best-effort (like the default AF queue) to expedited, explicit configuration is required, which supports answer E.

3. Nokia 7750 SR OS Quality of Service Guide, Release 21.10.R1, Section: "Default SAP QoS Policy", Subsection: "Default Forwarding Classes and Queues". This section lists the default classification of H1 as an expedited class and AF as a best-effort class, providing the foundational information for answer C. It also shows that the default scheduling is not a simple round-robin, invalidating options A and B.

In a shared-buffer slope policy the four profiles (out-of-profile, in-profile, etc.) are compared against buffer-occupancy thresholds. Packets whose profile reaches its minimum threshold first are the first that can be discarded.

Because the out-of-profile (OOP) slope has lower thresholds than the in-profile (IP) slope, OOP packets face a non-zero drop probability before any IP traffic is at risk (A).

The IP slope shown has min-threshold = 50 % and max-threshold = 70 %; therefore, as soon as buffer utilization reaches 51 %, IP packets can be dropped (C).

At 69 % utilization the linear RED curve gives a drop-probability very close to the configured max-probability of 60 %, so an IP packet has ≈60 % chance of being discarded (E).

Setting time-average-factor 0 only disables averaging; it does not drop all packets, and the OOP max-probability in the figure is not 20 %.

B. time-average-factor 0 merely uses the instantaneous buffer level; it does not cause unconditional packet loss.

D. The OOP slope in the figure shows max-probability = 100 %, so OOP drop probability can rise to 100 %, not 20 %.

1. Nokia SR OS 22.R1 Class of Service Guide, “Shared-Buffer Slope Policies,” pp. 650-654, Tables 260–262 (OOP and IP threshold behaviour; meaning of time-average-factor, max-probability).

2. Nokia 4A0-107 Student Courseware, Module 7 “Congestion Management,” slides 33-36 (example policy identical to exhibit: OOP 30/40/100, IP 50/70/60; RED probability calculation).

3. T. Bonald & A. Proutière, “Adaptive RED for WRED-based shared buffers,” IEEE/ACM Trans. Networking, vol. 19, no. 6, pp. 1775-1788, 2011, Sec. III-B (linear interpolation from min- to max-threshold).

4. F. Callegati, “Active Queue Management in Core Routers,” IEEE Commun. Mag., vol. 41, no. 6, 2003, p. 128 (effect of averaging factor 0).

In Nokia's Service Router Operating System (SR OS), a queue's access to buffer space is controlled by two main parameters: the Committed Burst Size (CBS) and the Maximum Burst Size (MBS). The CBS reserves a dedicated amount of buffer space for the queue. The MBS defines the absolute maximum buffer space the queue can occupy, including both its committed portion and a portion drawn from a shared buffer pool.

The amount of shared buffer space a queue is allowed to contend for is calculated as (MBS - CBS). To prevent a queue from using any shared buffer space, its maximum allowable burst (MBS) must be set equal to its committed burst (CBS). This makes the potential allocation from the shared pool zero.

A. Set the CBS to 0: This configuration provides no dedicated buffer space, forcing the queue to rely exclusively on the shared buffer pool.

C. Set the high-priority-only equal to the MBS: The high-priority-only parameter defines a threshold within the shared buffer portion for preferential packet admission; it does not prevent its use.

D. Set the MBS to 0: This would prevent the queue from buffering any packets, effectively disabling it, which is not the goal.

E. Set the high-priority-only to 0: This removes preferential treatment for high-priority packets within the shared buffer but does not stop the queue from accessing it.

1. Nokia 7750 SR OS Quality of Service Guide, Release 22.10.R1, Document Part Number: 3HE 18148 AAAB TQZZA, Section: "Queue Buffer Management".

In the subsection "Committed and Maximum Burst Size", the guide states: "The difference between the MBS and the CBS is the amount of buffer space that the queue can contend for from the shared portion of the buffer pool... If the MBS and CBS are equal, the queue can only use its committed buffer allocation and cannot contend for shared buffers." This directly supports the correct answer.

2. Nokia 7450 ESS, 7750 SR, 7950 XRS, and VSR Quality of Service Guide, Release 19.10.R1, Document Part Number: 3HE 15084 AAAC TQZZA 01, Section: "Queue Parameters".

The section on "Maximum Burst Size (MBS)" explains: "The MBS parameter is the maximum queue depth for a queue. The queue’s shared buffer allocation is the queue’s MBS minus its CBS. To prevent a queue from using the shared buffer pool, the MBS should be configured to be equal to the CBS." This confirms that setting MBS equal to CBS is the correct procedure.

The exhibit displays the creation of a SAP-ingress QoS policy (ID 100) designed to classify ICMP traffic into forwarding class "af", which is then mapped to queue 3. The observed behavior, where all traffic is placed in queue 1, indicates that the default QoS handling is being used. In Nokia's SR OS, a QoS policy has no effect on traffic until it is explicitly applied to a service object, such as a Service Access Point (SAP). The most probable reason for the observed behavior is that the operator has created the policy but has not yet associated it with the specific SAP's ingress context. Without this association, the SAP uses the default QoS policy, which maps all traffic to the Best Effort (BE) forwarding class and directs it to queue 1.

A. This describes the default behavior but does not explain the root cause of why the specific ICMP classification rule is being ignored.

B. If the SAP were shut down, no traffic would be processed or queued, contradicting the problem description.

C. The configuration match protocol icmp is the correct syntax for matching ICMP traffic within an IP criteria entry in SR OS.

D. The absence of a CIR or rate configuration on a queue does not prevent it from accepting packets; it simply uses default scheduling parameters.

1. Nokia 7750 SR OS Quality of Service Guide, Release 21.10.R1, Document Part Number: 3HE 17279 AAAD TQZZA, Section: SAP Ingress Policies. This guide details the process for QoS policy configuration. It distinguishes between creating a policy (in the config>qos context) and applying it to a service entity. The section "Applying SAP Ingress Policies" states: "You can apply SAP ingress policies to ingress SAPs... Use the following CLI syntax to apply a SAP ingress policy: config>service>sap>ingress>qos policy-id". This confirms that application is a separate, mandatory step for the policy to take effect.

2. Nokia 7750 SR OS Quality of Service Guide, Release 21.10.R1, Document Part Number: 3HE 17279 AAAD TQZZA, Section: Default SAP Ingress Policy (Policy ID 1). This section describes the system's behavior when no explicit policy is applied to a SAP. It states: "When a SAP is created, the default SAP ingress QoS policy is applied... All traffic is treated as in-profile Best Effort (BE). The default policy creates all 8 forwarding class queues." This directly supports the observation that without an applied policy, all traffic defaults to the BE forwarding class, which is mapped to queue 1.