Scaling of Per-VLAN Queuing on Non-Queuing MPCs
Per-VLAN (logical interface) queueing has been introduced on most MPCs supported on the MX platform. Table 1 shows the details along with the supported JUNOS release.
MPC |
MICs Supported |
JUNOS Release |
|---|---|---|
16x10GE MPC |
N/A |
13.2 |
MPC3E |
2x10GE XFP |
13.2 |
10x10GE SFPP |
13.2 |
|
2x40G QSFPP |
13.2 |
|
1x100GE CXP |
13.2 |
|
1x100G CFP |
13.2 |
|
MPC4E-32x10GE SFPP |
N/A |
13.3 |
MPC4E-2x100GE+8x10GE SFPP |
N/A |
13.3 |
MPC6E |
24x10GE SFPP |
15.1 |
24x10GE SFFP OTN |
15.1 |
|
2x100GE CFP2 OTN |
15.1 |
|
4x100GE CXP |
15.1 |
|
MPC5E-10G100G |
N/A |
13.3R3 |
MPC5E-10G40G |
N/A |
13.3R3 |
MPC2E-3D-NG/MPC3E-3D-NG |
20x1GE SFP |
15.1 |
2xGE-XFP |
15.1 |
|
40x1GE |
15.1 |
|
4xGE-XFP |
15.1 |
|
8OC3OC12-4OC48 |
15.1 |
|
4OC3OC12-1OC48 |
15.1 |
|
8CHOC3-4CHOC12 |
15.1 |
|
4CHOC3-1CHOC12 |
15.1 |
|
8DS3-E3 |
15.1 |
|
1xOC192-XFP |
15.1 |
|
16xCHE1-T1-CE |
15.1 |
|
8OC3-2OC12-ATM-CC-CE |
15.1 |
|
4COC3-1COC12-CE |
15.1 |
|
20xGE-SFP-E |
15.1 |
|
MPC3E-3D-NG |
2x10GE XFP |
15.1 |
10x10GE SFPP |
15.1 |
|
2x40G QSFPP |
15.1 |
|
1x100GE CXP |
15.1 |
To enable logical interface scheduling, you include the per-unit-scheduler statement at the [edit interfaces interface name] hierarchy level. When per-unit schedulers
are enabled, you can define dedicated schedulers for logical interfaces
by including the scheduler-map statement at the [edit
class-of-service interfaces interface name unit logical unit number] hierarchy level. Alternatively,
you can include the scheduler-map statement at the [edit class-of-service traffic-control-profiles traffic
control profile name] hierarchy level and then include
the output-traffic-control-profile statement at the [edit class-of-service interfaces interface name unit logical unit number] hierarchy level.
Table 2 shows the number of VLANs per port available in both 8-queue and 4-queue mode for 16x10GE, MPC3E, MPC4E and MPC6E.
MPC |
MIC |
VLANs/Port – 8-Queue Mode |
VLANs/Port – 4-Queue Mode |
|---|---|---|---|
16X10GE |
No |
21 |
44 |
MPC3E |
2x10GE with XFP |
20 |
42 |
10X10GE with SFP+ |
12 per group of 5 ports* |
34 per group of 5 ports* |
|
2X40GE with QSFP+ |
20 |
42 |
|
1X100GE with CXP |
20 |
42 |
|
32x10GE MPC4E |
No |
20 per group of 4 ports* |
48 per group of 4 ports* |
2x100GE + 8x10GE MPC4E |
No |
26 |
54 |
MPC6E |
24X10GE |
20 per group of 3 ports* |
42 per group of 3 ports* |
2X100GE with CFP2 OTN |
26 |
54 |
|
4X100GE MIC with CXP |
21 |
44 |
|
*The 10X10GE MIC for the MPC3E, the 32X10GE MPC4E, and the 24X10GE MICs for the MPC6E share VLANs across a port group. You can assign all of the available VLANs to one port within the port group or spread them across the ports in any combination. |
|||
Enabling and configuring per-unit schedulers on these interfaces
adds additional output to the show interfaces interface
name [detail | extensive] command. This additional
output lists the maximum resources available and the number of configured
resources for schedulers. Following is sample output showing the CoS
scheduler resource information on a non-queuing line card:
root@R1# run show interfaces et-2/2/0 detail
Physical interface: et-2/2/0, Enabled, Physical link is Down
Interface index: 165, SNMP ifIndex: 550, Generation: 168
Link-level type: Ethernet, MTU: 1522, Speed: 100Gbps, BPDU Error: None, Loopback: Disabled, Source filtering: Disabled,
Flow control: Enabled
Device flags : Present Running Down
Interface flags: Hardware-Down SNMP-Traps Internal: 0x4000
Link flags : Scheduler
CoS queues : 8 supported, 8 maximum usable queues
Schedulers : 0
Hold-times : Up 0 ms, Down 0 ms
Current address: 80:71:1f:10:e6:b4, Hardware address: 80:71:1f:10:e6:b4
Last flapped : 2013-05-07 16:17:01 PDT (03:16:41 ago)
Statistics last cleared: Never
Traffic statistics:
Input bytes : 0 0 bps
Output bytes : 0 0 bps
Input packets: 0 0 pps
Output packets: 0 0 pps
IPv6 transit statistics:
Input bytes : 0
Output bytes : 0
Input packets: 0
Output packets: 0
Egress queues: 8 supported, 4 in use
CoS scheduler resource information:
Maximum units supported per MIC/PIC: 20
Configured units per MIC/PIC: 1
Maximum units allowed per port: 20
Configured units on this port: 1
Queue counters: Queued packets Transmitted packets Dropped packets
0 best-effort 0 0 0
1 expedited-fo 0 0 0
2 assured-forw 0 0 0
3 network-cont 0 0 0
Queue number: Mapped forwarding classes
0 best-effort
If you enable more VLANs than the previously mentioned MPC/MIC combinations support, VLANs up to the supported numbers receive dedicated queuing resources. The additional VLANs share port queues. Scheduling for port queues cannot be controlled. However, port queues are guaranteed 1 percent of the physical interface bandwidth to avoid queue starving and buffer holdup.
In the case of MPC2E-NG/3E-NG, MPC5E and MPC7E/8E/9E SKUs, the following command needs to be configured to enable “flexible queuing” on the MPC. Configuration of this knob results in a reboot of the MPC. The per-unit-scheduler, hierarchical scheduling and 2 level hierarchical scheduling are supported. There are 32K queues enabled and they can be used for either ingress queueing or egress queueing. The 32K queues are available when all 8 queues are used per IFL.
chassis {
fpc 0 {
flexible-queuing-mode; }
}
Table 3 shows the number of VLANs per port available in both 8-queue and 4-queue mode for MPC3E-NG/MPC2E-NG, and MPC5E.
MPC |
MIC |
VLANs/Port – 8-Queue Mode |
VLANs/Port – 4-Queue Mode |
|---|---|---|---|
MPC3E-NG/MPC2E-NG |
Supported MICs |
32K |
32K |
MPC5E |
Supported MICs |
32K |
32K |
The number of logical interfaces with per-vlan queuing enabled should not exceed line card maximum. If the line card maximum is exceeded, then the queuing behavior is unpredictable. This could mean that some logical interfaces have queues assigned and some do not.