Sunday, January 26, 2014

Procurve Network OOBM

Network OOBM


Connect the Network patch cable you want to use for remote management (out of band) to the port marked “Management Port” (out of band).


Start up a terminal emulator program on a laptop and connect a console cable to the port marked Console (see above)


Set up OOBM

Basic Configuration

HP Switch# config                               Move to configuration context
HP Switch(config)# oobm                         Move into the OOBM context,
HP Switch(config)# enable
HP Switch(oobm)# interface enable               Enable the interface.
HP Switch(oobm)# ip address 10.1.1.10/21        Add IP address and default gateway to
HP Switch(oobm)# ip default-gateway 10.1.1.1    the interface.
HP Switch(oobm)# end                            Exit back to configuration context.

HP Switch# ping source oobm <router ip>         Verify connectivity w/an uplink device.



ProCurve Policy/Classifier based QoS

Policy/Classifier based QoS


Create Classes of traffic


class ipv4 "VOIP”

     10 match udp 0.0.0.0 255.255.255.255
0.0.0.0 255.255.255.255 range 16000 32000

     15 match udp 0.0.0.0 255.255.255.255 range 40000 41000
                  0.0.0.0 255.255.255.255 range 40000 41000
   exit

class ipv4 "VIDEO"

     10 match udp 0.0.0.0 255.255.255.255
0.0.0.0 255.255.255.255 range 35000 39000

     15 match udp 0.0.0.0 255.255.255.255 range 42000 43000
                  0.0.0.0 255.255.255.255 range 42000 43000
   exit

class ipv4 "VOIP_SIGNALING"

     10 match tcp 0.0.0.0 255.255.255.255
0.0.0.0 255.255.255.255 range 2000 2002

     15 match ucp 0.0.0.0 255.255.255.255
0.0.0.0 255.255.255.255 range 2000 2002

     20 match tcp 0.0.0.0 255.255.255.255
0.0.0.0 255.255.255.255 range 5060 5061

     25 match udp 0.0.0.0 255.255.255.255
                  0.0.0.0 255.255.255.255 range 5060 5061
   exit

class ipv4 "DATA"

     10 match tcp 172.31.2.0 255.255.255.0 0.0.0.0 255.255.255.255 eq 23
     15 match tcp 172.31.2.0 255.255.255.0 0.0.0.0 255.255.255.255 eq 22
   exit

class ipv4 "WEB"

     10 match tcp 172.31.2.0 255.255.255.0 0.0.0.0 255.255.255.255 eq 80
     15 match ucp 172.31.2.0 255.255.255.0 0.0.0.0 255.255.255.255 eq 80
     20 match tcp 172.31.2.0 255.255.255.0 0.0.0.0 255.255.255.255 eq 8080
     25 match ucp 172.31.2.0 255.255.255.0 0.0.0.0 255.255.255.255 eq 8080
     30 match tcp 172.31.2.0 255.255.255.0 0.0.0.0 255.255.255.255 eq 443
     35 match ucp 172.31.2.0 255.255.255.0 0.0.0.0 255.255.255.255 eq 443
   exit

class ipv4 "BULK_DATA"

     10 match tcp 0.0.0.0 255.255.255.255 0.0.0.0 255.255.255.255 eq 20
     15 match tcp 0.0.0.0 255.255.255.255 0.0.0.0 255.255.255.255 eq 21
   exit

class ipv4 "REMAINING"

   exit


Create Policies for classes


policy qos "VOIP_TRAFFIC"

     10 class ipv4 "VOIP"           action dscp af33
     30 class ipv4 "VOIP_SIGNALING" action dscp cs3
     70 class ipv4 "REMAINING"      action dscp cs0
   default-class                    action dscp default
   exit

policy qos "VIDEO_TRAFFIC"

     20 class ipv4 "VIDEO”          action dscp cs3
     70 class ipv4 "REMAINING"      action dscp cs0
   default-class                    action dscp default
   exit

policy qos "VOIP_SIGNALLING_TRAFFIC"

     30 class ipv4 "VOIP_SIGNALING" action dscp cs3
     70 class ipv4 "REMAINING"      action dscp cs0
   default-class                    action dscp default
   exit

policy qos "LAN_TRAFFIC"

     40 class ipv4 "DATA"           action dscp af22
     50 class ipv4 “WEB”            action dscp cs2
     70 class ipv4 "REMAINING"      action dscp cs0
   default-class                    action dscp default
   exit

policy qos "BULK_DATA_TRAFFIC"

     60 class ipv4 "BULK_DATA"      action dscp af11
     70 class ipv4 "REMAINING"      action dscp cs0
   default-class                    action dscp default
   exit

policy qos "UPLINK_TRAFFIC"

     10 class ipv4 "VOIP"           action dscp af33
     20 class ipv4 "VIDEO”          action dscp cs3
     30 class ipv4 "VOIP_SIGNALING" action dscp cs3
     40 class ipv4 "DATA"           action dscp af22
     50 class ipv4 “WEB”            action dscp cs2
     60 class ipv4 "BULK_DATA"      action dscp af11
     70 class ipv4 "REMAINING"      action dscp cs0
   default-class                    action dscp default
   exit


Apply Policies to VLANs and Interface(s)


Apply to an interface

interface Trk1
   service-policy "UPLINK_TRAFFIC" in
   exit

Apply to vlans

vlan 10
   name "VOIP_VLAN"
   service-policy "VOIP_TRAFFIC" in
   exit

vlan 20
   name "VIDEO_VLAN"
   service-policy "VIDEO_TRAFFIC" in
   exit

vlan 30
   name "VOIP_SIGNALLING_VLAN"
   service-policy "VOIP_SIGNALLING_TRAFFIC" in
   exit

vlan 40
   name "LAN_VLAN"
   service-policy "LAN_TRAFFIC" in
   exit

vlan 50
   name "BULK_DATA_VLAN"
   service-policy "BULK_DATA_TRAFFIC" in
   exit




ProCurve GVRP

GVRP—GARP VLAN Registration Protocol—is an application of the Generic Attribute Registration Protocol—GARP. GVRP is defined in the IEEE 802.1Q standard, and GARP is defined in the IEEE 802.1D-1998 standard.

When GVRP is enabled on a switch, the VID for any static VLANs configured on the switch is advertised (using BPDUs—Bridge Protocol Data Units) out all ports, regardless of whether a port is up or assigned to any particular VLAN.  A GVRP-aware port on another device that receives the advertisements over a link can dynamically join the advertised VLAN.

When you enable GVRP on a switch, you have the per-port join-request options listed in this table:

Learn            Enables the port to become a member of any unknown VLAN for which it receives an
(default)          advertisement. Allows the port to advertise other VLANs that have at least one other
                        switch port as a vlan member.

Block            Prevents the port from joining any new dynamic VLANs for which it receives an advertisement.
                     Allows the  port to advertise other VLANs that have at least one other port as a vlan member.

Disable        Causes the port to ignore and drop all GVRP advertisements it receives and also prevents the 
                      port from sending any GVRP advertisements.

Example:

      (config)# interface 1-2
         unknown-vlans disable

      #show gvrp
        ...
        GVRP Enabled [No] : Yes
        ...
        Port Type      | Unknown VLAN Join  Leave Leaveall
        ---- --------- + ------------ ----- ----- --------
        1    100/1000T | Disable      20    300   1000
        2    100/1000T | Disable      20    300   1000
        3    100/1000T | Learn        20    300   1000
        4    100/1000T | Learn        20    300   1000
        ...

A dynamic VLAN continues to exist on a port for as long as the port continues to receive advertisements of that VLAN from another device connected to that port or until you:

1. Convert the VLAN to a static VLAN

2. Reconfigure the port to Block or Disable

3. Disable GVRP

4. Reboot the switch

The time-to-live for dynamic VLANs is 10 seconds.


These steps outline the procedure for setting up dynamic VLANs for a segment. 

1. Determine the VLAN topology you want for each segment (broadcast domain) on your network.

2. Determine the VLANs that must be static and the VLANs that can be dynamically propagated.

3. Determine the device or devices on which you must manually create static VLANs in order to propagate VLANs throughout the segment.

4. Determine security boundaries and how the individual ports in the segment will handle dynamic VLAN advertisements.

5. Enable GVRP on all devices you want to use with dynamic VLANs and configure the appropriate “Unknown VLAN” parameter (Learn, Block, or Disable) for each port.

6. Configure the static VLANs on the switch(es) where they are needed,  along with the per-VLAN parameters (Tagged, Untagged, Auto, and Forbid) on each port.

7. Dynamic VLANs will then appear automatically, according to the configuration options you have chosen.

8. Convert dynamic VLANs to static VLANs where you want dynamic VLANs to become permanent.

If a port on the switch has joined a dynamic VLAN, you can use the following command to convert that dynamic VLAN to a static VLAN:

   (config)# static-vlan < dynamic-vlan-id >

For example, to convert dynamic VLAN 333 (from the previous example) to a static VLAN:

   2920(config)# static-vlan 333

When you convert a dynamic VLAN to a static VLAN, all ports on the switch are assigned to the VLAN in Auto mode.



ProCurve VoIP, LLDP-MED, and QoS



Link Layer Discovery Protocol (LLDP) is an ieee standard protocol defined in 802.1AB.  It's used to discover directly attached devices (switches, routers, IP Phones, etc.). LLDP advertisements are encapsulated in LLDP Data Units (LLDPDU) via Type Length Values (TLV).  The standard defines two types of TLVs: Standard and Optional.


LLDP devices must support the advertisement of Standard TLVs (Chassis ID, Port ID, TTL, etc.), and may also include optional TLVs'.  These optional TLVs' are where vendor specific advertisements would be defined (model, firmware, PoE, QoS, etc.).

VLANs

Typically IP Phones have a single port that supports both a PC and the phone vlan. Generally the data VLAN is left untagged and the voice VLAN is tagged.  There are three ways that the IP telephone can learn its voice VLAN:



1) Manually set it on the phone.

2) DHCP options – The phone sends DHCP requests over the data VLAN, and the DHCP server offer will include vendor specific DHCP options including the voice VLAN.

3) Use a discovery protocol – LLDP-MED or CDPv2.  Here, the phone will discover the voice VLAN from the switch. 



On Provision switches using LLDP-MED, you must first enable lldp globally and then set the parameter “voice” in the vlan context for the switch to accept LLDP-MED TLV advertisements.   

hpe(config)# lldp run

hpe(config)# vlan 100
hpe(vlan-100) name "VOICE"
hpe(vlan-100) tagged <interfaces >
hpe(vlan-100) voice  ß Use lldp-med TLVs sent from the phone
hpe(vlan-100) exit

QoS

The two commonly used QoS marking methods are:
  • Class‐of‐Service (CoS) at Layer 2, and
  • Differentiated Service Code Point (DSCP) at Layer 3.

CoS (aka 802.1p) has 8 levels, numbered 0‐7 which map to 8 priority queues on the (egress) of each Ethernet port.  The CoS marking is set in tagged frames only since the priority (802.1p) field is part of the VLAN tag.  Untagged frames are placed in the CoS level 0 priority queue (or normal queue)

On most HPE switches CoS values 0‐7 map to priority queues like so:
 
 [1] [2] [0] [3] [4] [5] [6] [7]
 
Therefore, a CoS 0 priority will be mapped to hardware output queue Q3 and take precedence over frames marked 1 & 2.   Most HP edge switches have 8 output queues and is configurable to have 4 or 2 if desired.

hpe(config)# qos queue-config
 2-queues              Set the number of egress queues for each port.
 4-queues              Set the number of egress queues for each port.
 8-queues              Set the number of egress queues for each port.



hpe(config)# qos queue-config 4-queues
 This command will modify the current running configuration,
 execute 'write memory' to replace the startup configuration,
 and then reboot.

 Egress queues will be configured as follows:
  4-queues


Do you want to save current configuration [y/n/^C]?


The default mapping of CoS values to output queues on an HP edge switch looks like this:
   +-------------+-----------------+-----------------+-----------------+
   | Cos Value   | 8 Output Queues | 4 Output Queues | 2 Output Queues |
   +-------------+-----------------+-----------------+-----------------+
   |      1      |        1        |        1        |        1        |
   |      2      |        2        |        1        |        1        |
   |      0      |        3        |        2        |        1        |
   |      3      |        4        |        2        |        1        |
   |      4      |        5        |        3        |        2        |
   |      5      |        6        |        3        |        2        |
   |      6      |        7        |        4        |        2        |
   |      7      |        8        |        4        |        2        |
   +-------------+-----------------+-----------------+-----------------+

 
In voice deployments, generally the Ethernet switch port is configured to support 2 VLANs, one for voice and one for data.  LLDP is enabled on these Ethernet ports and can be configured to advertise voice VLAN ID and QoS information using the Network Policy LLDP TLV. 



Generally set prioritization queues for voice, video, and switch control plane traffic on edge switches as:



Priority  Traffic
7              Control plane
6              Routing
5              IP voice
4              IP video
3              Voice signaling
0              Normal Data
1              http, bulk transfer, etc.
2              http, bulk transfer, etc.



The priority can either be set on the interface or the vlan. If you set the priority at the interface level, the switch will not be able to distinguish normal data traffic from voice. 


QoS can also be used to prioritize traffic globally on TCP/UDP port based traffic.  This can be used on soft phones since the switch doesn’t know the difference between normal data traffic and soft phone voice traffic.



VLAN based prioritization applies to the 802.1p flag in a tagged (trunked) packet.  Therefore it will only prioritize tagged traffic for the vlan it is set in.  As such, the untagged vlan doesn’t have the 802.1p header and therefore will remain at normal priority (priority 0). 







Trust Model
In TRUST model, the edge switch accepts (or trusts) the phones QoS markings without modifying the priority.   Procurve switches default setting is “trust”.



If the frame is routed, the MAC header is stripped before routing the packet to its destination.  To keep the priority across routers use DSCP.  To prevent users over-riding priority settings, use strict priority provisioning with CoS and/or DSCP as well.



Non-Trusted Model
In most environments the trust model is fine.  Otherwise you can strictly provision COS priority per interface or vlan.  You can also globally prioritize traffic based on TCP/UDP port numbers.  Additionally you can create a Classifier base QoS Policy to prioritize traffic.



The following example shows the use of LLDP-MED and strict QoS provisioning to modify the default settings.  The VoIP VLAN is set through LLDP-MED and qos is set via DSCP:




Enable LLDP globally
2920(config)# lldp run



Enable Differential Code Services proritization
2920(config)# qos type-of-service diff-services



Create and configure the telephony VLAN where all of the telephony equipment will run
2920(config)# vlan 10
2920 (vlan-10)# name VOICE_SIGNALING
2920 (vlan-10)# qos dscp 011000     (802.1p pri 3)



Create and configure the voice VLAN
2920(config)# vlan 20
2920 (vlan-20)# name VOICE
2920 (vlan-20)# qos dscp 011110             (802.1p pri 5)
2920 (vlan-20)# voice                                        ßAccept lldp-med TLV's



Create and configure the VIDEO VLAN
2920(config)# vlan 30
2920 (vlan-30)# name VIDEO
2920 (vlan-30)# qos dscp 011100                           (802.1p pri 4)



Create and configure the data VLAN.
2920(config)# vlan 40
2920 (vlan-40)# name DATA

2920 (vlan-40)# qos dscp 010010                     (802.1p pri 0 - Normal)


Advanced QoS on Provision

In some cases you may want to reduce the number of hardware queues to 4 or 2. This would allow resources that are by default shared across 8 queues to be shared across 4 or 2 queues.  Example:

hp(config)# qos queue-config 4-queues

This changes the switch to only use 4 queues (needs a reboot to take affect). You use the show qos queue to verify

hp# show qos queue-config
802.1p
Queue Pri   Memory %
----- ---   --------
1     1–2   10
2     0,3   70
3     4–5   10
4     6–7   10


The amount of packet buffer memory allocated to the queues is pre-defined and not user configurable.


However, the user can change the amount of Guaranteed Minimum Bandwidth (GMB) allocated to each queue. The default GMB queue settings are:




These defaults can be viewed with the following command:

hp# show bandwidth output 1-5

 Outbound Guaranteed Minimum Bandwidth %

  Port   Q1  Q2  Q3  Q4  Q5  Q6  Q7  Q8
  ------ --- --- --- --- --- --- --- ---
  1      2   3   30  10  10  10  15  20
  2      2   3   30  10  10  10  15  20
  3      2   3   30  10  10  10  15  20
  4      2   3   30  10  10  10  15  20
  5      2   3   30  10  10  10  15  20



These GMB values can be changed on interfaces. This command is for interface 1 through to 5:



hp(config)# int 1-5 bandwidth-min output 2 3 30 10 20 15 10 10



TThis is for 8-queues, if you had the switch set to 4-queues then then only 4 GMB values is used. The percentages allowed to the queues should add up to 100%. In the above example GMB for Q5 (video) and Q6 (voice) pulls bandwidth from the default settings of Q7 (routing traffic) and Q8 (net mgmt).




hp(config)# show bandwidth output 1-5

 Outbound Guaranteed Minimum Bandwidth %

  Port   Q1  Q2  Q3  Q4  Q5  Q6  Q7  Q8
  ------ --- --- --- --- --- --- --- ---
  1      2   3   30  10  20  15  10  10
  2      2   3   30  10  20  15  10  10
  3      2   3   30  10  20  15  10  10
  4      2   3   30  10  20  15  10  10
  5      2   3   30  10  20  15  10  10



Search Duke

About the Author

My photo
Central Florida, United States