Configuring ACI In-Band Management for Hardware Flow Telemetry Export

This document provides procedures for configuring Cisco ACI in-band management for Cisco Tetration hardware sensors.

Prerequisites for Configuring Cisco ACI In-Band Management for Hardware Flow Telemetry Export

These are the prerequisites for configuring Cisco Application Centric Infrastructure (ACI) in-band management for hardware flow telemetry.

Supported Hardware and Software

For supported ACI hardware switches, ACI software version and Tetration software version refer to the Tetration platform datasheet https://www.cisco.com/c/en/us/products/collateral/data-center-analytics/tetration-analytics/datasheet-c78-737256.html

Required inb VRF Under mgmt Tenant

You must use the inb VRF under the mgmt tenant because it is hardcoded in the Hardware Agent.

  1. In your Cisco APIC system, navigate to the bridge domain page under the mgmt tenant:

    Tenants > mgmt > Networking > Bridge Domains > inb

  2. On the Bridge Domain - inb page, click the Policy tab.

    The General subtab under the Policy tab should be selected automatically.

  3. Locate the VRF field and verify that the inb VRF is selected for the mgmt tenant.

Required Use of Bridge Domain Subnet as Gateway for Hardware Agent For Spine Hardware Sensor

For the spine hardware sensor only, you must use the bridge domain subnet as the gateway for the Hardware Agent, which means that you will also need an L3Out to reach the telemetry collector. This is because the spine switch doesn?t apply the ARP for the node in-band management IP address.

The following figure shows an example of this configuration.

Spine Hardware Agent

In-Band or Out-of-Band Considerations

  • If you use both out-of-band and in-band for external connections, in-band is preferred for packets sourced from the APIC by default (for example, VMM integration).

  • Cisco APIC uses the following forwarding logic:

    • Packets that come in from an interface go out from that same interface

    • Packets sourced from a Cisco APIC that are destined to a directly connected network go out from the directly connected interface

    • Packets sourced from a Cisco APIC that are destined to a remote network prefer in-band primarily, followed by out-of-band

  • If you prefer to use out-of-band for external connections, navigate to:

    System > System Settings > APIC Connectivity Preferences

    Then select ooband in the Interface to use for external connections field.

Configure the Pod Policy

Before you can configure in-band management, you must first configure the pod policy. Configuring the pod policy consists of these tasks:

  • Configuring the BGP route reflector

  • Configuring NTP

  • Enabling HTTP on the Cisco APIC

Procedure


Step 1

Determine which pod policy group is being used by your APIC system.

  1. Navigate to the Pod Selector page:

    Fabric > Fabric Policies > Pods > Profiles > Pod Profile default > default

    The Pod Selector - default page is displayed.

  2. Locate the Fabric Policy Group field and note the name of the name of the pod policy group displayed in that field.

Step 2

Locate the BGP route reflector, Date and Time, and Management Access policies used by the pod policy group.

  1. Navigate to the Pod Policy Group page:

    Fabric > Fabric Policies > Pods > Policy Groups > pod_policy_group_name

    The Pod Policy Group page is displayed.

  2. Locate the following fields in the Pod Policy Group page and note the policies used for each field:

    • BGP Route Reflector

    • Date and Time

    • Management Access

Step 3

Configure the BGP route reflector.

The ACI fabric route reflectors use multiprotocol BGP (MP-BGP) to distribute external routes within the fabric. To enable route reflectors in the ACI fabric, the fabric administrator must select the spine switches that will be the route reflectors, and provide the autonomous system (AS) number. Once route reflectors are enabled in the ACI fabric, administrators can configure connectivity to external networks.

  1. Navigate to the BGP Route Reflector page:

    System > System Settings > BGP Route Reflector

  2. Configure the following fields for the BGP route reflector, if they are not already configured:

    • Autonomous System Number: The autonomous system number must match the leaf switch-connected router configuration if Border Gateway Protocol (BGP) is configured on the router. If you are using routes learned using static or Open Shortest Path First (OSPF), the autonomous system number value can be any valid value.

      The Autonomous System Number can be in 4-byte asplain format from 1 to 4294967295.

    • Route Reflector Nodes: Configure up to two spine nodes as route reflectors. For redundancy, configure primary and secondary route reflectors.

Step 4

Configure NTP.

  1. Navigate to the Date and Time Policy page:

    Fabric > Fabric Policies > Policies > Pod > Date and Time

  2. Verify that NTP servers have been configured under the NTP Servers field in the Date and Time Policy page.

    For more information on configuring NTP, see the “Time Synchronization and NTP” section in the Cisco APIC Basic Configuration Guide on the APIC documentation page.
Step 5

Enable HTTP.

You must have HTTP enabled because the switches download the Hardware Agent from the APIC through HTTP.

In the Cisco APIC 6.0(1) release and earlier:

  1. Navigate to the Management Access page:

    Fabric > Fabric Policies > Policies > Pod > Management Access

  2. Locate the HTTP area in the Management Access page and verify that the entry in the Admin State field is set to Enabled.

    If the field is set to Disabled, change the setting to Enabled and click Submit.

In the Cisco APIC 6.0(2) release and later:

  1. Navigate to the Management Access page:

    Fabric > Fabric Policies > Policies > Pod > Management Access > policy_name

  2. In the Work pane, choose the Policy > Web Access tab.

  3. Locate the HTTP area in the Management Access page and verify that the entry in the Admin State field is set to Enabled.

    If the field is set to Disabled, change the setting to Enabled and click Submit.


What to do next

Go to Configuring Cisco ACI In-Band Management for Hardware Flow Telemetry.

Configuring Cisco ACI In-Band Management for Hardware Flow Telemetry

The following topology is used as an example configuration for these procedures.

Topology

Before you begin

Procedure


Step 1

Configure the VLAN pool.

  1. Navigate to the Create VLAN Pool page:

    Fabric > Access Policies > Pools > VLAN, then right-click and choose Create VLAN Pool.

  2. On the Create VLAN Pool page, perform the following actions:

    1. Enter a name for the VLAN pool.

    2. (Optional) Enter a description for the VLAN pool.

    3. In the Allocation Mode field, select Static Allocation.

      This is typically used when the pool will be referenced from a static source, such as a static path binding for an EPG for use with new-deployment servers.

    4. Click Add (+) in the Encap Blocks area to add an encapsulation block.

      The encapsulation blocks define the range of VLANs in the VLAN pool.

    5. On the Create Ranges page, enter the following information:

      • Range: Enter a value in this field. For this use case, we would enter 11 in this field.

      • Allocation Mode: Choose Static Allocation.

      • Role: Choose External or On the wire encapsulations.

      Then click OK to save the values entered on the Create Ranges page.

  3. On the Create VLAN Pool page, click Submit to save the values entered on this page.

Step 2

Configure the physical domain and AEP.

  1. Navigate to the Create Physical Domain page:

    Fabric > Access Policies > Physical and External Domains > Physical Domains, then right-click and choose Create Physical Domain.

  2. On the Create Physical Domain page, perform the following actions:

    1. Enter a name for the physical domain.

    2. In the Associated Attachable Entity Profile field, select the AEP that is used for APIC connectivity.

      This could be the default AEP or some other AEP. Select the AEP that is used for APIC connectivity so that you are deploying the management EPG for the APIC in-band management interface.

    3. In the VLAN Pool field, choose the VLAN pool that you configured in the previous step.

    4. Click Submit to save the values entered on this page.

Step 3

Apply the access policy to the interface connecting to the Cisco APIC.

  1. Create a leaf access port policy group by navigating to the Create Leaf Access Port Policy Group page:

    Fabric > Access Policies > Interfaces > Leaf Interfaces > Policy Groups > Leaf Access Port, then right-click and choose Create Leaf Access Port Policy Group.

  2. Enter a name for the leaf access port policy group, then, in the Attached Entity Profile field, choose the AEP that has the domain with the VLAN pool for the in-band management, and then click Submit.

  3. Create a leaf interface policy by navigating to the Create Leaf Interface Profile page:

    Fabric > Access Policies > Interfaces > Leaf Interfaces > Profiles, then right-click and choose Create Leaf Interface Profile.

  4. Enter a name for the leaf interface profile, then click Add (+) in the Interface Selectors area.

  5. In the Create Access Port Selector page, enter the necessary information, then, in the Interface Policy Group field, choose the leaf access port policy group you created in the previous steps.

  6. Click OK to complete the configuration in the Create Access Port Selector page, then click Submit to complete the configuration in the Create Leaf Interface Profile page.

  7. Create a leaf profile by navigating to the Create Leaf Profile page:

    Fabric > Access Policies > Switches > Leaf Switches > Profiles, then right-click and choose Create Leaf Profile.

  8. On the Create Leaf Profile page, enter the necessary information:

    • In the Leaf Selectors area, select the necessary leaf switches and configure the switch selector information for those leaf switches.

    • In the Interface Selector Profiles area, choose the leaf interface profile that you created in the previous set of steps.

  9. Click Finish in the Create Leaf Profile page.

Step 4

Configure the in-band management EPG.

  1. Navigate to the Create In-Band Management EPG page:

    Tenant > mgmt > Node Management EPGs, then right-click and choose Create In-Band Management EPG.

  2. Enter the necessary information on the Create In-Band Management EPG page, specifically these fields:

    • Name: Leave default as the name for the in-band management EPG.

    • Encap: Enter the access encapsulation. For example, for this use case, you would enter vlan-11 to match the information that you entered in Step 1.b.

    • Bridge Domain: The inb bridge domain under the mgmt tenant.

      This inb bridge domain is the bridge domain that was mentioned in the Required inb VRF Under mgmt Tenant section earlier in this document. Technically, this could be a different bridge domain, as long as it is in the mgmt VRF.

  3. Click Submit.

    Click the new in-band management EPG to be displayed under the Node Management EPGs area in the left navigation tree and verify that no fault is displayed for the new EPG.

Step 5

Assign in-band management IP addresses to the leaf and spine switches.

  1. Navigate to the Create Node Management Addresses page:

    Tenant > mgmt > Node Management Addresses, then right-click and choose Create Node Management Addresses.

  2. Enter the necessary info on the Create Node Management Addresses page:

    1. In the Select Nodes By field, choose Specific.

    2. In the Nodes area, select the specific nodes for the leaf and spine switches.

    3. In the Config area, choose In-Band Addresses.

    4. In the In-Band Management EPG field, select the default in-band management EPG that you configured in the previous step.

    5. In the In-Band Gateway and In-Band IP Addresses fields, set the in-band gateway and IP address range for the switches.

  3. Click Submit.

Step 6

Assign in-band management IP addresses to the APICs.

  1. Navigate to the Create Node Management Addresses page:

    Tenant > mgmt > Node Management Addresses, then right-click and choose Create Node Management Addresses.

  2. Enter the necessary information on the Create Node Management Addresses page:

    1. In the Select Nodes By field, choose Specific.

    2. In the Nodes area, select the specific nodes for the APICs (shown as controller under the Role column).

    3. In the Config area, choose In-Band Addresses.

    4. In the In-Band Management EPG field, select the default in-band management EPG that you configured in the previous step.

    5. In the In-Band Gateway and In-Band IP Addresses fields, set the in-band gateway and IP address range for the switches.

  3. Click Submit.

Step 7

Configure the inb bridge domain subnet.

  1. Navigate to the Create Subnet page:

    Tenant > mgmt > Networking > Bridge Domains > inb, then right-click and choose Create Subnet.

  2. Enter the necessary info on the Create Subnet page:

    1. In the Gateway IP field, enter the bridge domain subnet that will be used as the gateway for the hardware agent.

    2. In the Scope area, click Advertise Externally.

  3. Click Submit.

Step 8

Verify that the configurations have been completed successfully thus far.

  1. Navigate to the Node Management Addresses page for the leaf and spine switches:

    Tenant > mgmt > Node Management Addresses > Switches.

  2. In the Nodes Within the Policy area, verify that the leaf and spine switches are listed correctly in the In-Band Management IP and In-Band Management Gateway column.

  3. Verify that the APIC and the switches can ping each other:

    
    Leaf1# show ip route vrf mgmt:inb
    <snip>
    192.51.100.0/24, ubest/mbest: 1/0, attached, direct, pervasive
        *via 10.0.200.66%overlay-1, [1/0], 00:00:40, static
    192.51.100.1/32, ubest/mbest: 1/0, attached, pervasive
        *via 192.51.100.1, vlan6, [1/0], 00:00:40, local, local
    192.51.100.24/32, ubest/mbest: 1/0, attached
        *via 192.51.100.24, vlan6, [1/0], 00:00:40, local, local
    
    Leaf1# iping -V mgmt:inb 192.51.136.21
    PING 192.51.100.21 (192.51.100.21) from 192.51.100.24: 56 data bytes
    64 bytes from 192.51.100.21: icmp_seq=0 ttl=64 time=0.461 ms
    
    
Step 9

Configure the L3Out EPG.

  1. Navigate to the L3 Outside page:

    Tenant > mgmt > Networking > L3Outs, then right-click and choose Create L3Out.

  2. Enter the necessary information in the Create L3Out wizard, specifically:

    • In the Name field, enter a name for this L3Out (for example, L3Out-mgmt).

    • In the VRF field, select inb:mgmt.

    • In the External EPG pane, configure an external EPG for the L3Out.

Step 10

Create a contract between the L3Out EPG and the in-band management EPG.

  1. Navigate to the Create Contract page for the L3Out EPG:

    Tenant > mgmt > Contracts > Standard, then right-click and choose Create Contract.

  2. Enter the necessary information in the Create Contract page, then click Submit.

  3. Navigate to the External Network Instance Profile page for the L3Out EPG:

    Tenant > mgmt > Networking > L3Outs > L3Out-mgmt > Networks > Mgmt

  4. In the External Network Instance Profile page for the L3Out EPG, choose the contract that you just created in the Provided Contracts area.

  5. Navigate to the In-Band EPG page for the in-band management EPG:

    Tenant > mgmt > Node Management EPGs > in-band_management_EPG_name

  6. Choose the contract that you just created in the Consumed Contracts area.

Step 11

Configure the inb bridge domain for the L3Out.

  1. Navigate to the Bridge Domain - inb page:

    Tenant > mgmt > Networking > Bridge Domains > inb

  2. Click the Policy tab, then the L3 Configurations subtab.

  3. Click Add (+) in the Associated L3Outs area and choose the L3Out that you configured in Step 9.

  4. Click Submit.

Step 12

Verify that the switches can ping the telemetry collector IP address.


Leaf1# iping -V mgmt:inb 10.28.121.132

PING 10.28.121.132 (10.28.121.132) from 192.100.0.26: 56 data bytes
64 bytes from 10.28.121.132: icmp_seq=0 ttl=62 time=0.407 ms
64 bytes from 10.28.121.132: icmp_seq=1 ttl=62 time=0.455 ms
64 bytes from 10.28.121.132: icmp_seq=2 ttl=62 time=0.344 ms


Spine1# iping -V mgmt:inb 10.28.121.132

PING 10.28.121.132 (10.28.121.132): 56 data bytes
64 bytes from 10.28.121.132: icmp_seq=0 ttl=61 time=0.592 ms
64 bytes from 10.28.121.132: icmp_seq=1 ttl=61 time=0.433 ms
64 bytes from 10.28.121.132: icmp_seq=2 ttl=61 time=0.411 ms

Step 13

From the telemetry collector, download the hardware agent (RPM).

  1. In the telemetry collector, click the Action icon, choose Agent Config, then click the Hardware Agent Download tab.

  2. Locate the row with the latest version of the hardware agent and click the Download button in that row.

Step 14

Upload the hardware agent onto the APIC.

  1. In the APIC GUI, navigate to:

    Admin > Firmware > Images

  2. Click the Actions button and choose Add Firmware to APIC.

  3. In the Firmware Image Location field, select Local.

  4. In the File Name field, click Browse and navigate to the location on your computer where you downloaded the hardware agent in the previous step.

  5. Select that downloaded file, then click Submit on the Add Firmware to APIC page.

Step 15

Understand the upcoming steps on enabling leaf switches for analytics.

Before going through the next few steps in these procedures, it s helpful to understand what you will be doing and why.

  • Telemetry collection is supported only on EX-model switches and later.

  • EX/FX/FX2 model switches can run in one of these modes:

    • Analytics Priority

    • Netflow Priority

    • Telemetry Priority

    Analytics Priority is the mode used for telemetry collection, so this will be the mode that you will select in the upcoming steps.

  • You will create a Node Control Policy to enable Analytics Priority for consistency.

  • You will configure Node Control Policies under the Fabric Policies.

The following figure shows how the components that you will be configuring in the upcoming steps tie in with one another.

Enabling Leaf Switches For Analytics

 

Step 16

Configure the fabric node control policy.

  1. In the APIC interface, navigate to:

    Admin > Fabric > Fabric Policies > Policies > Monitoring > Fabric Node Controls > default

  2. In the Feature Selection area, click Analytics Priority.

    Analytic priority downloads the telemetry sensor software for installation on the switches.

  3. Click Submit.

Step 17

Create an analytics policy.

  1. In the APIC interface, navigate to:

    Admin > Fabric > Fabric Policies > Policies > Analytics, then right-click and choose Create Analytics Policy.

  2. On the Create Analytics Policy page, enter the necessary information to create the analytics policy.

  3. Click Submit.

Step 18

Create a leaf and spine switch policy group.

  1. In the APIC interface, navigate to the Create Leaf Switch Policy Group page:

    Admin > Fabric > Fabric Policies > Switches > Leaf Switches > Policy Groups, then right-click and choose Create Leaf Switch Policy Group.

  2. On the Create Leaf Switch Policy Group page, enter the necessary information, specifically the following fields:

    • Analytics Policy: Select the analytics policy that you created in the previous step.

    • Node Control Policy: Select the default fabric node control policy that you configured in Step 16.

  3. Click Submit.

  4. Navigate to the Create Spiue Switch Policy Group page:

    Admin > Fabric > Fabric Policies > Switches > Spine Switches > Policy Groups, then right-click and choose Create Spine Switch Policy Group.

  5. In the Create Spine Switch Policy Group page, enter the necessary information, specifically the following fields:

    • Analytics Policy: Select the analytics policy that you created in the previous step.

    • Node Control Policy: Select the default fabric node control policy that you configured in Step 16.

  6. Click Submit.

Step 19

Create the leaf and spine switch profiles.

  1. In the APIC interface, navigate to the Create Leaf Switch Profile page:

    Admin > Fabric > Fabric Policies > Switches > Leaf Switches > Profiles, then right-click and choose Create Leaf Switch Profile.

  2. On the Create Leaf Switch Profile page, enter the necessary information, specifically the following fields:

    • Switch Associations: Select the leaf switches and associate the leaf switch policy group that you created in the previous step.

  3. Click Submit.

  4. Navigate to the Create Spine Switch Profile page:

    Admin > Fabric > Fabric Policies > Switches > Spine Switches > Profiles, then right-click and choose Create Spine Switch Profile.

  5. On the Create Spine Switch Profile page, enter the necessary information, specifically the following fields:

    • Switch Associations: Select the spine switches and associate the spine switch policy group that you created in the previous step.

  6. Click Submit.

Step 20

Verify the configurations were set correctly.

  1. Log into the APIC CLI and enter the following:

    apic1# fabric 101 show flow monitor
    ----------------------------------------------------------------
     Node 101 (Leaf1)
    ----------------------------------------------------------------
    
    Feature Prio: Analytics
    
    
  2. Log into the leaf switch and enter the following:

    Leaf1# ps -ef | grep ta_agent
    root     19200 18286  0 04:42 pts/0    00:00:00 /usr/local/bin/node /bootflash/tetration/ta_agent/ta_agent.js
    admin    33433 32405  0 04:44 pts/2    00:00:00 grep ta_agent
    
    
    Leaf1# cd /.aci/mitfs/sys/analytics/inst-analytics
    
    Leaf1# cat summary
    # Netflow Instance
    mode           : analytics
    adminSt        : enabled
    childAction    :
    ctrl           :
    dn             : sys/analytics/inst-analytics
    ipFiltAct      : deny
    lcOwn          : local
    modTs          : 2017-12-18T18:22:16.751+00:00
    monPolDn       : uni/fabric/monfab-default
    name           :
    nwIssues       :
    operErr        :
    operSt         : enabled
    operStQual     :
    pltoperStQual  :
    policyDn       : uni/fabric/analytics/cluster-<cluster_name>/cfgsrv-<analytics_policy_name>
    rn             : inst-analytics
    status         :
    
    
    Leaf1# cd controller-<cluster_name>_<analytics_policy_name>
    
    Leaf1# cat summary
    # Controller Reachability
    name               : <cluster_name>_<analytics_policy_name>
    InstallOperSt      : success
    InstallOperStQual  : installed
    childAction        :
    descr              :
    dn                 : sys/analytics/inst-analytics/controller-<cluster_name>_<analytics_policy_name>
    dscp               : CS4
    dstAddr            : 10.28.121.132
    dstPort            : 5640
    imageUri           : http://10.0.0.1:7777/fwrepo/aci-analyticsagent-dk9.default.bin
    imageUri2          : https://10.0.0.1/fwrepo/aci-analyticsagent-dk9.default.bin
    imageVer           : 2.2.1.31
    lcOwn              : local
    modTs              : 2017-12-18T18:22:18.432+00:00
    monPolDn           : uni/fabric/monfab-default
    nameAlias          :
    rn                 : controller-<cluster_name>_<analytics_policy_name>
    srcAddr            : 192.5100.100.24/24
    srcIf              : unspecified
    status             :
    uid                : 0
    vrfName            : mgmt:inb
    
    
Step 21

Verify the configuration.

  1. In the telemetry collector, click the Action button, choose Agent Config, then click the Hardware Agent Conig tab.

    The leaf and spine switches should be displayed in this screen.