Prepare

SUMMARY Create agents for the fabric devices, ensure device profiles exist for those devices, and define the resource pools you want to use in your deployment.

Explanation of Procedure

With Apstra, you can design and build your fabric before you install the actual fabric devices in your data center. This is because you don’t need to assign physical devices to your blueprint until you’re ready to deploy. This decoupling of the design and build from the actual deployment gives you the freedom to work within the timelines and workflows suitable for your business. For example, you can use Apstra to design the fabric in all your data center sites before you even roll out equipment.

For expediency in this use case, however, you’ll install the fabric devices before you start your fabric design. Once you install the devices, you’ll create agents for these devices. An agent is an intermediary software service that translates API calls and responses between Apstra and the device. An agent can reside on the device (where supported) or on the Apstra server. There is one agent per device.

If you run with ZTP, Apstra automatically creates the agents for you. If you don't deploy with ZTP, you’ll need to create these agents manually, which is what you’ll do in this use case.

When you create an agent for a device, the agent reaches out to the device to determine what kind of device it is. Apstra uses device profiles for this purpose. Apstra ships with a set of profiles for all devices that are known to work with Apstra. A device profile contains information about a supported device including hardware model and supported software versions. When Apstra connects to a device for the first time, it gathers information from the device and looks for a matching device profile. Apstra allows you to create your own device profile, but this is beyond the scope of this document. For the purposes of this use case, your task is to familiarize yourself with the pre-existing device profiles that you’ll use.

Next, you’ll define the resource pools that you’ll use for your fabric. These are the IP address ranges, the BGP autonomous system (AS) number ranges, the VNI ranges, and so forth. These resources are typically specific to a data center location.

Create the Device Agents

Prepare your fabric devices. Connect to the console port on each of your fabric devices. Log in as a user with super-user access, and do the following:
1. Create the login credentials that you want Apstra to use to log in to the device. If you want Apstra to use an existing set of credentials, then you can skip this step. The login credentials that Apstra uses must have super-user capabilities on the device.
2. Configure the management IP address and subnet and connect the device to the management network over the out-of-band management interface (for example, em0). In this use case, the management IP addresses of the leaf devices are 10.123.151.1/16 and 10.123.151.2/16. The management IP addresses of the spine devices are 10.123.162.1/16 and 10.123.162.2/16.
3. Enable SSH and NETCONF on the device. Apstra uses NETCONF to configure the device. If you plan on having Apstra log in to the device as the root user, then be sure to enable root login over ssh on the device.
For complete information on prerequisite device configuration including how to configure a management routing instance, see Juniper Device Agent. .

Note:
Since you haven’t designed your fabric yet, you don’t generally wire up your fabric at this time (other than to connect the devices to the management network). Once you finish your design and build later, you’ll use Apstra to generate a cabling map for your cabling installers to follow.

Create an agent for each fabric device.

An agent acts as a proxy for the fabric device, translating commands from Apstra to commands that the device understands, and translating responses from the device to responses that Apstra understands.

There is one agent per fabric device. An agent can be onbox or offbox. An onbox agent resides on the device. An offbox agent resides on the Apstra server. For Junos devices, the agents are offbox.

Note:

To see what each icon in the left-nav bar represents, click the square at the bottom of the left-nav bar.

Log in to the Apstra UI. The default username is admin, the default password is admin.
In the left-nav bar, select Devices>System Agents>Agents to bring up the Agents page.
Select the OFFBOX tab at the top of the page.
Click Create Offbox Agent(s) to bring up the Create System Agent(s) window.

Fill in the required fields as shown in Table 1 and click Create.

Table 1: Offbox Agent Parameters
Agent Parameters	Description	Settings in this Use Case
Device Addresses	The management IP addresses of the devices in your fabric. Addresses can be a comma-separated list or a range.	10.123.151.1 10.123.151.2 10.123.162.1 10.123.162.2
Operation Mode	FULL CONTROL - deploys configuration and collects telemetry TELEMETRY ONLY - collects telemetry only	FULL CONTROL
Platform	The platform that these devices belong to.	junos
Username	The username to log in to the devices.	<username>
Password	The password to log in to the devices.	<password>
Note: If you have different login credentials for all your devices, then you must create your offbox agents one at a time.

The Agents page now lists the agents that you’ve just created. The Connection State indicates whether the agent can connect to the device. The Job State shows the progression as the agent starts up, logs in to the device, and retrieves system information to allow Apstra to identify the device hardware and software version. Once Apstra identifies the device, it assigns a device profile to use for that device. When successful, the Job State shows SUCCESS and the Connection State shows CONNECTED. The page also displays the device software version and the host name (if configured).

Note:

This step succeeds even if you have not configured NETCONF on the device. If you have not configured NETCONF properly, you’ll see errors during deployment.

Acknowledge the devices.
At this point, the agents are up and running and connected to the devices they represent. However, Apstra is not actively managing the devices yet. You must acknowledge the devices to bring them under Apstra management.
1. Select Devices>Managed Devices in the left-nav bar to bring up the Managed Devices page.
  This page shows the devices in OOS-QUARANTINED state, which means that the devices are out-of-service and not yet under Apstra management.
2. Select all the devices using the checkboxes on the left. A toolbar appears above the table (Figure 1).
  Figure 1: Acknowledge Systems
3. Click the Acknowledge selected systems icon. The devices all change state to OOS-READY, which means that the devices are still out-of-service but are now ready to be managed.

View the Device Profiles

Apstra ships with a set of predefined profiles for supported devices. In most situations, you should be able to find a predefined profile for your device.

Apstra also allows you to define your own device profile in the event that a predefined profile for your device does not exist. Creating a device profile is beyond the scope of this document. For information on creating device profiles, see Creating a Device Profile .

To view the predefined device profiles used in this use case:

Select Devices>Device Profiles to bring up the Device Profiles page.
Scroll through the list to confirm that the devices deployed in this use case are in the list. You may want to use the Query pull down and enter "Juniper" in the Manufacturer field to narrow down the list:
- Juniper_QFX5110-48S for the leaf devices
- Juniper_QFX10002-36Q for the spine devices

Click a device profile to see how each device is defined. Figure 2 shows the device profile for the QFX10002-36Q switch, and Table 2 provides a high level summary of the different sections in the profile.

Figure 2: Device Profile for QFX10002-36Q

Table 2: Device Profile
Section	Description
Summary	A summary of the device.
Selector	The set of matching criteria that Apstra uses to determine whether a device matches this device profile. The criteria include both hardware and software matching criteria. As part of agent creation, Apstra reads system information from the device and looks for a matching device profile based on the matching criteria in this section.
Capabilities	A summary of the hardware and software capabilities and characteristics.

Scroll down to the Ports section. Click a port in the port pictogram to see details, including how the interfaces are named on this device.

Since the interface name and speed may change depending on the capabilities of the transceiver or interface module, Apstra allows you to define mappings or transformations that capture the different personalities a port can assume.

The QFX10002-36Q has a port panel consisting of 6 cages with 6 ports per cage. Depending on the transceivers you install, you can configure any port in any cage for 10 Gbps (breakout) or 40 Gbps, Within each cage, you can install QSFP28 transceivers in the second and sixth ports to support 100 Gbps. Apstra uses transformations to capture all this information in the device profile.

Figure 3 shows the port panel with port 2 selected. The Transformation table confirms the port supports 10 Gbs, 40 Gbps, and 100 Gbps. If you click on port 1 you'll see that only 10 Gbps and 40 Gbps are supported.

Figure 3: QFX10002-36Q Port Transformations

Define Resources

Define the resources specific to your data center site. In general, each data center site has a different resource definition.

Specify the IP pools to use for your fabric.

Select Resources>IP Pools in the left-nav bar to bring up the IP Pools page.
Click Create IP Pool to create the IP pools for your network.
The Create IP Pool window appears.

Fill in the required fields to create the IP pool for the fabric loopback interfaces (Table 3) and click Create. These are the loopback IP addresses used for establishing BGP peering in the overlay.

Table 3: IP Pool for Fabric Loopback Interfaces
Parameter	Description	Setting in this Use Case
Name	A unique name for this IP pool.	DC1-Loopback-IP
Subnets	The IP subnet to use.	10.255.0.0/24
Create Another	Select to create another IP address pool.	Checked

Fill in the required fields to create the IP pool for the fabric interfaces interconnecting the switches (Table 4) and click Create. These are the IP addresses for the et interfaces in Figure 1. These addresses are used for BGP peering in the underlay.

Table 4: IP Pool for Fabric Interfaces
Parameter	Description	Setting in this Use Case
Name	A unique name for this IP pool.	DC1-Intra-Fabric-IP
Subnets	The IP subnet to use.	10.10.0.0/24
Create Another	Select to create another IP address pool.	Checked

Specify the IP pools to use for your per-overlay network loopback addresses. In this use case, you’ll create routing zones (or VRFs) called DC1-Green and DC1-Red representing two overlay networks. In this step, you’re defining the loopback IP address pools for the two VRFs.

Fill in the required fields to create the IP pool for the DC1-Green overlay loopback interfaces (Table 5) and click Create.

Table 5: IP Pool for Overlay Network (DC1-Green)
Parameter	Description	Setting in this Use Case
Name	A unique name for this IP pool.	DC1-Green-Loopback-IP
Subnets	The IP subnet to use.	10.192.0.0/24
Create Another	Select to create another IP address pool.	Checked

Fill in the required fields to create the IP pool for the DC1-Red overlay loopback interfaces (Table 6) and click Create.

Table 6: IP Pool for Overlay Network (DC1-Red)
Parameter	Description	Setting in this Use Case
Name	A unique name for this IP pool.	DC1-Red-Loopback-IP
Subnets	The IP subnet to use.	10.193.0.0/24
Create Another	Select to create another IP address pool.	—

Specify the AS number pools to use for your fabric. Apstra assigns each switch its own AS number. The assigned AS number is used to support the BGP peering used in the underlay. For the example network you only need four AS numbers. There is no harm in specifying a pool that is larger than needed in the event you expand the fabric at a later time.

Click Resources>ASN Pools in the left-nav bar to bring up the ASN Pools page.
Click Create ASN Pool to create the ASN pool for your fabric.

The Create ASN Pool window appears.

Fill in the required fields (Table 7) and click Create.

Table 7: ASN Pool
Parameter	Description	Setting in this Use Case
Name	A unique name for this ASN pool.	DC1-ASN
Ranges	The range of ASNs to use.	65000 - 65099

Specify the virtual network identifier (VNI) pools to use for your fabric. VNIs are used in VXLAN encapsulation to provide layer 2 separation for the overlay traffic in your fabric.

Select Resources>VNI Pools in the left-nav bar to bring up the VNI Pools page.
Click Create VNI Pool to create the VNI pool for your fabric.

The Create VNI Pool window appears.

Fill in the required fields (Table 8) and click Create.

Table 8: VNI Pool
Parameter	Description	Setting in this Use Case
Name	A unique name for this VNI pool.	DC1-VNI
Ranges	The range of VNIs to use.	5000 - 9999

ON THIS PAGE

Prepare

Explanation of Procedure

Create the Device Agents

View the Device Profiles

Define Resources