In Apstra 5.1, Intent-Based Analytics (IBA) is built on Apstra’s graph-based source of truth , where devices, interfaces, links, routing constructs, and services are represented as nodes with relationships. An IBA probe is effectively a processing pipeline (a directed acyclic graph of stages and processors) that ingests telemetry and then performs calculations, aggregations, and anomaly detection. To make any of that work, the probe must first determine which specific objects in the graph —for example, which leaf switches, which uplinks, which BGP sessions, or which interface counters—should be included in the analysis.

The probe element that selects those objects is the graph query. A graph query is evaluated against Apstra’s graph database to return a set of matching nodes/relationships; those query results then become the scope for ingestion and subsequent processing. In other words, the graph query defines “apply this probe to these devices/interfaces/sessions,” and it also provides the context used to bind telemetry identities (key-value pairs describing the metric source) to the correct logical objects in the blueprint. This is why Apstra documentation describes early probe processors producing outputs whose cardinality aligns with the number of results returned by the specified graph query(s). Without a graph query, the probe would not have a deterministic, intent-aligned target set for analytics, and the same probe definition could not be reliably reused across fabrics or blueprints.

In an EVPN-VXLAN fabric where all hosts are single-homed (each endpoint is attached to only one leaf/VTEP), the EVPN control plane still needs to advertise endpoint reachability and enable BUM handling across the overlay. Two EVPN route types are fundamental in this case: Type 2 and Type 3 .

EVPN Route Type 2 (MAC/IP Advertisement) is used to advertise learned MAC addresses and, optionally, associated IP addresses for endpoints connected to the local leaf. This enables remote VTEPs to learn where a given host resides (which VTEP to send unicast traffic to) without relying on data-plane flooding for MAC learning. In Junos v24.4 EVPN-VXLAN deployments, Type 2 routes are the core mechanism for distributing endpoint reachability (MAC and MAC+IP bindings) within the EVPN domain.

EVPN Route Type 3 (Inclusive Multicast Ethernet Tag / IMET) is used to establish the flooding scope for BUM traffic in EVPN-VXLAN. In VXLAN fabrics that use ingress replication (common in data centers), Type 3 routes help build the list of remote VTEPs that should receive replicated BUM traffic for a given segment.

By contrast, Type 4 (Ethernet Segment) routes are associated with EVPN multihoming (ESI-based) and DF election; with only single-homed hosts, Type 4 is not required. Type 7 is not part of the baseline single-homed EVPN-VXLAN host advertisement set in this context.

Verified Juniper sources (URLs):

https://www.juniper.net/documentation/us/en/software/junos/evpn/topics/concept/evpn-bgp-multihoming-overview.html

https://www.juniper.net/documentation/us/en/software/junos/evpn/topics/topic-map/assisted-replication-evpn.html

According to the Juniper documentation 1 , an interface map is a configuration template that maps interfaces between logical devices and physical hardware devices (represented with device profiles) while adhering to vendor specifications. An interface map specifies a connection between the interfaces of two devices, such as a leaf and a spine, a leaf and a server, or a leaf and an external gateway. An interface map can also specify port transformations, such as breaking out a 40 GbE port into four 10 GbE ports, or disabling unused ports. An interface map can be used to achieve the intended network configuration rendering and to enable features such as LAG, ESI-LAG, or MLAG. Therefore, the correct answer is B. An interface map specifies a connection between the interfaces of two devices.  References:  Interface Maps (Datacenter Design)

In Apstra 5.1, property sets are structured data objects (YAML/JSON) used to hold values that templates can consume at render time. Their most common use is with configlets , where property set key/value pairs are referenced as variables inside the template so Apstra can perform variable substitution during configuration generation. This directly supports statement B .

Property sets are also designed to be reusable . A single configlet can reference more than one property set (for example, one set for NTP servers and another for syslog collectors), allowing clean separation of data domains and easier lifecycle updates. This supports statement E .

Operationally, when you bring design content into a blueprint, the blueprint must have the required supporting objects available. In Apstra workflows, configlets that use property sets require those property sets to be present in the blueprint context (commonly accomplished by importing the relevant property set(s) from the catalog into the blueprint as part of bringing in the configlet and its dependencies). This aligns with statement A as the blueprint-level outcome: the property sets used by an imported configlet are imported/available in the blueprint for rendering.

Statements C and D are incorrect because property sets are not limited only to configlets (they are also used with Analytics probes), and the syntax is not vendor-specific—Apstra uses standard YAML/JSON structures independent of NOS.

Verified Juniper sources (URLs):

https://www.juniper.net/documentation/us/en/software/apstra5.1/apstra-user-guide/topics/concept/property-set-datacenter-design.html

https://www.juniper.net/documentation/us/en/software/apstra4.2/apstra-user-guide/topics/concept/property-set-datacenter-design.html

https://www.juniper.net/documentation/us/en/software/apstra5.1/apstra-user-guide/topics/ref/configlet-examples.html

In Apstra 5.1, capacity planning based on pre-built time-series telemetry (such as link utilization trends) is part of Intent-Based Analytics (IBA) . IBA is where Apstra ingests streaming telemetry from fabric devices, stores it as time-series data, and presents it through built-in analytics views (dashboards/widgets) and probes. Because the question specifically calls out “pre-built time series information regarding link utilization,” the correct UI location is the Analytics top-level tab within the blueprint.

The Active tab is primarily oriented to operational state and day-2 workflows (for example, viewing live state, queries, and device-level operational views). The Staged tab is where you modify intent (physical/virtual design, policies, catalog items) prior to committing and deploying. The Dashboard provides a high-level blueprint overview and navigation, but the drill-down and time-series analytics views that support trending and capacity analysis are accessed via Analytics .

In an EVPN-VXLAN fabric using Junos v24.4, link utilization time-series is particularly valuable because underlay congestion can degrade overlay performance (BGP convergence behavior, ECMP distribution effectiveness, and endpoint experience). Apstra’s Analytics tab centralizes these metrics so operators can evaluate utilization baselines, identify sustained hot links, and support proactive actions (rebalancing, adding capacity, or adjusting design intent) without relying on ad-hoc per-device CLI polling.

Verified Juniper sources (URLs):

https://www.juniper.net/documentation/us/en/software/apstra5.1/apstra-custom-telemetry-collection-guide/topics/concept/apstra-telemetry-and-intent-based-analytics.html

In Apstra 5.1, a logical device defines the abstract port layout and capabilities (including supported speeds), while an interface map binds that abstract port layout to the real, vendor-specific front-panel ports. Rack types then consume logical devices and interface maps to model the rack’s leaf/superspine roles. Once a logical device is referenced by interface maps and used inside rack types (and potentially templates that instantiate those rack types), Apstra treats the combination as a consistent contract: port counts, roles, and speeds must remain semantically valid for every object that depends on it.

The exhibit’s validation error indicates that after changing interface speeds on the logical device, the existing interface map(s) and their usage in rack types no longer match the logical device definition (for example, the map expects certain ports/speeds/roles, but the updated logical device would leave the map invalid). Because the logical device is being consumed in multiple places, the safest and required way to remove the error is to remove all dependencies—templates (if they reference the rack types), rack types, and interface maps—so Apstra can accept the new logical device definition without violating existing mappings. After updating the logical device, you then recreate or update the interface maps and re-associate them with the rack types/templates so the entire chain remains consistent under the new speed requirements.

A Juniper Apstra rack is a physical entity that contains one or more network devices, such as leaf nodes, access switches, or generic systems. A rack is used to organize and manage the network devices in the Apstra software application. A rack has the following characteristics:

It stores information on how leaf nodes connect to generic devices. This is because a rack can include generic systems, which are devices that are not managed by Juniper Apstra, but are connected to the network. A generic system can be a server, a firewall, a load balancer, or any other device that has a network interface.  A rack stores the information on how the leaf nodes, which are the devices that provide access to the end hosts, connect to the generic devices, such as the port number, the link speed, the LAG mode, and the roles 1 .

It has a rack type, which defines the type and number of leaf devices, access switches, and/or generic systems that are used in the rack. A rack type is a resource that is created in the data center design phase, and it does not specify the vendor or the model of the devices.  A rack type can be predefined or custom-made, and it can be used to create multiple racks with the same structure and configuration 2 .

It has a rack build, which assigns the specific vendor and model of the devices to the rack. A rack build is created in the staged phase, and it uses the rack type as a template.  A rack build can also assign the resources, such as the IP addresses, the ASNs, and the VNIs, to the devices in the rack 3 .

It has a rack deployment, which applies the network configuration and services to the devices in the rack. A rack deployment is performed in the active phase, and it uses the rack build as a reference.  A rack deployment can also monitor the network performance and compliance of the devices in the rack 4 .

The following three statements are incorrect in this scenario:

It stores information on how pods connect to super spines. This is not true, because a rack does not store any information on the pod or the super spine level of the network. A pod is a cluster of leaf and spine devices that form a 3-stage Clos topology, and a super spine is a device that connects multiple pods in a 5-stage Clos topology.  A rack only stores information on the leaf and the access level of the network 1 .

It stores IP address and ASN pool information. This is not true, because a rack does not store any information on the IP address and ASN pools. IP address and ASN pools are resources that are created in the data center design phase, and they contain a range of IP addresses and ASNs that can be assigned to the devices and the virtual networks.  A rack only uses the IP address and ASN pools to assign the resources to the devices in the rack build 2 .

It stores device port data rates and vendor information. This is not true, because a rack does not store any information on the device port data rates and vendor information. The device port data rates and vendor information are specified in the rack build, which assigns the specific vendor and model of the devices to the rack.  A rack only uses the rack build to apply the network configuration and services to the devices in the rack deployment 3 .

In Apstra 5.1, blueprint changes follow an intent workflow: you edit intent in Staged , then review the delta in Uncommitted , and finally Commit to activate those changes and create a new revision. If you have staged changes that are visible under Uncommitted but decide not to proceed, the Revert action is used to discard them. Selecting Revert clears the blueprint’s uncommitted intent delta and returns the blueprint to the last committed state (the currently active intended design baseline). In practical terms, it removes all pending edits that were made since the last commit—whether those edits were physical (links/topology), virtual (routing zones, virtual networks), policies (security policies), or catalog-driven operations—so that none of those changes will be deployed.

Revert is not a “single-step undo” limited to only the most recent change; it is a discard of the staged/uncommitted change set. It also does not roll back device configurations on its own (that is handled by revision operations such as Time Voyager rollbacks and subsequent deployment actions). Finally, Revert does not require a commit to take effect; it is used specifically to avoid committing changes. This behavior helps maintain clean operational control in EVPN-VXLAN fabrics by ensuring only validated and intentional intent updates are promoted to the deployed network state.

Verified Juniper sources (URLs):

https://www.juniper.net/documentation/us/en/software/apstra4.2/apstra-user-guide/topics/task/blueprint-commit-revert.html

https://www.juniper.net/documentation/us/en/software/apstra6.1/apstra-user-guide/topics/task/time-voyager-rollback-blueprint-revision.html

In Apstra 5.1, the worker node’s primary purpose is to add scalable runtime capacity to an Apstra cluster by hosting off-box services that would otherwise consume resources on the controller. Specifically, worker nodes run containerized services such as off-box device agents (used to communicate with and manage devices) and Intent-Based Analytics (IBA) components (such as probes and analytics-related services). This design keeps the controller node focused on cluster management and control-plane functions (API handling, cluster-wide state, blueprint control workflows), while shifting resource-intensive operational services to worker nodes.

As your fabric grows—more switches, more telemetry, more devices requiring agent connectivity—CPU and memory demand increases notably, especially when IBA is enabled. Adding worker nodes allows you to scale those container workloads horizontally without redesigning the fabric or reducing analytics coverage. In a Juniper data center built on EVPN-VXLAN with Junos v24.4 leaf-spine roles, this separation helps ensure that Apstra can continuously validate intent, process streaming telemetry, and maintain device communications reliably at scale. Worker nodes therefore exist primarily to offload and scale operational agents and IBA services, improving performance and resilience for larger deployments.