350-501
Free 350-501 practice questions
10 questions from our 350-501 practice tests, each with the correct answer and a written explanation. Read them with the answers shown, or switch to quiz mode and answer first.
Question 1 of 10, Architecture
An MPLS VPN provider must preserve the customer's original DSCP marking end to end, while the provider core queues traffic according to its own MPLS EXP marking and the egress PE applies the customer's per-hop behavior based on the original IP DSCP. Which MPLS DiffServ tunneling model is described?
- AExplicit-null uniform mode with EXP copied back to DSCP
- BShort pipe mode Correct answer
- CPipe mode
- DUniform mode
Answer and explanation: In short pipe mode, the provider marks EXP independently of the customer DSCP, the DSCP is never rewritten, and the egress PE queues on the customer's IP DSCP. Pipe mode also preserves DSCP, but the egress PE queues on the provider's EXP value instead. Uniform mode treats the network as one DiffServ domain, so a change to EXP in the core is copied back into the IP DSCP at the egress, which rewrites the customer marking. Copying EXP back to DSCP is the defining behavior of uniform mode, not a way to preserve the original marking.
Question 2 of 10, Architecture
An ISP edge router must drop packets whose source address is not reachable through the interface on which they arrive, on single-homed customer links. On multihomed customer links with asymmetric routing, it must only drop packets whose source has no route at all. Which configuration meets both requirements?
- AInbound ACLs permitting only RFC 1918 sources on every customer link
- BStrict-mode uRPF on single-homed links and loose-mode uRPF on multihomed links Correct answer
- CStrict-mode uRPF on all customer links with allow-default enabled
- DLoose-mode uRPF on single-homed links and strict-mode uRPF on multihomed links
Answer and explanation: Strict uRPF requires the source to be reachable via the receiving interface, which suits symmetric single-homed links, while loose uRPF only requires that a route to the source exists, which tolerates asymmetric multihomed paths. Reversing the modes would drop legitimate traffic on multihomed links and under-protect single-homed links. Strict mode everywhere breaks asymmetric links even with allow-default. Permitting only RFC 1918 sources would block the customers' public address space, which is the opposite of anti-spoofing.
Question 3 of 10, Networking
A customer with two sites in AS 65010 connects both sites to an ISP via eBGP. Site B never installs routes from site A. Which feature resolves the problem on the provider side without changes at the customer?
- ASite of Origin extended community on the PE sessions toward the customer
- Bas-override on the PE sessions toward the customer Correct answer
- Callowas-in on the PE sessions toward the customer
- Dnext-hop-self on the PE sessions toward the customer
Answer and explanation: Site B drops the routes because it sees its own AS 65010 in the AS path, and as-override on the PE replaces the customer AS with the provider AS, so the loop check passes without customer changes. allowas-in would have to be configured on the customer routers that receive the route, which is a customer change. next-hop-self does not affect the AS path. Site of Origin marks routes to prevent loops back into a multihomed site and does nothing to the AS path check.
Question 4 of 10, Networking. Choose 2.
Two PEs receive the same VPNv4 prefix from a dual-homed CE. The route reflector has both paths but sends only one to the ingress PE, and failover takes seconds. Which combinations allow the ingress PE to have a precomputed backup? (Choose two.)
- ASet the same RD on both PEs and increase the BGP hold timer
- BConfigure the MED on the backup PE to a value lower than the primary
- CEnable BGP Add-Path on the RR toward its clients for the VPNv4 family Correct answer
- DConfigure unique RDs per PE so the RR treats both paths as different prefixes Correct answer
- EDisable next-hop-self on the egress PEs toward the RR
Answer and explanation: With a unique RD per PE, each path becomes a distinct VPNv4 route and the RR reflects both, and Add-Path lets the RR send multiple paths for the same prefix; either way the ingress PE learns a backup for PIC. A shared RD with longer timers makes the RR pick a single best path and slows detection. Lowering MED on the backup would make it primary rather than a precomputed backup. Next-hop-self settings do not change how many paths the RR reflects.
Question 5 of 10, Networking. Choose 2.
Which statements about BGP graceful restart are correct? (Choose two.)
- ABoth the restarting router and its neighbors must support the capability, negotiated in OPEN Correct answer
- BIt keeps the BGP TCP session up through an RP switchover without any neighbor involvement
- CIt is configured only on the restarting router and needs no support from the neighbors
- DThe restarting router's neighbors keep forwarding through its stale routes while it restarts Correct answer
- EIt eliminates the need for BFD, because failures are detected by the restart timer
Answer and explanation: With graceful restart, neighbors acting as helpers retain the restarting router's routes as stale and keep forwarding while the session is re-established, and the capability is advertised in the OPEN message, so both sides must support it. Keeping the TCP session up without neighbor involvement is NSR, not graceful restart. GR is for planned restarts and does not replace fast failure detection. Neighbors must participate as helpers.
Question 6 of 10, MPLS and Segment Routing
After an IGP link comes up, traffic is black-holed for several seconds because IS-IS starts using the link before the LDP session across it is established. Which feature prevents this?
- ALDP session protection
- BLDP-IGP synchronization Correct answer
- CMPLS LDP autoconfiguration
- DLDP graceful restart
Answer and explanation: LDP-IGP synchronization makes the IGP advertise the link with maximum metric until LDP is operational on it, so traffic is not sent over a link without labels. Session protection keeps a targeted LDP session up during a link failure so labels are retained, which helps on link down rather than link up. Graceful restart preserves forwarding during a control-plane restart. Autoconfiguration only enables LDP on IGP interfaces and does not delay their use.
Question 7 of 10, MPLS and Segment Routing
An operator wants a subset of traffic to follow a low-latency path that is computed by each router in IS-IS itself, without a PCE or SR policies on the head ends, using measured link delay as the metric. Which feature meets this requirement?
- AIS-IS multi-topology with a separate delay-based topology
- BAn SR-TE policy on each head end with dynamic path computation on the TE metric
- CRSVP-TE tunnels with auto-bandwidth and affinity constraints
- DIGP Flexible Algorithm (for example, algo 128) using the delay metric Correct answer
Answer and explanation: Flexible Algorithm lets routers advertise a definition (metric type, affinities) and compute a constrained SPF per algorithm, producing prefix SIDs that follow the low-delay path with no PCE. An SR-TE policy requires head-end policy configuration, and the TE metric is not the measured delay. RSVP-TE needs signaled tunnels and does not compute on delay by default. Multi-topology IS-IS separates address families and does not use a delay metric.
Question 8 of 10, Services
A tier-2 ISP buys MPLS transport from a backbone carrier and wants to run its own MPLS VPNs, while the backbone carrier keeps only the tier-2 ISP's internal routes in its VRF, not the end-customer routes. Which service is described?
- AInter-AS option A with the tier-2 ISP as one AS and the carrier as another
- BAn EVPN E-LAN service carrying the tier-2 ISP's routers in one bridge domain
- CCarrier Supporting Carrier, with labels exchanged between CSC-CE and CSC-PE Correct answer
- DUnified MPLS with BGP labeled unicast inside the backbone carrier's IGP
Answer and explanation: In Carrier Supporting Carrier, the customer carrier's internal routes and labels (through LDP or BGP labeled unicast) are exchanged with the backbone carrier's VRF, so the backbone does not see the end customers. Inter-AS option A peers VRFs back to back and would carry customer routes on the border. Unified MPLS describes how one operator scales its own transport. An E-LAN service gives Layer 2 connectivity and is not the defined CSC model.
Question 9 of 10, Services
A customer runs OSPF as the PE-CE protocol at two sites and also has a backdoor link between the sites in the same OSPF area. Traffic between the sites always uses the backdoor link instead of the MPLS VPN. Which feature allows the VPN path to be preferred?
- ARedistributing the VPN routes into OSPF as external Type 2 LSAs on the PEs
- BConfiguring the PE VRFs with a different OSPF domain ID than the CE routers
- CAn OSPF sham link between the PE VRFs, with its cost lower than the backdoor link Correct answer
- DSetting the down bit on the Type 3 LSAs that the PEs send to the CEs
Answer and explanation: Routes learned through the MPLS VPN appear as inter-area routes, which lose to intra-area routes over the backdoor, and a sham link makes the VPN look like an intra-area link so cost can decide. A different domain ID makes the routes external, which is even less preferred. The down bit prevents loops but does not change preference. External Type 2 routes are always less preferred than intra-area routes.
Question 10 of 10, Automation and Assurance
An engineer needs to change the configuration of 200 IOS XR routers from a YAML inventory, using an agentless tool that connects over SSH or NETCONF and runs idempotent tasks from playbooks. Which tool fits this requirement?
- AAnsible Correct answer
- BCisco NSO with a Python service package
- CTerraform
- DA gNMI collector such as Telegraf
Answer and explanation: Ansible is agentless, uses inventories and YAML playbooks, and connects over SSH or NETCONF with idempotent network modules. Terraform is a declarative infrastructure-as-code tool with a state file, more common for cloud and controller APIs. NSO can do this but is a full orchestration platform that is not an agentless playbook tool. Telegraf collects telemetry and does not push configuration.