VMware SD-WAN Design and Deploy Skills 2023: The VeloCloud Transition

The VMware SD-WAN Design and Deploy Skills 2023 credential is a legacy VMware-era path built around the 5V0-42.21 VMware SD-WAN Design and Deploy Skills exam. The old exam covered VeloCloud architecture, design, Edge high availability, profiles, business policy, routing, gateways, Orchestrator, deployment, and troubleshooting.

The program context has changed substantially. Broadcom retired several VMware SD-WAN skills credentials on June 30, 2025, and effective July 1, 2025 VeloCloud was acquired by Arista. Broadcom support documentation now directs VeloCloud customers to Arista. That makes the VMware certification page historical even though the underlying SD-WAN technology and design skills remain relevant.

A useful 2026 article should therefore preserve the network-design knowledge without pretending a VMware credential remains current. The WAN and SD-WAN fundamentals provide a current concept base, while the old 5V0-42.21 material can still organize VeloCloud-specific historical skills.

SD-WAN architecture starts with control, transport, and policy

SD-WAN separates business intent from the manual configuration of individual branch links. The architecture typically includes edge devices at sites, centralized orchestration, gateways or cloud connectivity functions, and multiple available transports. Policy determines how applications use those transports.

The design question is not simply whether two sites can communicate. It is which path should carry which traffic, how the system reacts when performance degrades, where internet breakout occurs, how cloud access is reached, and how policy remains consistent across many branches.

The cloud networking fundamentals help reinforce routing and gateway thinking. SD-WAN extends those fundamentals with centralized policy and dynamic path selection.

Transport diversity is valuable only when failure behavior is designed

A branch may have broadband, MPLS, LTE or 5G, and other transports. Multiple links create options, but resilience depends on how those links differ and how policy uses them. Two circuits that share the same physical provider path may not provide the independence the diagram suggests.

Design should classify applications by latency sensitivity, loss tolerance, bandwidth, security, and business criticality. Voice and interactive applications may need fast failover and low jitter. Bulk data may tolerate slower transport. Critical applications may require stronger path preference or multiple viable routes.

Practice by removing one transport from the design and asking what happens to each application class. Then degrade a path without fully failing it. SD-WAN value is clearest when the design responds to performance as well as binary link state.

Business policy translates application intent into forwarding behavior

VeloCloud business policy historically allowed administrators to classify traffic and influence path, priority, service, and steering behavior. The transferable skill is policy design: define what the application needs before choosing a transport action.

Avoid building policy from application names alone. Consider user group, site type, security requirement, bandwidth, latency, SaaS destination, and fallback behavior. A policy that works in one branch may fail in another if transport or local breakout differs.

Document the default behavior as carefully as the special cases. Complex policy sets become difficult to troubleshoot when administrators remember exceptions but cannot explain what happens to traffic that matches none of them.

Routing still matters under an SD-WAN overlay

SD-WAN does not remove IP routing fundamentals. Branches still interact with LAN networks, data centers, cloud networks, dynamic routing protocols, static routes, and default gateways. Overlay intelligence must coexist with underlay reachability.

The SDN and SD-WAN concepts reinforce the distinction between centralized control and the networks that still carry packets. Candidates should be able to trace route selection through both layers.

Troubleshooting should ask whether the failure is local routing, overlay tunnel health, policy, gateway behavior, DNS, or application reachability. Reconfiguring SD-WAN policy cannot fix a broken LAN route.

High availability must include the edge device itself

Multiple WAN links do not create branch resilience if the only edge device fails. VeloCloud designs included Edge high-availability options to reduce that single point of failure. The design needs to consider device redundancy, state, link attachment, LAN connectivity, power, and failover behavior.

The high-availability principles are useful because they force the architect to identify the actual failure domain. Redundancy is meaningful only when duplicated components do not share the same hidden dependency.

Test HA operationally. A diagram may show two edges, but a failed upstream switch, shared power source, or incorrect cabling can make both unavailable at once.

Orchestration simplifies scale but becomes a critical dependency

Central orchestration allows profiles, policies, monitoring, and configuration to be managed consistently across many branches. This reduces configuration drift but also means the control and management plane must be secured, monitored, and understood.

Templates and profiles should encode stable design patterns, not every temporary exception. Excessive per-site overrides undermine the operational value of centralized orchestration and make troubleshooting unpredictable.

A good design defines ownership. Which settings are global? Which can vary by region or site class? Who approves exceptions? How are changes tested and rolled out? These questions remain relevant regardless of who owns the VeloCloud product.

Monitoring should focus on application experience, not just link up/down state

SD-WAN exists partly to make path decisions based on quality. The network observability model helps frame metrics such as latency, jitter, loss, bandwidth, and path change as evidence about user experience.

An interface can remain technically up while application performance is unacceptable. Operators should establish normal ranges by transport and site type, then investigate deviations with application and path context.

When a user reports poor performance, correlate the time of the complaint with path quality, policy decisions, route changes, and application traffic. This is stronger than restarting an edge because the link light appears normal.

The current path is a product transition, not a VMware exam update

The most important current-status fact is ownership. VeloCloud moved to Arista in July 2025, so new learners should not wait for a new VMware 5V0-42.21 successor. Current support, product documentation, and any future training direction belong with the product’s new owner.

That does not make the old material worthless. Architecture, transport diversity, business policy, routing, HA, orchestration, deployment, and troubleshooting remain useful network skills. Existing badge holders can continue to present the credential historically.

New candidates should study current VeloCloud/Arista material for product-specific operations and use the old VMware exam only as a map of concepts. Certification history and current product ownership should remain clearly separated.

Cloud and SaaS connectivity change the branch design problem

Modern branches often send traffic directly to SaaS and cloud services rather than backhauling everything through a central data center. SD-WAN designs therefore need to decide which destinations use direct internet access, gateways, hubs, or security services and how policy changes when a primary path becomes impaired.

Latency to cloud applications can matter more than distance to the corporate data center. Measure application experience from the branch and choose path policy based on real performance. A low-latency local breakout may be better for SaaS while regulated traffic still requires centralized inspection.

Document fallback behavior. A design is incomplete if it specifies the preferred path but not what happens when the preferred gateway, transport, or security service is unavailable.

Security and SASE considerations should be designed with routing

SD-WAN and security increasingly intersect through secure web gateways, firewalls, segmentation, and SASE services. Routing traffic to a security service changes latency, failure modes, DNS behavior, and policy ownership. Network and security teams need one shared traffic-flow model.

Do not bolt security on after the WAN design is complete. Decide which traffic must be inspected, where policy is enforced, what happens if the security service is unavailable, and how sensitive branches are segmented from one another.

Zero-trust access and SASE can reduce dependence on traditional network location, but they do not eliminate routing or transport requirements. Identity-aware access still needs a reliable path to the application and security service.

Deployment validation should prove the design under failure

After deploying a branch, validate normal and degraded operation. Test primary and secondary transports, application steering, dynamic routing, DNS, local breakout, tunnel establishment, HA behavior, and monitoring. A configuration that works only in the steady state is not a completed deployment.

Record baseline latency, jitter, loss, and bandwidth for each transport. Then deliberately introduce a controlled impairment in a lab or maintenance window and verify that traffic moves as designed. This gives operators confidence that business policy reflects actual behavior.

Finally, document the recovery path. When the preferred link returns, confirm that sessions and new flows behave as expected and that no stale routing or policy state remains.

Branch design should also account for operational scale. A policy that is easy to manage across ten sites can become unmanageable across hundreds if every branch has a unique exception. Standardize branch archetypes, transport classes, addressing, and policy whenever the business permits, then document the small number of justified deviations.

QoS and application identification require realistic validation. Marking traffic with a priority does not guarantee a good experience if the underlay is congested or the provider ignores the marking. Test how policy behaves across the actual transports and use measured loss, latency, and jitter to validate design assumptions.

Change control matters because centralized orchestration can amplify both good and bad configuration. Stage changes, use small pilot groups, preserve rollback, and monitor after deployment. Central management reduces repetitive work, but it also increases the blast radius of an incorrect global policy.

Because VeloCloud ownership moved to Arista, current administrators should also verify terminology and release behavior against Arista documentation before implementing historical VMware procedures. Keep the old topology and policy concepts where they still apply, but refresh software versions, supported integrations, lifecycle steps, and support processes from the current product owner.

The enduring skill is therefore architecture judgment: design predictable paths, test failure behavior, keep policy explainable, and use telemetry to confirm that the branch experience matches the design.

Study with ExamSnap to prepare for VMware SD-WAN Design and Deploy Skills 2023 Practice Test Questions and Answers, Study Guide, and a comprehensive Video Training Course. Powered by the popular VCE format, VMware SD-WAN Design and Deploy Skills 2023 Certification Exam Dumps compiled by the industry experts to make sure that you get verified answers. Our Product team ensures that our exams provide VMware SD-WAN Design and Deploy Skills 2023 Practice Test Questions & Exam Dumps that are up-to-date.