A bank or credit union branch needs a dependable connection to its core processor and other critical vendors, a backup that does not share a path with the primary, network separation between teller systems, ATMs, cameras and lobby Wi-Fi, and paperwork that shows all of it is managed. Regulators do not prescribe a particular circuit, but they do expect institutions to understand their dependencies, plan for outages, test those plans and oversee third parties, and carriers are third parties. The circuit is the easy part. The design and the evidence are where institutions get caught out.
This article is general information about common practice. It is not legal, regulatory or compliance advice. Your institution's risk assessment, your examiners and your advisors decide what applies, and supervisory guidance changes over time.
What runs on a branch network
- Teller and platform systems connected to the core banking system, usually hosted by a core processor or run in the institution's own data center.
- ATMs and interactive teller machines, which may connect through the branch network, through their own cellular modem, or both. Video teller machines add a live video stream.
- Phones and the contact center, often on a cloud platform shared with the call center at headquarters.
- Cameras, alarms and access control, some monitored by an outside security company.
- Cash recyclers, signature pads, document scanners and printers.
- Lobby Wi-Fi for members and customers, which must have no route to anything above.
Much of this is low bandwidth. What matters is that the branch can reach a short list of critical destinations, the core, the card processor and the ATM network, with predictable performance and a fallback.
Where the core lives decides the design
Most community banks and credit unions use a core processor that hosts the core system in its own data centers. Some run the core in-house. Either way, every branch needs a reliable path to wherever the core sits, and the core provider may have connectivity requirements of its own: supported connection types, encryption standards, specific IP addressing or a private circuit into its facility. Get those requirements before you choose a carrier.
| Approach | How it works | Trade-offs |
|---|---|---|
| Private circuits to the core | MPLS or private Ethernet from each branch or from headquarters to the core provider | Predictable and private, but less flexible and often costlier per site |
| Encrypted tunnels over internet | IPsec tunnels from each branch to the core or headquarters across ordinary internet circuits | Flexible and often cheaper; performance depends on the circuits underneath |
| SD-WAN | Several circuits per branch, managed centrally, with encrypted tunnels and policy-based failover | Strong resilience and visibility; needs careful configuration and change control |
Our comparison of MPLS vs SD-WAN walks through the choice. Many institutions land on SD-WAN across the branches with a private Ethernet link between headquarters, the operations center and the core provider or data center.
What regulators tend to expect, in general terms
US banking agencies publish the FFIEC IT Examination Handbook, which includes booklets on business continuity management and on architecture, infrastructure and operations. Banks also look to interagency guidance on managing third-party relationships, issued by the federal banking agencies in 2023. Credit unions are supervised by the NCUA and state regulators, which publish their own guidance on vendor due diligence. Information security standards that implement the Gramm-Leach-Bliley Act apply to both.
None of these tell you which circuit to buy. Read broadly, they point to a consistent set of expectations for connectivity:
- Know your dependencies. Document which branches rely on which circuits, carriers and vendors, and what fails if one goes down.
- Plan for disruption. Have a business continuity plan that covers loss of connectivity at a branch, at headquarters and at a key vendor.
- Test it. Failover that is only on paper is weak evidence. Record when you tested, what happened and what you fixed.
- Oversee your vendors, carriers included, in proportion to how critical they are.
Our guide to a business continuity plan for connectivity covers how to write the connectivity section of that plan.
Redundancy that holds up
Two circuits from two carriers can still share a conduit, a pole line or a building entrance, and fail together. For critical sites, such as headquarters, the operations center and large branches, ask each carrier how its circuit reaches the building and confirm in writing that the paths are diverse. Our guide to route diversity and redundancy explains how to check.
- Small branches: a wired primary with a business cellular backup is common, and the cellular path is naturally separate from the street.
- Full-service branches: two wired circuits from different networks, or wired plus cellular, on an SD-WAN device that fails over on its own.
- Headquarters and operations: dedicated internet access with an SLA, a second carrier on a confirmed diverse path, and private links to the core or data center.
- ATMs: many institutions give off-site and drive-up ATMs their own cellular connection, with or without the branch line as a second path.
Put network equipment on a UPS, and on the generator where the branch has one, so a short power cut does not take the branch offline before the backup matters.
Due diligence on carriers
Telecom carriers are not always used to the questions financial institutions ask, so ask early and keep the answers in your vendor file.
- What are the SLA terms for availability, repair time and notification of outages and maintenance?
- Who do we call, at what hours, and how are trouble tickets escalated?
- Does the service use subcontractors or another carrier's network for part of the path?
- What security documentation can you provide? Some carriers can share audit reports for parts of their business; others offer summaries or attestations. Note what you received.
- What notice do you give before changing the service, the route or the equipment?
- How do we exit, and how long does it take to move the service?
The point is not to collect paper for its own sake. It is to be able to show you understood the dependency and chose with your eyes open.
Security on the branch network
- Segment everything. Teller and platform, ATMs, cameras and alarms, phones and lobby Wi-Fi each on their own network, with rules between them.
- Encrypt between sites, whether over private circuits or the internet, according to your policy and your core provider's requirements.
- Change control. Record network changes, who approved them and how they were tested. Examiners and auditors often ask.
- Monitoring and logging that someone actually reviews. Our cybersecurity and managed network services can cover firewall management and alerting if you do not have staff for it.
The contact center and remote staff
When a branch goes down, members call. That makes the contact center part of the continuity plan, not a separate project. If the phones run on a cloud platform, decide in advance where calls go when headquarters loses its connection: another site, staff working from home, or a group of mobiles. Remote staff who handle member data need the same care as a branch, with managed devices, an encrypted connection back to the institution and a clear rule on what may be done from home. And because 911 calls from a multi-line phone system must carry a usable location, the platform needs a current location for anyone working away from a branch.
A worked example
Hypothetical: the credit union, branch count and design are invented to show the reasoning. They are not a real institution or a quote.
A credit union has six branches and a headquarters, uses a hosted core, and has one carrier providing every circuit on a single private network. A storm cuts the carrier's fiber in the street and four branches lose the core at once. The branches fall back to limited offline processing, and members are turned away from the drive-up lanes.
The redesign puts every branch on SD-WAN with two circuits: a wired primary and either a second wired line from a different network or business cellular. Headquarters gets a second carrier on a diverse path, confirmed in writing, and a private Ethernet link to the core provider's facility as the core provider requires. ATMs get their own cellular connections. The credit union documents the design, tests failover at each branch and records the results, and updates its continuity plan and vendor files. No single carrier outage can take more than one path at each site.
Opening or moving a branch
Qualify connectivity before the lease or purchase is final. A site that needs a construction build can take 60 to 90 days for dedicated fiber, and the core provider's setup and testing add their own time. Cellular can bridge the gap for a temporary branch, but plan the permanent design early so the branch opens with both paths in place.
Questions to ask your provider
- Does your service meet our core provider's connectivity requirements?
- How does your circuit reach each building, and does it share a path with our other carrier?
- What SLA, repair and notification terms apply, in writing?
- What documentation can you provide for our vendor management file?
- Can all branches be managed on one design with one renewal calendar?
Getting your branches priced
Send us your branch addresses, your core provider's connectivity requirements and what each site runs. Several carriers bid on each location, and we bring back the best option per site, the total, and the route information you need for your diversity review. The carrier pays FiberX, so nothing is marked up, and our lowest-price promise applies to qualified like-for-like quotes. If your current setup already wins, we will tell you. The consultation is free, you usually hear back the same day, and Phil Morales is your one contact. Send us your addresses, call 478-758-8091 or text (347) 870-0965.
