Keep the site open when the WAN fails.

Challenge

The site makes money or serves patients only while a single circuit holds, and the circuit fails at the worst hours of the year.

• The average retail location experiences 14 to 18 hours of unplanned network downtime a year. One national retailer put a single hour of POS downtime on Black Friday weekend at $480,000 across its fleet, and 67 percent of customers who hit a payment problem do not return to finish the purchase (Jumpmind, April 2026).
• Cloud only store systems with basic caching go read only or stop when the link drops; sales, returns and inventory updates wait, and reconciliation afterward takes hours (same source).
• Chick-fil-A runs Kubernetes at the edge in every restaurant for low latency, scale and continuous business, with equipment that keeps working when the connection to the central site is down (Network World). Few chains, banks or clinic networks have the platform team that built it.
• Branches and clinics carry the same exposure with higher stakes: payments, scheduling, pharmacy and lab systems reached over one circuit.

Approach

Norseman gives every site the services that must survive an outage, two ways out, and one team running the fleet.

1. Deploy the Odin's Edge Site profile in a half rack or wall cabinet at each site, integrated and tested in Norseman's lab, shipped staged, installed by site staff or Norseman field engineers.
2. Run the services that must survive locally: the store, branch or clinic application tier, product and price data, a local identity cache, payment and print gateways, and a local copy of the schedule or roster, all syncing with the central system when the link returns.
3. Give every site two paths: primary broadband plus cellular or satellite failover with automatic switchover, managed as part of the platform.
4. Standardize the fleet: one blueprint, changes delivered to every site through staged rollouts with rollback, no store by store projects.
5. Operate it: 24/7 monitoring of services and links, patching under SLA, backup verification. Or hand it to the customer's team with the same console and runbooks.

Typical Outcomes

The site keeps selling, booking and dispensing while the circuit is down, and nobody on site has to do anything. Items marked target are design goals until a pilot reports.

• Transactions complete during outages and customers do not notice (target; measured as transactions completed while the WAN is down).
• Hours of disconnected operation become a published number per site (target; measured in the lab per profile).
• One blueprint across the fleet, so a change lands everywhere in a day instead of a quarter.
• No IT staff at the site; the store, branch or clinic manager has one number to call.
• The same site runs the AI when the mission is ready: loss prevention, queue and shelf analytics, document assistants in the branch.

Procurement Paths

• Commercial chains, banks and clinic networks: per site subscription or direct purchase from Norseman, with financing and a customer catalog for repeat site orders.
• State and local public sector (DMVs, county clinics, campus retail): GSA MAS through cooperative purchasing where the state allows it, or state contracts.
• Federal retail and service sites (exchanges, commissaries, base clinics): NASA SEWP VI and GSA MAS, with a hardware line and a service line per site.
• A budgetary estimate and a business case the sponsor can carry to finance come with the first conversation.

Compare vehicles and ordering steps on How to buy.

Partner Technology Examples

• Connectivity and failover: Ericsson Cradlepoint, Peplink, Fortinet
• Platform: SUSE Rancher and Harvester, or Spectro Cloud Palette
• Hardware: Dell Technologies, Hewlett Packard Enterprise, Supermicro
• Data sync: Jumpmind SymmetricDS (open source), PostgreSQL
• Security and identity: Palo Alto Networks, Keycloak (open source), Yubico
• Backup and operations: Veeam, ServiceNow, Grafana and Prometheus (open source)