Telco BSS modernization without downtime

Adaobi: Patrick, BSS. Set the stage. Patrick: Business Support Systems. Customer management, billing, charging, mediation, partner management. Anything between the network and the money. Our legacy BSS was a thirty-year-old mainframe-derived system that nobody in the building had originally architected. Replacing it had been on the strategic plan for ten years. Every previous attempt had been killed by the cutover risk. Kwame: The strangler approach? Patrick: We did not replace it. We surrounded it. We built a facade in front of the legacy system that spoke a modern API. New products were routed through the facade to the new BSS. Old products continued to go to the legacy. The facade reconciled state between the two. Over thirty months, we migrated products one by one. The legacy got progressively smaller. On the day we turned it off, it was serving fewer than three percent of subscribers and they had been pre-migrated in a quiet weekend. Adaobi: Nine months of dual-running? Patrick: At peak, yes. Every subscriber had a record in both systems. Reconciliation ran every thirty minutes. Discrepancies above a threshold paged on-call. We saw about eighty discrepancies a day at peak — mostly rounding differences in fractional currency. We closed every one. By the end of dual-running, the discrepancy rate was below five per day. Kwame: The cutover weekend you never had? Patrick: That is the punchline. There was no cutover weekend. By the time we were ready to turn off the legacy, ninety-seven percent of traffic was already on the new system. The last three percent migrated over a Saturday afternoon by ops staff working remotely. The CEO did not even hear about it until the following Monday. Adaobi: Advice to other telcos? Patrick: Stop planning the big cutover. Start planning the strangler. Modernization is patience and a facade. There is no other way. Kwame: Patrick, thank you. Patrick: Pleasure.