What would be the recovery time-frame for a telecom breakdown?

Phone-number scramble could be fixed almost immediately by bringing a redundant database online. You’d never know the difference. Big stuff like telecomms is typically very redundant. And they get great contingency planning.

A hardware failure would take some hours, but most service would stay intact and the calls routed around the hole.

24 hours if you set up an old number/new number website that people could update themselves. @Nullo The database would have to be updated first with the new numbers.

@flutherother They can’t just switch over to a backup on a second server and keep the old numbers? Even on a cell network?

@Nullo I think you are right. If the exchange knew nothing about the change of number when you called the old number it would follow the old route to the correct person. If the exchange did know about the change of number it could easily link the old and the new.

Nullo is correct. It is as easy as pressing a button to activate the stand by side of the database.

Pretty quick with caller ID. You call your own number, and the person who answers tells you what number appears, and you tell them what number you dialed. You both know your new number. And, a lot of phones identify what number your phone is anyway.

Just as long as they don;t mess with area codes too….

As long as mail was still functioning about 2–3 days max.

I was a communications technician with AT&T for a few years. I worked in a central office on call-processing equipment (DMS 100).

Hardware redundancies provide hardware backup but rarely software backup, unless the duplex processors aren’t running in parallel. Most of the time though, their software content is duplicated, and in the case of global data mismatches, all the processors would have the erroneous assignments.

Most central offices have a duplicate of the earlier (hopefully very recent and correct) data on some form of removable media (RM), such as a disk or external drive.

What they would do is remove one processor from service, and load it with the previous data stored on the RM. Once that processor was successfully returned to service with the correct data, then almost everyone would be back in service. The only phone numbers that wouldn’t be back in service would be those assigned after the last system image was copied to the RM.

I retired from AT&T in 2003. Based on the technology at that time, it would take about an hour to remove a processor from service and load the correct information for about 60,000 customers.

Given a rough estimate of the rate at which new phone numbers are assigned or changed, and our practice of taking an office image weekly, it would take about another hour to restore service to those customers whose correct data wasn’t on the RM.

If all other central offices were at least at our level of readiness and responded at about the same time, then I guess it would take about 2–3 hours, depending on the size of the office.

The most populous central office that I know of has about 120,000 customers. An office that size could take about 4 hours to restore service.

The above scenario would be with intraoffice mismatches. If the mismatches involved randomly reassigning area codes and prefixes to different central office processors, then that could take at least a week of reassigning, unless there are RM’s with multiple-office and interoffice-routing data.