AT&T Service Outages Hit The Midwest Friday (wgntv.com)

An anonymous Slashdot reader quotes WGN: Customers using AT&T for phone and Internet reported widespread outages across Chicago and the Midwest Friday starting around 3 p.m. and going into the evening, according to the website DownDetector. In addition to Chicago, customers in St. Louis, Milwaukee, Indianapolis, and Cincinnati also reported issues like receiving a "Call Failed" message or a recording when they tried to make phone calls, or Internet simply not working. Outages in the Chicago area seem to have peaked around 8 p.m. The number of people registering issues on the website Outage Report seems to have peaked around that time as well.
The outage "has been repaired," reported an Indiana TV station early Saturday morning. Anyone else having connectivity issues? Share your stories in the comments.

Comments

World Series?

[seven of five]

Cubs home for World Series... everybody tuning in... wouldn't be surprised if ATT choked on traffic.

Unsurprising

[Zak3056]

Legacy telco does not seem to be able to build bulletproof networks anymore.

Telephony is the original give nines application. In the past, and outage like this would probably have resulted in congressional hearings, but there have been numerous wide area outages from AT&T and other baby bells over the last few years, and the frequency of these outages just seems to get higher as years go on.

Re:Unsurprising

[grumling]

When the goal is to deliver 4KHz worth of audio that hasn't changed at all over the last 100 years, It's pretty easy to get to 99.99% up time. Note that 99.999% uptime was only for SONET level circuits. When the product is narrowly defined by you, you get the ability to define what reliable service is. Helps to have the ability to raise rates every few years even when your operating costs drop. Also helps to control the central office upgrade cycle. That way you can make damn sure that switch's software is mostly bug-free.

Once you start running advanced networks it gets a little more hairy. And now that flat-rate pricing, instead of per-minute billing, is the norm, people are using the network much more than they used to. Because the upgrade cycle is less than 10 years, it makes sense to centralize as much as possible, especially for CPE provisioning and services. Years ago, switching for local calls was handled entirely within the local CO. These days there's going to be a virtual switch that could be located in another state, or maybe even just one or two for the entire country. If the local CO's link fails, or there's a badly managed maintenance order, what would have been isolated to a single town can now affect whole regions or even be national. But it helps keep the margins high and maybe keeps the competitors at bay.

Not saying it's right, just saying that current network management practices encourage centralization. And maybe there's something to that. Fixing one (hot backed up) thing that can restore service to millions in a few minutes might be cheaper and better than single point of failure devices all through the network. Certainly easier to patch and maintain a few redundant servers than thousands of individual units.

Re:Unsurprising

[Waffle Iron]

I doubt that they ever got even 99.99% uptime for end users. I distinctly remember occasionally getting "fast busy" signals when I tried to make calls (major metro area in the 70s), and it these downtimes almost certainly happened more than 52 minutes per year. Maybe the circuits were just overloaded, but my phone was still useless while it happened.

They very well may have had *less* uptime than a decent ISP has now. The main difference is that with almost constant use of the internet, you're more likely to notice any outages and get annoyed.