This story has been floating around since the 14th and it’s only now that I’ve had a chance to wrap my head around it. The guys at Fierce Wireless, eagle eyed as they are managed to uncover some juicy details in an FCC filing regarding an Android Instant Messaging program that almost brought down the entire T-Mobile network in an unknown city.
“T-Mobile network service was temporarily degraded recently when an independent application developer released an Android-based instant messaging application that was designed to refresh its network connection with substantial frequency,” Castle wrote in the filing. “The frequent refresh feature did not create problems during the testing the developer did via the Wi-Fi to wireline broadband environment, but in the wireless environment, it caused severe overload in certain densely populated network nodes, because it massively increased signaling–especially once it became more popular and more T-Mobile users began downloading it to their smartphones. One study showed that network utilization of one device increased by 1,200 percent from this one application alone. These signaling problems not only caused network overload problems that affected all T-Mobile broadband users in the area; it also ended up forcing T-Mobile’s UMTS radio vendors to re-evaluate the architecture of their Radio Network Controllers to address this never-before-seen signaling issue. Ultimately, this was solved in the short term by reaching out to the developer directly to work out a means of better coding the application.”
This whole thing is beyond my scope of development knowledge but when the carrier has to reach out to the developer to fix a problem, that’s obviously not a good sign. As Fierce Wireless points out, this one again highlights the troubles facing wireless carriers in this new age of significant data traffic flowing along their networks. T-Mobile Director of National Planning, Grant Castle argues that carriers must be allowed to “prioritize data and manage applications via techniques including scheduling algorithms, channel selection and, possibly, dividing applications up by frequency
The boys at Fierce Wireless put together an interesting piece regarding this and it’s worth hitting the link for the read!