To save a huge amount of energy on the AT&T network, we harness Pig and Python
October 16, 2017
A handful of data scientists at AT&T are feeling pretty wired about the projects they’ve been working on.
They’re saving enough energy in the AT&T network in 2017 to power every household in a city of 225,000.
At your house, you can save energy if you remember to turn off the lights when you leave the room. But imagine if you ran a communications network with thousands of switching offices, tens of thousands of antennas, and millions of pieces of equipment.
Flipping down the switches gets a little trickier.
It’s no secret millions of customers are moving from our legacy telephone network and services onto our IP-based network and services. It’s natural to migrate onto newer and better systems. But how do you know when it’s OK to turn off a piece of equipment – or to combine 2 pieces of equipment when only one is needed?
First, some background:
Once upon a time, a single phone call required a dedicated circuit from end to end. Then phones got kind of popular, so AT&T developed “time-division multiplexing” or TDM – combining many calls onto fewer cables by digitizing them.
TDM is the current “old” landline technology, and it still requires masses of specialized equipment spread throughout the nation. Indeed, our inventory database has 100 million records of 100 types of equipment. But it’s not always obvious what’s being fully or partly utilized. It’s hard to eyeball a phone cable and see what’s going through it.
That’s where data analytics comes in. AT&T data scientists and network specialists are loading mounds of data into Hadoop clusters, writing code and finding the exact location, rack and type of switches and other hardware that can be physically turned off by a local technician – without affecting customers.
Click! No more power needed. Thanks for your loyal service, surplus SONET MUX or DS-3 cable. We’re turning down some legacy equipment and consolidating other legacy equipment over the next several years.
We’re proud to be working hard with data analytics across AT&T, using platforms like Apache Pig and coding language like Python. (And some others without cool animal names.) We believe a data-driven enterprise is more efficient and provides better customer service.
And this particular data story is yet another indicator of the benefits of our Software Defined Network. As we convert to SDN on our IP network, we don’t need so many specialty parts. We can use out-of-the-box routers to send phone calls, tweets and streaming video wherever they need to go. We simply upgrade the software as needed.
If you want to read more about our energy-saving efforts and greenhouse gas emissions goals, please see our latest AT&T Energy Management Issue Brief. Last year, AT&T realized more than $100 million in annualized energy savings from more than 25,000 energy projects.