AT&T enabled 34% of its network in 2016 with software, bringing the telco another step closer to reaching its goal to virtualize a large portion of its network with software by 2020.
The service provider has set a large goal to virtualize 75% of its network with software by the year 2020.
John Donovan, chief strategy officer and group president for AT&T, told attendees at the 2017 Internet, Media & Telecommunications Conference that it surpassed its goal to virtualize 30% of its network.
“Our target for last year was to be at 30%, so we hit 34%, which we’re really thrilled with,” Donovan said. “The hardest part of how big scale projects go is you have a year of planning, a year at 5% and a year at 30% and a year that takes you to 50% and then you start to burn down the tail.”
Donovan added that getting to 34% puts AT&T in a position to complete its virtualization goals for 2017.
“The hardest part is going from 5 to 30 because that’s where you have to get really good at scaling,” Donovan said. “For us to hit 34% of our network functions in software really bodes well for us for 2017 to get the things done that we want to do.”
Already, AT&T is seeing the fruits of its software labor take shape in terms of reducing costs and efficiency in its network.
To create network diversity and redundancy, the service provider does not have to deploy multiple hardware elements, for example.
“This industry grew up where if you wanted to have high reliability and you had to have a redundant card and a redundant shelf and a redundant route and all of that stuff was done with physical hardware,” Donovan said. “Much like the web scale players today where software moves around that’s the way our network works in the future.”
Donovan likens the new software-centric network as enabling AT&T to provision services and functions much like an app store.
“When we deploy in our network a node, we drop software in there and we’re not buying an appliance to do firewall, or buying an appliance to do routing,” Donovan said. “It’s a software load and we drop the software load in and we can move that around to get network redundancy.”