USAID has two core locations on the U.S. coasts, with a small footprint to connect its cloud-service providers.
“As far as the user base is concerned, we have 13,000 to 14,000, with more than half of those overseas and the rest here in Washington, D.C.,” Singh said on Federal Drive with Tom Temin.
The so-called “interconnects,” also called USAID’s “co-locations,” on the West and East coasts, are sort of like satellite downlink centers.
“When I say interconnect, they are direct connects to the major data centers for the cloud service providers,” Singh said. “We don’t go through the general internet, basically. So it is a back-end connection to the cloud providers.”
USAID uses an MPLS to connect domestic users to its core locations and data centers on the East Coast, where the bulk of missions go through. Meanwhile, many overseas users are routed to the West Coast location because USAID can cut down latency anywhere from 60 milliseconds to 100 milliseconds, in some cases, Singh said. Some applications do run at the centers, but about 95% of the data storage is in the cloud.
“We do have some very specific applications in the co-locations and they are mostly our management-and-control kind of applications, whether we want to control our security stack or our wireless or our network management,” Singh said. “Those kinds of applications or services are in co-locations and are very specific because of security and the way we have designed the overall network management.”
But to be clear, overwhelmingly, the data for USAID is in the cloud.
“We’re not dumping any bulk data in the co-locations,” Singh said. “There is also some backup related data that we do use or store at the co-locations, but not very much.”
And when it comes to where USAID data is stored, Singh said they go by the book.
“We use FedRAMP, you know, public cloud, and really go by the security controls and requirements, Singh said. “We do ask certain questions as to where our data is, which, in most cases, is in the continental United States. Some providers store parts of the data in different locations, so they can have redundancy or backup in case there is an issue with the cloud service provider, resulting in the need to recover the data.”