MobiledgeX, in its description of the infrastructure edge network deployed in Germany, provides a bit of insight on where infrastructure edge facilities are deployed, at least in the first iteration.
To support 60 million smartphones in Germany (or other devices and use cases), 40,000 tower sites connect into 900 access sites, which connect to 11 mobile core network locations.
The mobile core connects into transport networks that connect to 3 “Availability Zones” in Frankfurt.
The takeaway is that DT supports 900 edge computing locations, supporting 40,000 tower sites, roughly a 1:44 ratio of infrastructure edge facilities to towers.
And that might illustrate a practical implication of early infrastructure edge deployments: they will not generally include computing sites “at every macrocell tower,” as sometimes suggested might be the case.
Greg Pettine, EdgeMicro EVP, says his firm is not looking to build edge computing data centers at the base of cell towers, mainly because there isn't enough fiber in those locations to satisfy customers' edge computing requirements.
Vapor IO, another start-up firm, also isn't putting its edge computing data centers at the base of every single cell tower. it's "not relevant to talk about sitting at the bottom of one tower," said Vapor IO CEO Cole Crawford.
Instead, the company is looking to place edge computing sites at aggregation sites.
"We never actually said that they had to be located at cell towers," Crawford said, noting that Vapor IO's first edge computing data center in Chicago is running inside a Distributed Antenna System hub with a significant amount of nearby fiber.
If you think about it, that is a practical approach for any young or small firm with limited capital to invest. Eventually, should the business develop as expected, more sites, closer to the radio edge of the network, might be needed.
That is an established pattern in the communications or any other business with high capital requirements. Initially, a few sites are mandated both by capital limitations and demand as well. Over time, sites multiply.
Also, note that the architecture for infrastructure edge computing, or edge computing generally, has a layered structure. In a large country such as the United States, one might find hundreds of hyperscale data centers, and tens of thousands of infrastructure edge computing sites.
For reference, recall that there are perhaps 375,000 U.S. macrocell sites (counting all carriers on towers), including perhaps 175,000 “tower” locations.
So using the base of 175,000 macrocell sites, infrastructure edge computing sites, even when fairly well built out, might have only “tens of thousands” of such sites. Call the number 30,000. That implies a 1:6 ratio of edge computing facilities to towers.
That, in turn, suggests aggregation sites are the most-likely places where edge computing occurs, much as modern mobile networks try and centralize processing, removing cost from radio sites.
The point is that infrastructure edge computing is more likely to occur at more-centralized aggregation sites than at “every macrocell site.”
Keep in mind that early deployments are not the same as later, fully-built networks. With any new capital investment at the edge, it is rational to build incrementally. In fact, even with unlimited capital resources, that still would make sense, as there are other constraints (permitting, availability of construction crews, acceptance testing and so forth).
No comments:
Post a Comment