Historically, more data has moved on local area networks of all types than across public and private data networks. That was true in the client-server era and arguably remains true in the cloud computing era. Even looking at consumer data consumption, which one might arguably claim is 80-20 in favor of WAN-delivered content versus local on-premise data sharing, most of that end-user-consumed or sent data is over private Wi-Fi--hence local area networks--even if the access connection implies that most data is device to cloud and cloud to device.
Given wide use of Wi-Fi as the LAN medium, device-to-cloud and cloud-to-device data transfers usually feature nearly-identical amounts of data movement "across the LAN" (Wi-Fi) as "across the WAN" (device to cloud to device).
The point is simply that growth of public network data volumes is a reflection of internet app use overall, and hence derivative of that use. That, in turn, has made data centers the focal point of data transfer requirements.
By some estimates there are 8,000 data centers globally, looking primarily at public (multi-tenant) and other large data centers supporting cloud computing. Of course, that only enumerates the big public and hyperscale data center sites. Forecasts suggest we are headed for more than 1,000 such centers by perhaps 2026.
If we include all private enterprise and other entity data centers, there could be as many as seven million data centers operating globally. We sometimes forget how those facts have shaped global data networks.
Some 50 or 60 years, global data traffic largely moved between telephone company central offices. Data network traffic was logically and often physically distinct, moving between private data centers on dedicated facilities and logical networks.
These days, data moves between data centers and to end users of data from data centers. Central offices still function, but their contribution to volume is negligible.
Electricity consumption is one example of footprint, but the bigger implication for much of the connectivity business is the shaping of data communications volume. Traffic volume moving between data centers is just about equal to data served up to, and coming from, actual end users globally.
Internet workloads in the cloud computing era drive traffic across networks, both between data centers and to end users who are the ultimate consumers. Far more traffic moves within each data center.
Edge computing might eventually shift some traffic off wide area networks and confine it locally. Just how much impact edge computing will have on WAN traffic is hard to determine right now. It might be quite fair to predict that cloud WAN traffic will continue to grow even if edge computing moves some workloads off remote data centers.
At a high level, traffic moves between places where internet use is highest. Up to this point, that has meant east-west between Europe, North America and Asia. South-to-South and Europe-to-Africa are growth areas.
No comments:
Post a Comment