Defining Networks in Terms of Latency

One new way of looking at a network is to define it not only in terms of bandwidth but also latency. And that might be true for edge computing value as well as dense fiber deployments or 5G. 

Using 5G, some form of local off-premises edge infrastructure or a dense fiber network, latency might be defined as within five milliseconds to 12 milliseconds for round-trip latency, though 5G air interface latency might be as low as one millisecond, some note. 

source: STL Partners 

Some say round-trip latency--when using edge computing--means a server or computing facility can be located up to 150 miles distant and still have a round-trip latency of five milliseconds. Round-trip latency of 10 milliseconds involves computing facilities up to 300 miles distance. 

Round-trip latency of 20 milliseconds implies a computing location up to 600 miles away. Round-trip latency of 50 milliseconds can mean communicating with a computing facility or server up to 1,500 miles away. 

source: STL Partners

 

source: Equinix

Edge Role Bound to Grow

To the extent that application performance is a significant problem for supporting remote workers, and to the extent many enterprise managers expect high levels of remote work in the future, the demand for edge computing to address latency and improve end user performance seems rather self evident.

With the caveat that the situation remains volatile, a new survey by Aryaka finds enterprise managers reporting they have closed 25 percent to 50 percent of their sites.

A quarter of respondents expect 51 percent to 75 percent of their employees to remain hybrid post-pandemic, and another 43 percent expected remote work to be the pattern for 25 percent to half of employees.

Should this continue, we should expect to see other changes, especially changes in real estate spending and probably on-site data centers as well, given the secular move to “everything as a service.” About half of respondents said they would be closing in-house data centers as part of wider shifts. 

XaaS highlights the importance of remote application delivery using the internet, which in turn heightens the importance of access and security as though everyone were using internet access and delivery. The “zero trust” shift provides an example of the change. 

source: Aryaka 

Much of the rise in management complexity also can be explained by the shift to XaaS and high percentages of remote workers to support. Cost and complexity are issues, to be sure. But application performance is a big issue.

source: Aryaka 

And edge computing is among the primary tools for ensuring application performance.

PTC'22 — Executive Interview: Bill Barney, CEO & Founder, Turbidite

Turbidite will start with edge services in places like Cambodia, Philippines, Vietnam, Thailand, Guam. Eventually, hyperscale will follow. 

Is Moore’s Law Going to Become Obsolete?

It matters!

Edge Computing Matters as Cloud Architectures Shift

Nobody ever is surprised about cloud spending growth, either public, private or hybrid. Soon we’ll add edge computing to that list, though “edge computing” service revenues will remain a fraction of total cloud computing volume. 

As always, though, edge computing matters because it is a new revenue growth driver of some size, even if it initially represents less sales volume than does cloud computing. Also, if, as expected, edge becomes an integral part of the cloud computing fabric, there are strategic reasons for cloud computing services providers to incorporate edge capabilities. 

That is almost certainly the case if new use cases based on ultra-low latency and high computing requirements become popular. Even when the goal is ultra-low latency processing, the additional impact is to reduce data network transit. 

source: Researchgate 

That might be a corollary benefit of edge computing: reducing hairpinning. 

Right now, it is difficult to estimate actual retail value from edge computing services, as forecasts combine sales of infrastructure (hardware, software) with the value of edge computing services actually bought by customers, which distorts our understanding of the service provider market. 

Software as a service and infrastructure as a service generally lead in terms of end user sales volume, for cloud computing as a whole. At least for the moment, it seems that will be the likely pattern for edge computing as well. 

source: Gartner, ZDnet  

source: I-Scoop, Gartner