YOU ARE AT:5GEdge virtualization: Overcoming deployment hurdles

Edge virtualization: Overcoming deployment hurdles

While it has become somewhat commonplace to take for granted that network operators are working hard on virtualizing aspects of their networks, it is important to remember that the industry is still in the early days of shifting from a physical networking environment to a virtual networking environment. To this end, if progress is to be made, it is worth examining some of the key virtualization deployment hurdles that continue to stand in the way of widespread implementation of VNFs in operator networks.

While the following is not a comphrhensive list, the two points outlined below are commonly cited by network operators as key challenges to be overcome.

VNF footprint(s)

It is a commonly accepted notion that VNFs are less expensive to purchase, deploy and operate than PNFs.  At the same time, the resources needed to run workloads in a telecom grade environment can be substantial. As such, in some cases, operators are finding that the compute resources required to run VNFs in their network can actually work against the claim that virtualizing network elements helps operators to cut their network OpEx.

Beyond the raw compute power to run some VNFs, operators are also finding that, as with PNFs, some level of customization can be required to make the VNF a good fit for their network.  In many cases, changes to one aspect of the VNF can have downstream implications on its performance. To this end, the need to customized VNFs, while theoretically easier in a software environment (especially in an open source software environment), can take quite a bit of time, and cause delays in the ability to execute on a virtualized network implementation.

For these reasons, operators are looking to their supplier ecosystem for innovative solutions to these problems. Two vectors of product development that operators are keen to see progress on can be summarized as:

  • Compute resource optimization – the ability to follow Moore’s law as applied to VNF development can be a powerful tool in helping to reduce the number of cores needed to run a VNF in a production environment.
  • Microservices – Beyond reduction in raw computing resources needed to run a VNF, operators are also interested in using microservices to compartmentalize functions within the VNF. In this sense, it could then become possible to make changes – add or remove features – within the VNF without changing the entire structure of the VNF.  What’s more, using microservices also allows for increasingly agile design methods to reduce the cycle times required to change and/or optimize particular aspects of the VNF.

Service assurance – testing and diagnostics

As networks become increasingly virtualized, it can become substantially more complex to run testing and diagnostics, and as a result, perform service assurance.  At a basic level, virtualizing network functions allows for the possibility of multiple suppliers to be contributing features to a VNF, or a system of VNFs. As this happens, the instrumentation requirements needed to run the required testing and diagnostics can increase. Going further, because development cycles for VNFs are becoming more agile in nature, there is a need for continuous testing throughout the design process. This also holds true in a production environment where the dynamic nature of virtualized networks can also create this need for continuous testing. In turn, this also contributes to the VNF footprint issue described above.

Similarly, as service delivery over virtualized infrastructure becomes more complex – with massive IoT, ULL and increasingly tailored video content all bringing different service delivery requirements – testing and assurance requirements will become even more critical. Most telecom-grade OSS/BSS vendors offer comprehensive testing and assurance solutions, as do testing specialists. However, adopting these solutions without increasing the level of complexity involved in running  a virtual network remains a challenge that operators are working through.

For additional reading on the subject:

ABOUT AUTHOR

Jason Marcheck
Jason Marcheck
Founder and principal analyst at Layne Bridge and Associates. Jason is a 20 year veteran ICT industry analyst covering 5G, IoT, cloud and virtualization strategies for clients across a range of vertical industries. Prior to founding Layne Bridge, Jason worked for 14 years at Current Analysis/GlobalData as a research leader and consulting director.