Yesterday I was rather disparaging about 5G. It’s only natural. Over the years I’ve watched the announcements of EDGE/3G/LTE and it’s always been the same story… it starts with a promise of infrastructure changes that will revolutionise business models and human engagement… and it ends with web browsing being a bit faster (which is nice). Surely 5G will be the same, it will improve bandwidth, it promises to finally improve latency (which would be really nice), but innovation is going to happen at the device/application level.
Maybe. But I went to a 5G panel session at MWC today. I’m a naturally cynical person, but the line-up was excellent with the CTOs of Ericsson, NTT, Telstra, and SKT, so it was worth a look.
I have to admit the vast majority was the usual; lots of talk about “platforms”, “disruption”, “innovation”, “driver-less cars”, “IoT”, etc. But two comments jumped out at me and suggested that 5G may be more interesting than I thought.
First – apparently it is massively cheaper for operators to rollout than 3G or LTE. This is the first time I’ve heard of the new infrastructure being cheaper. It implies that rollout will be faster, broader, that the operators will have money left over to invest in the interesting bits, and that the equipment manufacturers like Ericsson and Huawei will also have to come up with compelling on-top products to sell.
Second – re-configuration and provisioning sound like a core part of 5G. This opens up possibilities. Consider the impact that server virtualization and orchestration have had on web services and enterprise IT, consider the impact that DevOps is having on software production, and here we have the same mechanisms being introduced at the core of the network. It’s hard to say exactly what this will lead to (in the same way it was for server virtualization), but there’s clearly potential there.
So I’m a little excited about 5G. It started me thinking about dynamically re-configuring networks which automatically responded to application behaviour to optimise performance. Fanciful but cool. An absolute nightmare to test. But as I walk around MWC I see two things which feel like they are going to happen, are going to make test an order or magnitude more complicated, and are going to require the test tools industry to step-up a little from quibbling about the best way to structure an object repository. Systems will soon be too dynamic to verify behaviour by comparing results to an “expected result”. The application layer, the middle-ware layer, the OS layer, the hardware layer, the network layer, etc. are getting more-and-more blurred and effective testing is going to have to deal with that; not declare it “out of scope”. Exciting!
Antony Edwards is TestPlant’s CTO, leading the product vision, implementation, deployment, and ensuring customer success.
