This week, I had the great pleasure of attending Light Reading’s Big Telecom Event in Chicago, including several of the breakout sessions and panel discussions. As expected, virtualization, NFV, VNF, and SDN were the key discussion topics generating the greatest attention. In particular, attendees and participants wanted to know how network virtualization is bringing new innovation to the telco industry.
Among all the talks, there was one topic that especially caught my attention; in a virtualized world, are SLAs and five 9s still relevant? Even further, why not three 9s? Does this new generation of users even care about drop calls? Valid questions, but something just did not seem right. After all, the discussion about network innovation and all the advances that virtualization is making in the telco space, it seems that we reverted back to the old, familiar measuring stick (five 9s, drop calls, etc.).
Many of us are a byproduct of the old big telcos, so the five 9s expectations run through our blood stream. However, that doesn’t necessarily mean they’re still relevant in the new virtualized telco world. Can I instantly post my new pic on Facebook? Can I play my YouTube video without buffering? After all, if I need to call someone I’ll use FaceTime or Skype (thank you very much, dialing pad that I can’t seem to find among my many apps). What if the traditional KPIs are no longer relevant, and the user-perceived experience is now the new measure of network performance?
As the conference progressed, it became increasingly apparent that network providers should focus on gradually innovating the performance measurement perception and language, in addition to innovating the network architecture. In other words, focus on demonstrating the perceived per user and even per application user experience instead of providing overly complicated and often regurgitated old telco network wide KPIs.
Oh, and I almost forgot about the three 9s. Isn’t this going in the wrong direction? With innovation comes expectation of improved performance. Why not “1” “double 0”? After all, as network virtualization innovators, this is exactly where we can be better than any legacy proprietary hardware solution. In a virtualized network, you should be able to achieve infinite redundancy. Yes, infinite. This should be true in a vertical multi-layer/multi-path redundancy, as well as a horizontal “always-on” geo-redundancy, with a virtual machine always available to replicate and pick-up traffic. I never seem to know “where” my pictures are stored in the cloud; they just always seem to be there. The same should be true for the telco clouds as well. We seem to use an old measuring stick to measure new innovations. In my honest opinion, virtualized telco networks should strive for a 100% user-perceived reliability and should enable data analytics capability that measure and showcase what really matters in this new data driven user model.