Thanks to all who joined us for The Modern Telco is Open, Part 3 — Intelligent Virtualized Operations and Automation for Telco, sponsored by Red Hat and Intel. During the webinar, we looked at how automation tools — like Red Hat Ansible Automation and Red Hat CloudForms — use a playbook-driven and open source API approach to streamline complex operational processes for for network automation, hybrid cloud, software-defined networking/network functions virtualization (SDN/NFV), and Internet of Things (IoT) applications. After the webinar, we took questions from the audience, but unfortunately ran out of time before we could get to all the questions. Below is the full “The Modern Telco is Open, Part 3” Q&A.
Looking back at Slide #7 – Journey to an Open Telco Architecture – What is the expected timeline for this journey?
If we look at the left side of the evolution, parts one and two — the modernization of the infrastructure — have been occurring for the past four to five years, and so many of our operators have worked to build a common cloud platform to deploy virtualized applications and network functions. If we look at the next part of the journey, the modernization of applications and services will evolve at a similar timeline as the evolution of 5G architecture, which will enable MEC applications and services. As with the LTE architecture, one would expect that this part from section three to section four will take five to eight more years as the operators will need to monetize the services to ensure a good ROI on their 5G/MEC investments from the innovative new services in order to build out this infrastructure.
What aspects of their [RedHat’s] business are service providers/telcos currently focused on from an automation perspective?
A lot of it has to do with it getting operational in a centralized environment — getting the basic services up and running. In a multi-vendor world of virtualized apps and capabilities, the key issue is how do I standardize bringing new services to life — some people call it VNF [virtual network functions] on-boarding. The second piece on their minds is that as we evolve to a world containerized microservices is: how do these things interact in a containerized and VM [virtual machine]-based world, and how do I operationalize that in a consistent fashion? That is what is hot at the moment, we are seeing some work starting in the data management platform area as well, but it is still early for most operators.
To expand on that, we have moved very quickly in last year or so on being able to automate networks devices — a nice thing to have — and it’s now a mandatory function across the organization. As we go through that journey and are now moving that proverbial needle into the next phase, it was fine to get network devices to push configs and standardize at the network infrastructure layer. Now we are seeing how we tap into the ephemeral state data (data that only lasts a short amount of time) and how we can leverage that. Not just to validate that configuration changes are successful in completion, but to use the ephemeral data to make better decisions about how we deliver services based on that information as we move to that containerized and microservices world.
What regions are we seeing progress in automation of the NFV and SDN use cases for mobile or business services?
In almost all regions — from APAC to Australia and Japan — they are deploying both mobile and business services. As we bring it back to Europe, we are also seeing an advanced approach, including with SDN in these deployments. The next regions that will go and drive it will be in the EMEA, India, and LATAM markets as they come on strong and need to keep up. In North America, both Canada and the U.S. are deploying mobile and business services to take advantage of automation with both SDN/NFV. Across the board, we are seeing it in all regions, definitely in the Middle East, India, and Africa as well. It is a fundamental technology that enables organizations to deliver on their services and meet their SLAs.
Where do analytics and service assurance fit into service provider automation plans?
What we have seen are that the tools we had previously used for physical networks were pretty much a polling type of architecture, in that you will poll the devices to figure out what was going on and get a view for the operators of the red/green lights from all these devices.
As we move to a new set of capabilities with virtualization, SDN, and a new set of technologies, a whole set of new efforts has begun for this virtualized infrastructure to bring service assurance telemetry in order to take the information out of the infrastructure and use it in an event-based mechanism. With this approach we can take advantage of only the data we need, apply the analytics to it along with data science algorithms, and use this data for self organizing networks (SON) to control radios or the WAN [wide area network] on the fly.
As we take this out of the platform, operators can apply analytics and computer science algorithms to improve their service capabilities and SLAs. With this type of information in hand, operators can use it to bring better value to their customers. Much in the way we talk about taking and leveraging automation platforms to unify operations from a configuration perspective, we can use that same platform to expand on getting telemetry from the infrastructure cloud and network — and not just go dump all the data from the box every 15 minutes. We get more intelligent about the data we get, and we can send that data to a broad set of systems.
One does feel bad for the folks that still have to sort through all that network data. It would be better for them to use automation platforms to provide that first level filter, push that information into the systems, and normalize it so that the operations and business teams can take advantage of it.
NFV-MANO related question… Any example of VNFM integration in terms of Ansible playbooks? I do have a case where I have playbooks to install application and components to OpenStack infrastructure created with Heat templates. Now I would need to created VNF descriptions (TOSCA templates) to create VNF with VNFM (like OpenStack Tacker) but I would like to re-use already existing playbooks for application part. I suppose this is possible. Can you share any good practices?
As part of our Virtual Central Office demonstration efforts, we used Ansible playbooks to deploy OpenStack, as well as OpenDaylight, and the associated VNFs to enable both virtualized CPE and consumer services.
In addition , these playbooks are available on github via the OPNFV VCO demo website as well.