As service providers begin to roll out 5G POCs and initial deployments, there is no question that these cloud native platforms will be huge Capex and Opex investments on their part. These platforms should be able to handle diverse workloads such as telco, edge and enterprise applications. They must be efficient as well as robust from an operational standpoint. They must also serve the operator well for years. While it is never my intention to be hyperbolic or combative, the question is if Openstack is a viable 5G platform – when compared with hyperscaler clouds.
OpenStack as a platform for 5G deployments?
Longtime readers of this blog will remember various posts on Openstack as well as on the increasing footprint of hyperscalers (primarily AWS) in hosting telco clouds. https://www.vamsitalkstech.com/5g/six-areas-where-hyperscalers-telcos-should-collaborate-in-2021/
https://www.vamsitalkstech.com/5g/deep-dive-into-5g-core-architecture-in-the-cloud/
While most telco operators have existing private clouds that are based to some extent on OpenStack versions, the move even with their enterprise groups to the public clouds is undeniable. We have also covered the validated reference architecture approach followed by the likes of AWS not just for enterprise workloads but also for 5G.
- OpenStack needs to prove its stability and operational robustness – Many large telcos and M&E (Media and Entertainment) companies have deployed OpenStack over the last decade, run into operational headaches, dealt with many stability hurdles, and have largely given up on hosting mission-critical workloads. Specifically, with telco workloads, the upgrades of various OpenStack components as well as subsystems such as Ceph etc, have in the past introduced compatibility issues. These cause concerns ranging from high availability to operational cost to any unplanned downtime.
- The “5G ISV on OpenStack” challenge – Everything in Telco is an application. So this is simple..how many Telco ISVs actively test on OpenStack?
- The Automation Gap – I have spent tons of blogs arguing the point that telco clouds are essentially some hardware but mostly software. Then automation is key. So can OpenStack vendors provide closed-loop and end-to-end automation as highlighted in these posts – https://www.vamsitalkstech.com/5g/devops-for-5g-service-updates/
- Service Assurance – End to end Service Assurance is a key issue in 5G environments as Operators are now more integrators. 5G deployments will have everything in terms of network infra-appliances, VMware-based environments, custom solutions, Openstack environments, different clouds and K8s-based CNFs. The goal for Operators is to build platforms that can offer Service Assurance that covers everything from onboarding to monitoring to triaging issues
- Time to Market considerations – The real issue with Openstack as applied to 5G and services beyond 5G Core and RAN (e.g slicing) is that cobbling all of the above complex services and deploying ISV functions introduces delays in rolling out new services for business and individual consumers. The lack of service chaining and the need to bolt in 3rd party orchestration introduces delays in provisioning, commissioning and activation of new services.
Conclusion
Enterprise architecture groups within telcos have historically been somewhat disconnected within the larger CSP organization, but 5G calls for closer integration. While OpenStack has been a somewhat popular platform for enterprise deployments, 5G will be a challenge in the above five areas.