This is part three of a five-part series discussing how Juniper Networks is helping communication service providers (CSPs) partner, collaborate and succeed within the public cloud paradigm. See part two here.
With a blend of mission-critical legacy and virtualized services in production today, the path to cloud-native isn’t-trivial. Building upon a weathered metaphor, we’re not just “turning a big ship,” we’re building a new ship around existing passengers who expect an uninterrupted and high-quality service experience. Customers applaud innovation but are the first and loudest to object when services are impacted.
The journey’s complexity is relative to the starting point, but CSPs face several universal challenges when evolving to cloud-native.
- Investing in the future while preserving the past – After years of investing in virtualization, CSPs have developed expertise, processes and tools to deliver NFV infrastructure (NFVI) and mission-critical cloud services. Preserving the harnessed automation systems, CI/CD pipelines, preferred OpenStack and Kubernetes distributions and robust security policies is necessary during the cloud-native evolution. Maintaining the continuity of business operations and protecting the customer experience are critical aspects of this process. A strategy based upon “rip and replace” is simply not feasible.
- “Multicluster Sprawl” – The number of systems and services is increasing for the operations teams who manage them. Personnel previously responsible for a dozen devices are now expected to manage hundreds; this is merely the tip of the iceberg. Designed for multivendor interoperability, cloud-native microservices like 5G are deployed within their own cluster resulting in cluster- or Kube-sprawl – as RAN and edge clouds are distributed to tens-of-thousands of sites. Disaggregation and cloudification become an operational quagmire without software-defined control and automation that delivers a single point of operations.
- Protecting Centers of Data – Cloud workloads are ephemeral, spun up dynamically to match temporary load, throughput, latency or resiliency requirements before being spun down. With cloud-native, these workloads can be further deconstructed into microservices, which create a greater dependency on both network isolation and security policies to protect API calls and control traffic that was previously localized. If not implemented and managed correctly, these dynamic and auto-scaled workloads can present new attack surfaces for malicious actors.
- Development or Acquisition of Skills and Processes – CSPs are challenged to develop the people and processes needed to support the evolution to cloud-native. Beyond investments in personnel training, skills acquisition costs and software development, time is the most precious commodity as CSPs compete to hit an aggressive 5G market window. Hyperscalers offer a compelling option to achieve time to market, but CSPs must carefully weigh the near-term benefits against the long-term risks.
- Loss of Flexibility and Control – Proprietary and vendor-locked, first-generation virtualized clouds limited the flexibility and control of CSPs, who remained tied to vendor software release cycles and roadmaps and lacked the leverage to qualify alternative vendor solutions. Cloud-native offers multivendor freedom and economics, but if the cloud infrastructure, operations and management are outsourced to a hyperscaler, do these benefits become the currency to fund time to market?
- Observability and Troubleshooting – Cloud-native microservices are geographically distributed throughout the network. Control and data plane components that were once local to a single server are now connected across physical and logical networks making observability and troubleshooting more complex. Service quality monitoring and SLA assurance are equally difficult; NetOps staff rarely see network problems before customers, thereby impacting service quality and generating customer turnover.
Managing and mitigating these challenges requires a well-defined telco and edge cloud architecture. Ideally, it would build upon DevOps automation and embedded security, support all existing and future cloud applications and standardize on methodologies to support multivendor collaboration and hybrid cloud networking. It sounds challenging but Juniper has a long history of designing, building and deploying cloud solutions for operators like Deutsche Telekom and BT. With practical cloud expertise based on real-world deployments and runtime operational experience, Juniper brings valuable ‘know-how’ and an experienced professional services organization to any CSP cloud deployment.
To Build or Not to Build? That is the Question.
As hyperscalers have emerged with enticing telco services and drop-ship infrastructure, should CSPs rethink their strategy to build telco and edge clouds? Not a chance. Whether hosted in private or public clouds, the Federal Communications Commission (FCC) holds CSPs responsible for network outages, 911 interruptions, public safety systems, and digital surveillance (e.g., CALEA, wiretap). Combined with the stringent uptime and quality metrics for fixed and mobile voice services and telco workloads like BNG, 4G, 5G and IMS, CSPs are uniquely qualified to manage the rigors of a regulated communications industry.
Hyperscalers do offer CSPs welcome relief from time to market pressures and skills gaps. However, new competitive telco services for RAN, 5GC and private 5G, paired with expansive enterprise offerings, have CSPs reconsidering if hyperscalers are friend or foe.
In part four of our blog series, we’ll describe the rise of the hyperscaler and their evolving CSP relationships.
Solutions for Service Providers
Cloud-Native Contrail Networking