How to optimize an enterprise data architecture with private cloud and multiple public cloud options? As the inexorable drive to cloud continues, telecommunications service providers (CSPs) around the world – often laggards in adopting disruptive technologies – are embracing virtualization. Not only that, but service providers have been deploying their own clouds, some developing IaaS […]
As the inexorable drive to cloud continues, telecommunications service providers (CSPs) around the world – often laggards in adopting disruptive technologies – are embracing virtualization. Not only that, but service providers have been deploying their own clouds, some developing IaaS offerings, and partnering with cloud native content providers like Netflix and Spotify to enhance core telco bundles. With the rollout of 5G and virtualized network infrastructure, the telco network cloud is adding robust enterprise networking offerings to the virtual portfolio, to the point that some CIOs are looking around and asking what isn’t virtualized yet, and why not?
This is all happening in the context of a global pandemic that has accelerated digital transformation and changed consumer behavior to the point that reliability and resilience are now primary considerations in selecting a communications provider. Connectivity has become too critical to people’s lives to be compromised for a relatively small amount of cost difference. More and more mature markets are adopting growth market practices of having multiple providers (where SIM-swapping has long been common practice to optimize cost) to assure connectivity. Enterprise end-customers too are looking to service providers to provide SLAs and contracted assurances of availability and uptime, which is adding to the pressure.
These dramatic, lurching changes are difficult to manage. Data has never been more important. Network, customer, finance, partner, and operational data all contribute to a comprehensive view of business performance and service delivery that can make the difference between the right strategy at the right time, and a decision that maybe shouldn’t have been made. Not only that, but broad-based integrated data is allowing telcos to train highly accurate AI models executing across the business, from predictive maintenance and intelligent anomaly detection on the network, to pre-emptive customer support and lifetime value optimization.
The data sources are many and varied, from the network itself to its management systems, through BSS including billing and care, and a marketing automation environment that has become a sub-architecture all to itself in recent years, combining multiple cloud services and applications with interfaces to product catalogue and campaign management solutions. These are in turn distributed across on-prem, private and public clouds, some visible at an infrastructure level to the enterprise architect and some buried in an OEM. Different departments in the organization have different strategies, and in many instances their own ‘shadow IT’. When IT was a Cap-Ex play, it was clear that the CIO should have ownership, especially given the requirement for landed hardware. The shift to cloud however has moved much of IT to Op-Ex, or subscription services, opening multiple opportunities across the enterprise for agile solution providers and LOB leaders alike. Pulling all of that together is the real challenge for the telco CIO.
Many service providers are making strategic cloud partnership decisions now, looking in most cases at tie-ups with Google, Amazon or Microsoft, each a strong Cloudera partner. Economies of scale become possible, both in terms of the cost profile that these providers can offer, and in terms of the kinds of commitment service providers can make in return. As Cloudera customers design their journey to cloud, many are choosing the Cloudera Data Platform (CDP) on one of these hyper-scaler clouds to take advantage of lower infrastructure costs.
However, there remain three major sticking points on the cloud journey. First, dependence on a single cloud vendor will create a vendor lock-in situation. To some service providers, it reminds them of the first few years of mobile in the late 1990s and early 2000s, when CSPs tended to choose a Network Equipment Provider (NEP) with whom to build the network, got locked in, and costs ballooned. Within five years of launching, almost every mobile operator in the world had moved to a hybrid network architecture.
The second issue is with regulatory, security and privacy concerns about moving certain workloads and certain data sets beyond the firewall, and in many cases out of the country. This particularly relates to PII data, but also to data from enterprise end-customers on their network that relates to financial, government or healthcare services. Even where hyper-scalers have points of presence within the jurisdiction, there has been some concern expressed by some service providers about certain services routing data outside of the jurisdiction for processing, particularly in AI. For those workloads, private cloud options are preferred, even where the costs may be higher.
The third issue is with the network cloud strategy of the service provider. The future shape of the enterprise business for the carrier is virtual: virtual network functions, running in dedicated virtual enterprise environments, possibly in a dedicated slice of a 5G network, and supporting edge workloads that penetrate deep into the service delivery infrastructure of that enterprise end-customer. The challenge for the telco is that the hyper-scalers see this opportunity also. Enabling the hyper-scalers in-market through offloading workloads to their clouds only emboldens them in encroaching upon traditional telco business. This is a difficult tightrope to walk.
There’s no getting away from the fact that service providers have to work with the hyper-scalers, for cost, ease-of-use, and time-to-value reasons. At the same time, defensive strategies require that service providers make sure there are off-ramps in their relationships, and that there are on-prem options for sensitive workloads. Having established the need for a hybrid data cloud strategy, the question becomes how to manage that environment. Separating the data workloads does not mean that data governance needs to be separated as well, causing fragmentation, inconsistency and redundancy; Cloudera CDP – and in particular the Shared Data Experience (SDX) – embeds comprehensive data security and governance into the platform that allows service providers to manage data and analytics in both private and multiple public clouds with a single consistent data context. SDX is a fundamental and integral part of Cloudera Data Platform architecture. Cloudera recognises that data engineers and analysts need to continuously identify relevant data sources and assess their value; SDX’s data catalog delivers just that capability. In addition, data lineage is persisted across those clouds, assisting in auditability, transparency and explainability which are increasingly important to prove compliance as well as for AI workloads.
Further, standards and policies around data privacy, security and user access can be unified across all data sources and applications, making sure that CISOs, Chief Privacy Officers, and Chief Compliance Officers can be confident that enterprise policy is being distributed to the different data assets and resources that the business relies upon. The Cloudera Trust Centre ensures that security and privacy are at the forefront of product development, keeping our technical teams up to speed on global regulations and standards.
This also means that the orchestration of enterprise data workloads is retained under the control of the service provider themselves, and is not reliant on one of the hyper-scaler environments, which would further complicate strategic telco options going forward. As new workloads are contemplated, and new data sources engaged, decisions can be made at that time about how to design the data environment based on relevant criteria, such as cost, sensitivity, and proximity to relevant other workloads.
As Cloudera continues to operationalize open-source Data and AI software in a cloud data architecture, telecommunications companies around the world are migrating their workloads to Cloudera CDP. Already, 80% of the world’s top 100 service providers run on Cloudera. It is uniquely suited to the demands of the industry, with use cases and user stories across the telecommunications landscape, from networks to care, finance to marketing, and aligning with complex ecosystems and platform businesses as they emerge.
Find out more at https://www.cloudera.com/solutions/telecommunications.html