It is clear that operators are looking to the cloud as a new business channel and hope to fill any gaps left by the established cloud providers, such as Amazon and Google. The established providers, having had issues with disruption of service, leave some skeptical and looking to a cloud provider that may have closer ties to the network and a better grasp on security, scalability, QoS, service management and performance monitoring.
So for telcos infrastructure as a service (IaaS) is definitely the lowest-hanging fruit. Obvious expectations from telcos are that IaaS must be ubiquitous, always available, have low latency and have a high level of reliability and QoS.
This service, tapping a telco's connectivity infrastructure and skills, promises to deliver a new revenue opportunity that will compensate for the inevitable decline in telephony revenues, as millions of people across the globe start consuming virtual infrastructure across networks.
This article looks at the core requirements for IaaS, as identified by service providers that are rolling out IaaS services, along with the infrastructure capabilities that need to be in place to achieve those requirements.
Scale. Process automation and management tools need to support hundreds and even thousands of companies, and potentially millions of end-users, and also provide the highest levels of customer visibility, control and end-to-end assurance.
Converged network and IT capacity pool. Virtualization and service management tools need to span IT and network boundaries, so that the pool appears seamless and both servers and network work together to support end-to-end user and application SLAs.
Self-service and on-demand capacity. A customer portal that gives customers visibility into and control over "their" virtual IaaS environment. Process automation and management tools to enable customers to turn infrastructure resources up/down in software, on demand, without needing to dispatch a technician to provision more capacity is also critical.
High reliability and resilience. Automated distribution of applications across the virtualized infrastructure (LAN and WAN) for resilience and SLA management.
Integrated BSS/OSS. Automated IaaS management and operational processes to support flow-through service provisioning, rating, settlement, etc.
Service management. Integrated SLA and class-of-service management to support customer requirements for different application types, e.g., business-critical or non-business-critical. In the service management context convergence of eTOM and ITIL is a critical step, especially for telcos looking to successfully serve enterprise customers. The pressure is on operators to provide differentiated, profitable cloud services. A streamlined, automated BSS/OSS will play a pivotal role in enabling service providers to quickly create, deliver and monetize differentiated cloud services. The key underlying software requirements they need to provide flawless cloud services include:
Customer and business management layer. A CRM-like self-service and customer care solution is needed to manage customer expectations and needs. Both the customer and the customer service representative should have access to a broad range of service-level data, related usage rates and metering characteristics.
Revenue management. Billing and charging need to be linked directly to a mediation-like layer to capture all metered records and provide the necessary chargeback to individual or departmental levels. They also need to interface with the self-service engine to provide the necessary usage/chargeback information to users. The flexibility of pricing, complex bundling capability and embedded business intelligence to provide customized offers need to be critical features of a revenue management solution set.
Different charging/rating scenarios that need to be supported include:
- Usage-based consumption (duration, events)
- SLA violation calculations
- Storage (GB-month, million I/O requests)
- Bandwidth (public Internet in/outbound, same cloud, regional cloud)
- Computing (CPU hours, RAM hours, service units)
- Configurable server instance types
- Partner management, contract management to determine individual terms with each partner on a per-service basis
- Model contractual clauses in all aspects of the financial relationship including payment terms, disputes, taxation and currency requirements
- Commission calculations for value-added resellers and agents, as well as referral fees for referring partners
- Allocating SLA penalties across the value chain based on fault
- End-to-end subscriber management, which includes account relationship, terms, real-time balance management, customer hierarchies, subscription type, etc., should also be part of the revenue management solution set.
Cloud analytics. Analytics capability to provide reporting, analytics and data-mining capability. Data can be pulled from an existing CRM system, self-service portals, or from the resource management layer itself to gain insight into customer usage patterns, trend analysis of service consumption and changing demands, and capacity planning of existing resources, ensuring future cloud service requirements can be met.
Service management layer. A streamlined fulfillment system needs to feed requests from the self-service engine through the resource management platform to provide access to the required cloud services (storage, computing power).
The performance and service assurance layers need to manage top down from the application end-user's perspective things such as performance, availability and behavior patterns, which will help in SLA/QoS management for private clouds. Integrated SLA and class-of-service management need to support customer requirements for different application types, such as business-critical or non-business-critical.
Network resource management layer. The necessary hardware, software and services are required to create shareable computing power, development environments and applications. This layer also provides IT resource management, application lifecycle management, database management and usage metering.