CloudDB: Unified integrated data layer for cloudified networks

2016.06.12 By Xue Qingfeng

The proliferation of mobile Internet has created a competitive battlefront between telcos, over-the-top (OTT) content providers, and virtual operators. Traditional business models built on mobile, broadband, and traditional data services no longer satisfy customers’ fast-changing requirements. Equally, the decline of telcos’ numbers advantage when it comes to customer base is creating the dumb pipe scenario. So, they’re turning to service innovation for the answer.

5G services and the Internet of Things will soon be hitting communications networks hard in both requirements and the diversity of what people want. Telcos will need flexible network architecture that can scale down TTM to days or even hours. 

But, to match the speed and service diversity of IT companies, they’ll need innovative communications tech. They also need to learn how to better correlate and aggregate data from different network elements and service systems. Then, they can open up the goldmines data operators are sitting on, because siloed data prevents the monetization of this extremely valuable resource – an area that quickly needs resolving. 

Services and data: We're better off apart

With all major telcos having developed network virtualization strategies, a new round of cloudification and virtualization on telecom networks will come into being, bringing with it opportunities and challenges. Telcos are asking how agile networks can stop them competing with OTT players and instead get them cooperating to share new market opportunities. The best way to do this involves flexible network architecture that’s designed to better satisfy diverse requirements.

Core networks should be the first candidates in mobile communication networks for virtualization. Software decoupling and hardware generalization means that the virtualization process will cut TCO heavily. Reconstructing software architecture can separate applications and data, enhancing network flexibility to meet different service requirements and ensuring a lossless service experience. Once applications and data are separated, traditional network elements are separated into a standalone service layer (VNF or virtualized network function) and a data layer (database). In this type of set-up, the service layer only manages the service logic of applications, while the data layer handles the application data. This allows for greater service flexibility, rapid deployment, and network elasticity.

A standalone cloudified unified data layer meets the requirements of applications for data storage, data processing, and carrier-grade reliability, while cloudified networks with unified data layers also present a solution for data siloes. 

Unified data layers can integrate and aggregate subscriber data (HLR/HSS), session data (MME/IMS), policy data (PCRF), and a range of different IT systems data, so data can then be shared on the unified data layer between VNFs. Data analysis and data opening using the unified data layer also lets telcos carry out rapid data monetization with carrier-grade reliability. OTT and third parties can cooperate, enabling both sides to benefit.

Unity with Huawei CloudDB

CloudDB is a unified, open integrated database for cloudified networks. Built on Huawei's industry-leading SDM (subscriber data management) solution, it incorporates the ultra-high reliability of HLR/HSS, which is noted for its nine-year zero downtime. 

The next-gen evolution of the SDM database solution has the following core functions:

Multiple application data integration: CloudDB manages the subscriber and session data of a number of traditional core network elements, including HSS, UPCC, DRA, IMS and MME. It provides cloudified database services for traditional telecom IT systems and third-party applications, and supports a wide range of data types, such as profile, session, and unstructured data, as well as data models like relational, key-value, and JSON. The solution provides rapid online time for new applications and services thanks to a large number of data access interfaces, including LDAP, SQL, Reset, SOAP, and SPML; flexible data models; and the online configuration of service provisioning functionality.

Super-high capacity and high performance: CloudDB uses a fully distributed architecture design with non-polar linear expansion and theoretically unlimited capacity. The database capacity is lab-verified for up to 1 billion users, and the solution can process 1.6 million operations per second with 99 percent of data latency under 1 millisecond.

Ultra-high reliability: CloudDB ensures 99.9 percent general hardware reliability and 99.9999 percent software and data reliability. This means it can deliver data services with high availability, ensuring service continuity for applications. CloudDB has no single point of failure due to its system-wide distributed redundancy architecture that runs alongside rapid software and hardware fault detection and recovery capabilities. 

Automated seamless switching and takeover mechanisms ensure a lossless service experience, while multiple active datacenters are on hand for backup and disaster recovery. Automated data conflict decision-making mechanisms ensure data consistency among multiple active data centers if a fault occurs.

Ultimate levels of integration: CloudDB employs automatic data compression and storage, and allows for the mixed storage of memory and hard disks. The data dictionary automatically compresses and merges identical data, with flexibly defined policies identifying active and inactive data, and migrating inactive data from the memory to hard disks. Moreover, the wake time for hard disk data is down to the millisecond. Compared with an integrated SDM database, CloudDB raises integration by 40 percent, bringing substantial savings in storage and memory space.

Flexible data resiliency: CloudDB supports a data layer unified load balancer that automatically performs lossless elasticity, scalability, and data migration based on session loading and storage space conditions. With a multi-datacenter heterogeneous disaster recovery architecture, the automatic flex time of CloudDB’s databases is less than 10 minutes. Each database is independent, requiring no association between databases, which simplifies resource orchestration.

Flexible deployment: To meet different needs, CloudDB supports flexible data redundancy and data consistency policies based on different applications and different data property configurations, including strong consistency and final consistency. Data caching is deployed on the application side to reduce data access latency, and N-Way multi-site disaster recovery and All-Active multi-datacenter backup architecture are supported. Any datacenter can act as the primary data node for providing data services without read and write restrictions, which maximizes system resource utilization and reduces bandwidth load requirements.

Built-in data analysis and opening capabilities: CloudDB's built-in big data analysis engine can perform convergence analysis on the data it stores, and collect other network data using data federation technology, on which it can perform correlation analysis. The system can also perform historical data analysis on subscriber information collected from each network element, with data mining creating complete subscriber profiles. The system supports data opening, and provides a large number of API interfaces, which facilitates flexible and diverse new service development and slashes new-service TTM. 

The face of the future

Huawei is leading the industry charge to roll out cloud datacenters. It has been working with Telenet in Belgium, which is set to commercially launch a cloudified SDM solution based on CloudDB. At the same time, the Indonesian operator Telkomsel has performed much better since service innovation, including O&M and user experience, thanks to a unified data aggregation and open platform (Dynamic Subscriber Profile or DSP) based on the CloudDB solution. The platform aggregates and opens up subscriber information network-wide in real time. As well as software and hardware decoupling, network virtualization is also driving network architecture analysis and reconstruction. 

The unified integrated data layer can facilitate network de-layering, simplify network O&M, drive down TCO, and open and monetize data – for the cloudification and virtualization of traditional telecoms networks, the unified data layer is the face of the future. 

qrcode comm88 en

Mobile reading