Data Virtualization: the smart approach for bringing BSS data silos together

A typical telecom BSS deployment has multiple database instances, database technologies and data models. BSS data is spread out across many disparate data systems. Furthermore, many communication service providers (CSPs) are considering or already starting to move BSS applications into the cloud which may lead to a hybrid environment – the already complex BSS data landscape is going to get more complicated.

Data thriving consumer applications like Analytics and AI/ML need BSS data cumulatively and holistically, as opposed to just needing individual records from the data sources. It would be quite cumbersome for the data consumers to have point-to-point integration with multiple data sources, especially when this is not the primary task but a mandatory pre-requisite.

CSPs are very familiar with the concept of data repository and, in fact, many of them already have at least one. A data repository holds data from many sources either in the native format (data lake) or transformed format (data warehouse) at any scale. Stored data can be shared and analyzed throughout an organization.

A CSP may have multiple BSS data repositories. A unified view of data, without creating another data repository can be built with data virtualization. According to our ebrief, Data Virtualization: the smart approach for bringing BSS data silos together, data virtualization can play a pivotal role in enabling many use cases in BSS.

According to analyst firm Gartner, data virtualization is now a mature data integration style. Data virtualization and its vendors are getting included in their various reports. A 2020 report from market researchers, Reports and Data, projects the global data virtualization market to grow at a compound annual growth rate (CAGR) of 20.6 percent from USD 2.45 billion in 2019 to USD 10.87 billion in 2027. This and other similar reports and forecasts indicate that data virtualization is a mature option that can be utilized in the BSS domain as well.

What is data virtualization?

Data virtualization is a layer between data sources and data consumers, essentially a middleware. It abstracts data consumers from the nitty-gritty of data sources, but still allows a simplified, unified and integrated view of the data.

Data virtualization layer contains no source data, only the metadata (data location, format, access protocol etc.) required to access the applicable sources. The data remains in the data sources. When a data consumer submits a query, data virtualization layer collects the relevant data from the data sources optimally, performs the necessary join and transformation, creates an integrated view and delivers the results to the data consumer – all on the fly, without the data consumer knowing about the true location, format and protocol to access the data.

Data virtualization to integrate multiple BSS data repositories

There can be various circumstances that lead to multiple BSS data repositories. Having multiple data repositories may not be a conscious choice but one with pragmatic tradeoffs. And this is where data virtualization comes into the picture. Data virtualization can be deployed as a thin layer on the top of multiple data repositories. This layer will appear as a single, unified repository to all data consumers

Figure 1: Data virtualization integrating multiple BSS data repositories Source: Ericsson

There are many use cases which can be supported by data virtualization:

• Providing logical views of business entities without making physical copies of data: Data related to a business entity (e.g., customer, product offering, resource etc.) can be generated and processed by multiple applications. For example, a product offering is configured in the catalog application, ordered in the ordering application, charged in the charging application and billed in the billing application. The data virtualization layer can provide a holistic view of the product offering by collecting and combining all the data spread out in multiple systems.
• Exposing data to enterprises in domain specific language: Enterprises (customers and business partners) of a CSP may want to get access to the BSS data for their various in-house data consumers. These external data consumer applications would prefer to get data in the language which they understand better – a domain-specific language (DSL). The data virtualization layer can expose curated domain specific views, derived from the common native data, while hiding underlying complexities and technical details of BSS data models.
• Combining a variety of data repositories: There are many use cases like customer churn management, lifetime value prediction for enterprise customers etc. which fall under the BSS area, where data only from the BSS applications would not be sufficient. Data from other sources, such as, OSS (quality of services, SLA, social media (posts on different platforms) would also be needed. We can’t wish that all these data would be available in a single huge data repository. Since physical data consolidation would be inappropriate, data virtualization comes to rescue by integrating this variety of data sources.
• Enabling location agnostic data access to smooth the cloud migration journey: Migration to the cloud would create a hybrid environment where data is scattered throughout various on-premises and cloud data sources. Data virtualization layer can span across both cloud and on-premises data sources and can provide location agnostic data access to the data consumers.

Business drivers for data virtualization

The following benefits make a compelling business case for data virtualization:

• Access to all data: Expose all data needed by users, whenever they need it, on demand: enabling data driven development and decision making by getting insights faster.
• Increased productivity: Users spend more time analyzing and using data and less time searching, preparing and managing it.
• Modernization: Legacy systems can be swapped with modern applications or can be migrated to the cloud without affecting data consumers.
• Increased agility: Agility to on-board new data sources or retire existing ones. Rapid development to deliver new data sets for any project need.
• Cost optimization: Improves utilization of existing server and storage investment. Reducing number of data replications and hence saving on data storage and governance expenses.
• A consistent data governance: Centralized layer to define and enforce consistent data access and security policies across heterogenous data sources.

Bring data silos together via data virtualization

CSPs do not just want to store data to retain historical records, they also want to generate value by getting insights from the data. CSPs may have multiple data silos, and these silos need not be demolished, but can be integrated via data virtualization. Data virtualization can eliminate data barriers by providing easy yet secure location agnostic access to all data. It can help CSPs to use BSS data more widely, effectively, and efficiently in a faster and cost-effective manner.

At Ericsson, we employ data virtualization to fast-track data driven development while minimizing BSS data replication. Data virtualization takes care of most of the data integration aspects and abstracts out data consumers like AI/ML and Analytics from the BSS transformation journey.

Learn more

• For further info on the ebrief, please refer to Data Virtualization: the smart approach for bringing BSS data silos together
• For further info on the BSS to cloud journey, please refer to Futurum | BSS-to-cloud journey: Innovation across the digital value chain | Report
• For further info on driving the agility flywheel, please refer to Driving the agility flywheel: Journey to agile | Report