Bridging the gap between IT and Business through Big Data Analytics
The role of IT and networking specifically as we know it, is changing in the eyes of many Cisco stakeholders, primarily: (1) IT Experts and (2) Business Leaders. This combined with the changing expectations of the millennial generation will have a huge impact on the next market transition. Simplistically, network’s power to influence business outcomes cannot be underestimated. By combining network data with secondary sources of business critical information: cost, service management, user and compliance data and device diagnostics, companies like Cisco can empower businesses to optimize operations, profitability and decision-making. Concisely, using telemetry we can improve the productivity at any level of the organization.
Let’s have a look at the different personas who interact with disparate data sets and their respective wants / needs:
IT Experts want network data aggregated and combined with secondary sources of information (e.g. usage or user statistics, IP information, location analytics, application stack visibility) to make optimization decisions across the network value chain to minimize downtime and improve availability e.g. bandwidth optimization, critical vs. non-critical application differentiation and peak usage management. Not only this, IT experts want to aggregate data in a way that the network can then use algorithms to self-configure itself and implement machine learning to define or recommend corrective actions e.g. re-routing non-business critical applications through public routes rather than clogging bandwidth on business critical WAN links.
Business Leaders want network data aggregated and combined with business critical data (financial management information like application cost, fixed line costs, variable contract costs, labor costs and other productivity considerations) to gain visibility over networks’ gross profit and other cost contributions. They want to treat network information as a major cost-reduction driver in business decision making. Again, they want the data to first aggregate and then correlate automatically with optimal business targets in terms of spend, cost allocation and cost optimization. This correlation will then be the foundation stone towards the network re-configuring itself.
After understanding the motivations of the primary stakeholders, it is also important to empathize with the needs of the millennial generation who will be the next generation of IT users, consumers and business leaders. Millennials care about simplicity, easier to understand user interfaces, stories through data and experiences with technology that are consistent with the overall experience they have in today’s technology interactions e.g. Amazon or Netflix styled recommendation algorithms, Instagram or Facebook styled simplicity and Twitter or WhatsApp styled usability.
Use Case: Business Relevant Analytics using Networking Data
Keeping these user needs and perspectives in mind, when different data sets are correlated powerful outcomes are created. Let’s deep-dive on one specific use-case: Traditionally the network is viewed as a key infrastructure enabler in any enterprise. However, many enterprises either do not have a good network utilization visibility or do not have a good understanding of how the cost is associated with services provided and applications supported. This lack of visibility is especially true for cost break downs in terms of regions, sites, departments, applications and users. This is definitely a service gap for every large enterprise since it makes impossible to monetize the network services and solve problems like prioritized capacity planing, service roll-out and what-if scenarios modeling using machine learning and big data mining.
It is extremely difficult to gain these granular levels of insights for many enterprises and even if certain enterprises get to this level, it is proven to be a time and labor intensive task. This difficulty is due to the following reasons:
- First, the relevant data are fragmented or stored in silos in different vendor products. There is no end-to-end collection or storage. For example, the cost data might be stored in an IT Financial Management (ITFM) system, network utilization information might exist in a network-monitoring tool, and user identity might be captured in another AAA service.
- Second, the acquisition and analysis of the data needs in-depth networking tribal knowledge combined with programming and data analytics skills. This type of talent is either hard to find or very expensive.
- Third, the local optimization done on the data to define tangible actions is subjective and limited by the capability of the user. By aggregating data and combining it with relevant sources of information, automated insights and recommendations to reconfigure and improve the operational efficiency will be realized.
We’ve laid out the problem and the intricacies to consider. However here is how Product Managers at Cisco are driving innovation to make the network data more relevant for business outcomes: By utilizing data pattern recognition and machine learning we will be able to identify application to bandwidth utilization ratios e.g. 20% applications that account for 80% of the bandwidth, most used applications and reallocate the non-business critical applications to public or low priority links to ensure data fall-out/packet drop rate is minimal for critical traffic items as determined by customer defined cost policy. These re-allocations will form the basis to automate network reconfiguration by pushing customized policies in different points of the network based on real-time cost, network and other data accumulations. By comparing historical cost and utilization data it can be determined how to reroute traffic on paths with lower utilization at specific time periods to ensure minimal packet drop rates. The idea is for the network to automatically reconfigure itself based on cost thresholds, data utilization limits and peak time usage preferences. With this end-to-end view, we believe that we can uniquely optimize network traffic and save costs by moving non-business critical traffic on a different path compared to business critical traffic while allowing the network to rely on collected data to determine optimal transmission logic between links.
The power of data is just about to be unleashed.