ChannelLife New Zealand - Industry insider news for technology resellers
Story image
Expert opinion: Cisco and IBM adopt hybrid analytics architecture to drive IoT growth
Wed, 8th Jun 2016
FYI, this story is more than a year old

The new Internet of Things (IoT) partnership between Cisco and IBM promises to enhance the companies' IoT offerings by bringing edge analytics to IBM and centralised analytics to Cisco. Integration of both types of analytics in a hybrid analysis platform is critical to customers using an evolutionary IoT strategy, an increasingly popular approach that is essential to unlocking the potential of IoT. The partnership will accelerate each company's IoT revenue growth and will drive overall IoT market growth as the role of each type of analytics becomes clearer.

For IBM, the availability of edge analytics lowers a barrier of entry to IoT for customers. With edge analytics costs are lower because not all data need be transported, stored and analysed in centralised locations or the cloud. Customers can build IoT applications that do not use centralised analytics at all, reducing costs and providing a stepping-stone to larger systems that use centralised analytics. Partnering enables Cisco to strengthen its IoT offering by providing a version of Watson analytics on networking and server hardware and easily integrating Cisco applications with Watson cloud-based analytics.

For customers, the partnership promises to alleviate a past IoT conundrum: where to start with analytics. The companies message that customers can now have analytics both ways. If and when combined with centralised analytics such as IBM's Watson, edge analytics can result in a more powerful, flexible and cost-effective hybrid IoT analytic capability, relative to a centralised platform alone.

For IBM and Cisco, the integration greatly improves their analytics offerings, strengthens their IoT strategies and promises to increase their respective IoT revenues. The agreement simplifies the construction of IoT solutions that integrate the capabilities of both companies; company spokespeople have said many of their 25,000 mutual customers had been asking for that capability.

Edge and cloud analytics: You can have one without the other

Centralised, usually cloud-based analytics and edge analytics have different purposes and different cost structures. Centralised analytics aims at discovery of new value in big data repositories, while edge analytics aims at rapid response to incoming data.

Centralised analytics take place in the cloud or in company data centers. This incurs costs for data transmission, storage and processing as well as for sophisticated personnel (i.e., data scientists) to prepare the data and perform analyses. The centralised approach lies at the heart of what Technology Business Research (TBR) calls “big IoT.” Big IoT follows the same assumptions as other big data applications: There is strategic business value in large data aggregations, which companies can discover through analytics and cognitive computing. Centralised analytics is driven by the need to discover these new insights. However, it is not required for every IoT implementation.

Edge analytics is driven by the need to act on incoming data. Compared with centralised analytics, it is tactical and less expensive. It is implemented on gateways, routers, switches and near-the-edge servers to apply insights about incoming IoT data to drive rapid actions by devices or people. Based on edge analysis, summary data may be transmitted to a central location, but in any case, data transmission, storage and processing costs are lower than with centralised analytics.

Edge analysis is not looking for insight; it is looking for actionable data patterns. The patterns are dictated by rules, and the rules can be written based on domain knowledge or can be generated by centralised analytics. IBM would prefer centralised Watson generates insights into data and embodies these insights into rules to be run by Watson's edge component, but use of centralised Watson can be deferred or even omitted entirely.

The hybrid IoT system IBM and Cisco are promoting with this announcement integrates centralised and edge analysis, but systems can run either entirely at the edge or entirely centrally. If the rules for the edge are generated by domain experts, or from separate analyses of other bodies of data, then no centralised analysis is necessary. This is the model for most point projects in “little IoT,” the fastest-growing segment of the IoT market. TBR estimates little IoT accounts for 80% of IoT projects being initiated.

On the other hand, analysis for insight and action can be performed centrally, without edge analytics, with all data brought to the cloud or data center. This is the model for entirely cloud-based IoT, like that promoted by Amazon Web Services, and is the model IBM has promoted until now. Cisco favored its own fog computing edge analytics but also featured how external central analytics can be integrated.

Hybrid IoT analytics: Lower costs and increased ROI

Much of the excitement surrounding IoT in 2015 was about extracting value from the big data IoT generates, and IBM was at the forefront of generating that excitement. Vendor companies emphasised strategic transformation of companies implementing IoT. What IBM and other vendors pursuing big IoT have found over the last year, however, is that many customers see IoT as strategic but are hesitant to commit to large-scale IoT projects. Customers interested in strategic transformational IoT projects often experiment with proofs of concept or tactical implementations as they evaluate their commitments to big IoT. This is a costly and time-consuming process for vendors and resulted in slower IoT growth rates than expected by companies counting on big IoT.

Customers see edge-based IoT as safer, with greater assurance of a more rapidly realised ROI. The problem with strategic IoT based on centralised analysis is the cost of big IoT is large and the benefit is less certain than that for little IoT. The benefit comes from the insights generated by centralised big data analytics, and the value of those insights cannot be known until they are generated and used. With little IoT, ROI may be more modest, but it can be estimated with more confidence. Because of the lower cost of little IoT, the risk is smaller.

Partnering with Cisco enables IBM to offer its customers a smooth path between edge-based IoT and centralised big IoT. At the same time, performing analysis at the edge reduces the cost of centralised analysis by reducing the amount of data transmitted, stored and processed. Together, these two benefits smooth customers' transitions and lower costs, making it easier for customers to move from little IoT to big IoT. TBR estimates IBM's CY15 IoT revenue was approximately $3 billion, a gain of 5% year-to-year. With the contribution of hybrid IoT analytics and its resulting pull-through of other products and services, IBM's IoT business could reach north of $3.5 billion for CY16, according to TBR estimates.

Cisco can now offer its customers an easy bridge to the power of centralised analytics. Cisco customers had been using edge analytics unless they created their own integration with centralised analysis platforms; the Cisco platform relied on the company's fog computing edge analytics as its analytics platform. TBR estimates Cisco's CY15 IoT revenue at approximately $1.7 billion, a 14% gain over CY14. With the addition of Watson analytics and its respective pull-through of other Cisco IoT components, TBR estimates Cisco's IoT business stands to hit $2 billion for CY16.

Hybrid IoT analytics is a winner for IBM and Cisco

Combining the IBM Watson brand with Cisco's infrastructure and fog computing capabilities is attractive to IoT customers. IBM is a leading provider of IoT analytics software and professional services, and partnering with Cisco will ease major adoption hurdles by providing a more unified solution across IBM and Cisco assets. These assets have been tested and validated and will help customers reduce implementation time and costs.

The technology paradigm underpinning this partnership is not new. Vendors such as Microsoft, Dell and GE have edge and centralised analytics and computing capabilities for IoT that provide similar cost savings and time benefits. However, the IBM-Cisco partnership will create proofs of concept and increase customer awareness of hybrid IoT. It also will allow IBM to put Watson in the driver's seat in more situations.

For Cisco, the partnership adds a vital third component to its IoT System, bringing in a powerful centralised analytics engine to pair with its platform (Jasper) and connectivity plays. TBR believes the partnership ultimately strengthens Cisco's class revenue driver — edge routing — while complementing its IoT-based software and security revenue opportunities.

The partnership allows the companies to approach IoT with joint sales and go-to-market initiatives where it is most beneficial for both vendors. Because Cisco and IBM have separate cost structures under this agreement, we expect IBM to pursue partnerships with additional vendors, such as Juniper, to spur IoT market growth. At the same time, Cisco will explore tighter integration with other central analytics vendors such as Microsoft and GE.

We believe the partnership will also help IBM better differentiate against IoT analytics and cloud competitors such as Amazon Web Services (AWS), Google and Microsoft by keeping analytics and Watson at the center of its cloud and IoT portfolios. With AWS established as the dominant leader in public cloud infrastructure, IBM is working to promote its hybrid IT, cognitive and cloud services abilities as part of a focused portfolio, which allows clients to use cloud-enabled technology to achieve industry-specific business outcomes. AWS struggles to provide comparable business solutions to enterprise customers because it messages a broader set of solutions, typically at the developer level. IBM's and Cisco's industry focus and expertise will resonate well with IoT customers that typically require customisation and will ultimately provide each vendor with an additional avenue to promote their cloud solutions.

Hybrid IoT is the emerging architecture for the broader industry

While not all implementations of IoT will have both edge and centralised analytic components, the availability of both is a benefit to customers and a necessity for vendors focused on providing a comprehensive IoT solution. IBM and Cisco had hybrid solutions prior to this partnership, but they required more effort by customers. With this announcement, both companies are offering a more complete and flexible solution. However, they are not alone. Dell, GE, Microsoft and SAP also offer centralised and edge analytics. AWS remains the only major player without significant direct assets in this capability.

As this hybrid approach becomes more familiar, it will accelerate the trend toward evolutionary IoT, where companies move to their transformative IoT goals through steps, each one providing benefits to the company. TBR believes this is the future of commercial IoT.

Article by Technology Business Research Incorporated (​Ezra Gottheil, principal analyst, Andrew Smith, analyst, Patrick Filkins, analyst, Molly Gallaher Boddy, research analyst)