In-memory analytics plugs real-time need

Business analytics based on in-memory computing will continue to gain traction even as companies try to make sense--in real-time--of the ever-increasing volume of data generated, noted an analyst.

Daniel Zoe-Jimenez, program manager of enterprise applications and information management at IDC Asia-Pacific, said that in-memory computing, which he defined as "accessing and analyzing data sets without it being written down in storage", has been in the market for almost a decade. This means that a product such as SAP's Hana appliance, which is based on in-memory computing technology, is not a new offering, he pointed out.

That said, he noted during a recent interview with ZDNet Asia that SAP's packaging of in-memory computing and preconfigured software as a data warehouse appliance, which was launched in December 2010, is the only one of its kind in the market right now and "quickly gaining traction" among customers.

SAP corroborated the IDC analyst's observations, saying that Hana is a "game-changing technology" and it is seeing very strong interest from key customers across the Asia-Pacific region.

"Our pipeline for Hana is already in triple-digit millions and we expect the appliance to generate significant revenues in the future," said Stephen Watts, SAP's president of Asia-Pacific and Japan, in an e-mail.

Zoe-Jimenez also observed more companies are starting to open up regional offices around the world. As this takes place, there's a need for business reports and predictive analyses to be delivered real-time from companies' data warehouses housed in their headquarters to frontline staff and managers without latency, he noted.

To address this need, data appliances are a good fit, he suggested. "The market has witnessed an uptake of [the] data warehouse appliance as an alternative delivery model due to the immediate benefits that it brings, including lower total cost of ownership, higher performance and simple to use.

"In addition, since the software is optimized and pre-integrated with the hardware, the deployment times are typically faster," he said.

Besides SAP's Hana appliance, Zoe-Jimenez also mentioned Oracle's Exadata Database Machine and IBM's Netezza products as offerings that are "gaining increased traction".

Big Blue forked out US$1.7 billion last year to acquire Netezza, pointing to the ease of installation as a way to quickly get analytics into the hands of business units such as sales, product development and human resources, according to an earlier report.

Cathy Huang, industry analyst for enterprise and services, ICT, Frost & Sullivan Asia-Pacific, noted that embedding business intelligence (BI) or analytic functionalities within the hardware "enables a unified BI architecture" that uses the same query engine, charting engine, metadata, administration model and application programming interface (API).

"More importantly, the single user interface enabled by the unified architecture provides a common interactive paradigm for all users' reporting and analysis need, which significantly helps to improve user adoption rates and manage IT support costs more effectively," she said in an e-mail.

Additionally, by accessing and analyzing data from inside the RAM (random access memory) instead of the storage disks, BI based on in-memory computing bypasses some challenges encountered by traditional disk-based analytics such as slow query response times, Huang pointed out. The disk-based cubes and warehouses are also costly and time-consuming to set up and maintain, the analyst added.

Predicting the future
Beyond quicker analysis and reporting times from one's data warehouse, however, Remus Lim, country manager of information management at IBM's software group, said that it is seeing companies trying to make sense of "data in motion". This, said the executive, referred to analyzing both structured and unstructured data such as videos, audio or social media content, before they are stored in the database.

"This provides consumers with real-time, data-in-motion analytics for [almost instantaneous] decision-making capabilities," he said in a phone interview.

Lim cited a hospital that made use of Big Blue's InfoSphere Streams product for real-time analytics to find correlations on physiological data streams such as blood pressure and body temperature. The report generated, which is up to 24 hours faster than typical practices, would then help with the early detection of potentially life-threatening conditions.

"Early intervention leads to lower patient morbidity and better long-term outcomes," he pointed out.

SAP's Watts also mentioned the company's "co-innovation program" with a milk processing company in Europe which is utilizing Hana to achieve real-time visibility into the latter's production line. In-memory computing can analyze the complex dependencies between different product lines, explained the executive, so if waste from one production line is accumulating, the company can see where to reroute the foodstuff to.

"For a company dealing in large quantities of perishable goods, this kind of instant insight is key to making the best use of raw materials," said Watts.

IDC's Zoe-Jimenez, however, urged companies not to be swept up by the buzz in-memory computing might be generating. Rather, they should consider carefully whether their business model requires such real-time analytical capabilities.

"It is clear that not all organizations require real-time technologies. Its adoption will depend on the end-user's needs and specific requirements and should be analyzed on a case-by-case basis," he surmised.