03:05 PM
Connect Directly

Gauging The Benefits Of In-Memory Computing For Insurance

SAP has made a splash with its Hana in-memory database technology, but its utility for insurance will depend on carriers' need for speed.

In-memory computing has made it to the big leagues, in a manner of speaking, with the announcement by the National Basketball Association that it has adopted SAP's Hana in-memory database to let fans analyze NBA statistics. Fans will be able to run queries on every statistic going back to the league's origin in 1946.

Big Data's ImpactThe April 2013 digital issue of Insurance & Technology examines four key business areas -- marketing, product development, customer service and risk management -- where insurers are leveraging big data to optimize results. We also analyze the challenges insurers face in making the business case for investments in analytics and data management. To read more, download our April 2013 digital issue now.

That capability will no doubt give basketball geeks no end of fun, but it remains to be seen whether mainstream companies, including those in the insurance industry, can put in-memory database technology to profitable use.

In-memory technology, which relies on main memory for data storage, is on the rise as memory costs decline and companies look to process ever-larger amounts of data much faster than ever before, says Gartner analyst Donald Feinberg. "The slowest thing you can do on a computer today is read or write to a spinning disc," Feinberg told I&T. "When you look at processing time for an in-memory function, you're talking about orders of magnitude faster -- depending on what you're doing, it could be as much as 1,000 times faster."

SAP has gotten a jump on its competitors with Hana, which has expanded from a real-time analytics play to now powering the full SAP Business Suite of enterprise applications. "Other companies can do this but they haven't yet, so SAP has a big head start," Feinberg says. "We believe that over the next few years you'll see this technology coming out of Oracle, IBM and Microsoft."

Among the benefits of SAP Business Suite on Hana is that it accelerates routine reporting functions -- customers can run end-of-period financial reports in seconds or minutes instead of tying up a database with a batch run for hours, according to Ovum analysts Tony Baer, Madan Sheina, Nicole Engelbert, Carter Lusher and Warren Wilson, who published a note on Hana in January. But faster processing speed is the least of Hana's benefits, the Ovum analysts say.

"The in-memory architecture allows data views to be generated on-the-fly, a benefit that not only reduces database footprint and storage requirements, but also potentially simplifies the modeling and deployment of data and the design of analytics or other complementary applications that run atop Business Suite," their note states. "Hana's in-memory architecture also allows analytics to be embedded with transaction processing, enabling companies to become more agile. For instance, SAP customer John Deere achieved positive ROI on its Hana investment based solely on the benefits of implementing it for pricing optimization."

Gartner's Feinberg says Oracle's Exadata database appliance has been misleadingly touted as in-memory when in fact the product uses memory off the server board rather than on. Oracle has come around to calling Exadata's capability "database in memory," rather than in-memory database. The distinction is worth noting because of the significant difference in speed, Feinberg insists, but it may still leave companies wondering whether it matters. (Feinberg hastens to add that Oracle's Exalytics appliance uses true in-memory capabilities and competes with Hana as a standalone analytics application, but Oracle Business Suite doesn't use in-memory computing.)

"The question to a corporation is: If I'm running Exadata today, will it be fast enough that I won't need to convert to Hana?" Feinberg says. "In many situations the answer will be yes."

Insurers, especially top-tier carriers, should be asking similar questions, says Martina Conlon, an analyst with New York-based Novarica.

Conlon sees in-memory computing giving insurers three major advantages. "First, they can churn through greater amounts of information to derive insights quicker," she says. "There is also the potential to make better predictive models and to drive real-time business intelligence."

For example, Conlon sees the technology helping to drive underwriting earlier in the insurance life cycle. "Instead of underwriting occurring after a submission," she says, "some companies are performing abridged processes, whereby they're launching marketing campaigns and reaching out to certain prospects -- perhaps in certain geographies, performing a kind of pre-underwriting and understanding that they want that risk before they attempt to approach the prospect. In-memory technology will enable the injection of more information into that process, from social media, data bureau sources, etc., and develop a more comprehensive understanding of the risk and the prospect."

Gartner's Feinberg cites the examples of handling, analyzing costs and detecting fraud on insurance claims. "Ultimately, it's a matter of whether there's a business benefit for getting something that may take 30 minutes down to one minute," he says. "That's the question you have to ask."

Anthony O'Donnell has covered technology in the insurance industry since 2000, when he joined the editorial staff of Insurance & Technology. As an editor and reporter for I&T and the InformationWeek Financial Services of TechWeb he has written on all areas of information ... View Full Bio

1 of 2
Comment  | 
Print  | 
More Insights
Newest First  |  Oldest First  |  Threaded View
User Rank: Apprentice
3/28/2013 | 4:43:04 AM
re: Gauging The Benefits Of In-Memory Computing For Insurance
One of my favorite links for additional details on SAP HANA is Feel free to reach out to me directly if you have any questions.
User Rank: Apprentice
3/25/2013 | 1:05:10 PM
re: Gauging The Benefits Of In-Memory Computing For Insurance
I've been building PoC, of various kinds, using Postgres + R + PL/R for quite a while. Do that with SSD as primary storage and normalized data, and you get about as fast as you can. With the analytics called from the SQL. Is that real-time enough for you?

PL/R should work with any of the PG extensions, such as Greenplum; so long as they maintain the C language interface for UDF (R's hooks are in C, not R syntax).
User Rank: Apprentice
3/23/2013 | 4:41:07 PM
re: Gauging The Benefits Of In-Memory Computing For Insurance
SAP certainly did not invent in-memory database technology, and in this case it is NOT optimized for analytics. Companies like Kognitio have had in-memory analytical platforms in market since the early 1990s... Accelerating routine reporting is necessary, but not sufficient...

The key to differentiation in insurance and in all industries isn't reporting (like the simple stuff HANA can do with SAP BW), but rather the advanced analytics that can be accomplished with statistical algorithms like R - especially if it can be embedded and massively parallelized in the database.

SAP has added a great deal of marketing heft to the equation, but it is relatively immature, having an offering in place for less than 2 years.
James McGovern
James McGovern,
User Rank: Apprentice
3/23/2013 | 11:42:55 AM
re: Gauging The Benefits Of In-Memory Computing For Insurance
HP's Vertica technology not only accelerates based on in-memory but also via columnar operations. For insurers that want lower costs and higher performance, drop me a note at
Michael Dwyer
Michael Dwyer,
User Rank: Apprentice
3/22/2013 | 3:31:04 PM
re: Gauging The Benefits Of In-Memory Computing For Insurance
I think Hana has taken an interesting niche play here but I'm not sure I agree that they are "the first" to seriously move into in-memory database computing. Also, while I understand the challenges found with Hadoop-style solutions (real-time ad-hoc queries), I'm dubious that a combined Oracle/SQL-Server transactional system (both of which can do significant in-memory computing) combined with a Hadoop-style solution (probably in the cloud) doesn't make the most sense here. I realize you can probably mix Hadoop and Hana but I'm not sure why you would.
User Rank: Apprentice
3/21/2013 | 5:05:40 PM
re: Gauging The Benefits Of In-Memory Computing For Insurance
Recommend readers look at TIBCO Active Spaces for true in-memory computing with full redundancy across physical servers.
Register for Insurance & Technology Newsletters
White Papers
Current Issue
Insurance & Technology Digital Issue
Innovation? Check. Core modernization? Check. Security? Check. Today's insurance IT challenges don't stump this year's Elite 8.