Web Explosion

It does not take a rocket scientist to realize that we are in the midst of an online revolution. This online revolution mostly has centered on consumer-centric (b2c) online businesses - such as the retail, travel and hospitality industries - that have witnessed meteoric year-to-year growth. AMR Research estimates a $208 billion market potential for online businesses by 2007.

Yet more recently, another trend has emerged. Enterprises in multiple industries, including financial services, have begun to leverage the online channel for both intra-enterprise as well business-to-business (b2b) applications. These applications support core functions such as client services, sales and information delivery. Such an online strategy enables even smaller firms to function as large, global organizations, while enabling the larger enterprises to become more nimble and deliver services without bureaucracy.

Going Online: Boon for the Business; Bane for IT?

Web applications provide an ideal channel for businesses such as financial services and investment firms to enhance their abilities to serve their customers and partners better, and consequently, drive more revenue. For instance, retail brokerage has been an area in which the online model has been embraced for several years now. Today, the range of services offered cater to both casual traders as well as power traders, who now have access to sophisticated online tools to wade through complex markets such as options and futures. On the institutional side, broker-dealers can utilize the Web as a medium to service their buy-side clients through portal applications. To a great extent, this can alleviate the pain that a typical client-service desk encounters in providing real-time order positions and trade confirmations to its customers. Further, the advent of regulations such as MiFID has elevated the levels of pre-trade and post-trade data transparency required. Here, again, the Web serves as an ideal platform to publish information to clients and ensure fair execution of trades.

These application scenarios illustrate how the online service delivery model is based on three foundational tenets:

Personalization. A user is made to feel that the portal or the Web page has been designed and tailored to his or her specific behavioral needs.

Intimacy. It is often a comforting feeling for a user to know that there is a resource that provides up-to-date information on a 24/7/365 basis, creating a stronger bond between the customer and the service provider.

Scale. A Web application enables a business to cater to customers that it does not see, hear or even know exist. Consequently, thousands of customers in increasingly far-flung locations can be reached and serviced in a scalable manner.

But, while business executives reap the benefits of serving customers, partners and employees via the online model, their IT counterparts have the unenviable task of building, deploying and managing these applications in a scalable and reliable manner. The fact that only 14 percent of newly developed Web applications meet response time requirements (according to Gartner research) comes as little surprise, as the infrastructure part of the Web application equation has not been adequately addressed until now.

IT groups face a litany of challenges when it comes to deploying large-scale online applications. First, these applications have to be designed and deployed to handle and scale to support unpredictable loads, where the duration and nature of client access is unknown. For instance, poor response times - or even worse, the unavailability of brokerage Web sites operating during market open and close hours - often is a consequence of infrastructure deficiencies. Further, the volume of data, especially dynamic or fast-changing data, that needs to be managed within online environments such as portals that service buy-side clients or publish pre-trade/post-trade data for compliance can cripple most infrastructures.

To make matters more complex, end users expect enhanced or, at a minimum, equal levels of service with online applications as with off-line alternatives. What a small number of employees once managed physically now must be delivered automatically via the Web without any degradation in the quality of service. Barriers to entry are low in the online world, and as a consequence, IT teams are challenged with deploying more and more innovative, value-added services to entice customers and prevent them from defecting to competitors.

Enterprise Data Fabric: Building a Strong Foundation

According to Meta Group, approximately 60 percent to 70 percent of the latency and scalability problems of online infrastructures exist outside of the application tier and reside in the data-sources layer. Given this statistic, it is obvious that a robust data platform is essential to ensure a pleasant, consistent and reliable user experience with online applications. Industry pioneers are defining an Enterprise Data Fabric (EDF) as a solution to address this need. EDF is a pervasive and responsive information infrastructure that manages data predominantly in memory and across distributed nodes to enable the delivery of the right kind of data at extremely low latencies to handle transactions, analytics and decision support in online applications.

The relevance of an EDF as an infrastructure piece stems from its ability to offer speed, high availability and real-time intelligence, all of which have significant bearing on end user satisfaction and customer retention.

- Speed/Performance. By caching data in memory (RAM) and close to the point of use, an EDF guarantees blazing performance and scalability by eliminating disk input/output. Furthermore, by pooling memory across distributed hardware nodes, the RAM limitations of a single node are overcome. Hence, large volumes of frequently accessed data - such as securities reference data and user profile information - or transient entities such as user sessions can be held in the EDF layer to ensure instantaneous data access. Data can be held in multiple formats - Java/C++/C# objects, XML documents or relational tables - so that applications can consume data in a format of choice without the need for expensive data transformations. Such an approach greatly improves the performance of trading and client-service portals.

- High Availability. One of the major challenges with online applications is the ability to guarantee 100 percent uptime. An EDF provides sophisticated in-memory data replication that prevents data loss even if applications fail. Additionally, data that is held in the EDF can be persisted to disk or written to a system of record such as a database synchronously or asynchronously. Such a strategy is far less expensive and more efficient compared to failover and recovery models currently used in online infrastructures.

- Real-Time Intelligence. Another relevant feature offered by an EDF is real-time data analytics. Dynamic entities - such as market data streams and order flows - can be monitored, analyzed and queried on a continuous basis. Such a capability can be used by financial services firms to provide intelligent updates about order status, margin violations or compliance irregularities to their retail as well as institutional investors. With an EDF, online users can predefine patterns of interest to be analyzed, or add scenarios or queries dynamically in real time.

Foundational Principles

An EDF strategy for online infrastructures is anchored on the following salient principles:

- Address performance and scalability at an architecture level. From an application-development and maintenance standpoint, an EDF approach is more cost-effective than constantly patching a fundamentally flawed infrastructure, which is what most IT organizations typically must do.

- Manage data close to the point of use. In-memory data caching enhances performance and enables the delivery of relevant information to an online user in real time.

- Manage data as a dynamic entity. The delivery of intelligent, personalized information to users in a scalable manner is accomplished through the distributed topology of an EDF coupled with the ability to continuously query and analyze data.

Online business models are information-centric in nature, wherein data architectures are absolutely critical. As Web-based channels become an increasingly significant source of revenue for enterprises, an EDF-based data strategy becomes essential to guarantee a pleasant online customer experience, drive more transactions and increase revenues.

About the Author

Bharath Rangarajan, Director of Product Marketing, GemStone Systems

As director of product marketing for GemStone Systems, Bharath Rangarajan oversees product positioning and market strategy for the GemFire product line. He has more than seven years of experience in enterprise software, dealing with data management issues in functional areas such as EAI, B2B collaboration and supply-chain management.

Rangarajan has led technical and product teams at i2 Technologies, SeeBeyond Corp. and Candle Corp., among other companies.