Google Groups
Subscribe to Software Outsourcing [ Hire Dedicated Group ]
Email:
Visit this group

Saturday, June 2, 2007

Building Interoperable Insurance Systems with .NET 3.0 Technologies

Building Interoperable Insurance Systems with .NET 3.0 Technologies


Mike Walker

Microsoft Corporation

December 2006

Applies to:
Microsoft .NET Framework 3.0

Summary: This white paper will use an insurance-industry scenario to demonstrate interoperability capabilities of the Microsoft platform. Using protocol-level standards alone is not enough; capturing the business side of the messaging transactions is key to making interoperability work for your business. This is true across all industries, not just insurance. (15 printed pages)
Contents

Introduction
Insurance-Industry Forces
Business Terms Used in This Document
Life-Insurance Policy Scenario
Architecture Overview
The Insurance Agent Policy System
The Insurance Carrier Systems
What Is the Value?
Conclusion
Resources
Introduction

The purpose of this white-paper series is to provide guidance around integration challenges.

Through this white paper, we will use an insurance-industry scenario to demonstrate interoperability capabilities of the Microsoft platform. Through maturity of many enterprises, we live in a world where there is more than one stack of technology. These platform stacks range from legacy mainframe-based COBOL or FORTRAN types of applications to the more modern solutions based on .NET, Mobile Systems, or Java—and everything in the middle. As a result, as enterprises have iterated through technologies and technology trends, there has been more than a few bandages applied to the various technologies.
Insurance Interop Series

This white paper will serve as a guide for architects who are facing integration challenges in the insurance industry. We will show you how to use Microsoft integration technologies to integrate disparate systems in your enterprise. Additionally, this document will provide pragmatic design guidance for building interoperable solutions using open standards such as WS-*. Additional documents in this series will include the following:

Architecture Overview to Building Interoperable Insurance Systems

Securing Insurance Solutions

Scaling and Operational Management

Deploying Enterprise Solutions

Developing Composite Applications

Technologies that will be covered include:

1. BizTalk 2006. The integration technology for this solution. The solution also uses the BizTalk business rules and workflow orchestration.
2. Windows Communication Foundation (WCF). The programming model to develop Web service messages and manage protocol-level communication by using the WS-* protocols.
3. Windows Workflow Foundation (WF). To create compelling workflows using smart-client technologies.
4. SQL Server 2005. The repository for all of the application and customer data.
5. Windows Server 2003. The server platform.

This scenario will give us a glimpse into the business process. Like many businesses, each insurance company has its own unique way of handling its process. However, there are some similarities that these businesses share at the platform level. The purpose here is to demonstrate that there is a way to leverage these common platform services to build Service-Oriented Architectures (SOA) giving an organization more agility with the processes that differentiate their specific business.
Insurance-Industry Forces

In the insurance industry, there are many technologies at play, ranging from mainframe to UNIX to Windows. With this wide range of platform technologies, it is increasingly difficult to manage and operate while trying to be agile in an ever-changing financial market. For years, organizations have been building and buying technologies to meet these needs. Interoperability has become a necessary evil after the solution has been built and/or implemented. This has left us with point-to-point integrations that address very specific problems only at the application or system level, but not at the business-function level.

Figure 1. The result of point-to-point integrations

If care is not taken, point-to-point integrations over many years result in:

* IT portfolio management becoming unmanageable, given the duplication of systems, multiple variations of integrations, management of dependencies of applications, and so forth.
* Increased cost of IT systems, dramatically rising because of the number of custom integrations.
* Loss of agility, because development of systems is slowed significantly as a result of increased code complexity, limited reusability, and lack of standardization in the enterprise.

So, what does this mean to many insurance carriers? It means that interoperability is at critical importance—not only as an efficiency issue, but also as a competitive differentiator. In these days of modern competition, companies must increase the return on investment (ROI) of their IT systems by streamlining the processes and becoming more agile to stay competitive.

Our goal is to address the industry challenges with a set of enterprise-ready technologies on the Microsoft platform. We use the following principles in the examples:

* Enterprise-class solution
* Standard communications:
o Use WS-* standards
o ACORD messages
* Ensuring interoperability with existing solutions

Business Terms Used in This Document

ACORD—ACORD (www.acord.org) is a nonprofit association whose mission is to facilitate the development and use of standards for the insurance, reinsurance, and related financial-services industries.

Order system—Creates requests for external data, transmits them to the appropriate third-party data provider, manages responses received, and matches responses to the appropriate original requestor.

Third-party service provider—An external system to fulfill an underwriting requirements request (for example, a credit-rating system).

Underwriting process—Implementation of the business process for assessing and processing a new business.

Broker system—Possible smart-client front-end system for order entry and progress monitoring used by an insurance broker. Other front-end systems are also possible, such as a Web portal to brokers, or a Web UI for self-service order entry by customers.
Life-Insurance Policy Scenario

The customer, Robert, wants to purchase a platinum-level 1 million dollar life-insurance policy. The broker, Tom, enters Robert's policy application using his smart-client application. The policy is sent to the Order system, where it is processed and routed to the appropriate systems to begin the underwriting process. While in the Order system, third-party services are kicked off. For this scenario, we will use a Paramed, a third-party service that verifies insurers' health insurance and medical records.

The built-in business logic can also generate requests to third parties if a certain condition is met. This could be the broker or another partner of the insurance company.

Figure 2. Business process used for our scenario
Architecture Overview

This section will walk through the high-level logical architecture used in the scenario. The details around specific aspects such as security, messages, development, and deployment will be provided in other papers in this series.

To ensure applicability with real-world challenges, we derived a set of high-level requirements.
Requirements

The following requirements are for an enterprise-class solution:

* Must interoperate with existing, commercial off-the-shelf applications. As discussed earlier, many organizations purchase and customize software. It is critical to address this.
* The integration technology must be Web services. Many forms of communication, such as binary communication, are proprietary. Until the emergence of Web services, there was no standardized way to communicate messages. Web services provide a way to communicate across heterogeneous platforms.
* WS-* standards must be used. Web services using SOAP and WSDL have been industry integration standard for years. However, these traditional Web services lack the robustness needed for messaging. The WS-* standards provide these necessary features without the usage of binary communication.
* Long-running workflow. Management of long-running orchestrations has been difficult, especially when that workflow spawns many smaller external workflows, in which case reconciliation and transaction management can become complex.

We use BizTalk as the message hub for this solution, given its rich capabilities and the strong need this insurance solution has for tying multiple systems together and managing multiple external workflows.

Figure 3. Using message-bus technology

Shown in Figure 3 is an enterprise view of BizTalk as an enterprise service bus (ESB). Remember that it is not a requirement that this is used as an ESB. This white paper refers to this layer as just a message layer, so that you can incorporate it into your solution in either case.

The rationale for using BizTalk is that it provides a centralized platform for the following capabilities:

* Business-process management—Centralizing reusable business process not only lends to service orientation, but also provides a mechanism for organizations to augment existing or purchased commercial off-the-shelf–based (COTS-based) applications without the complexity of modifying them.
* Workflow orchestration—Management of multiple workflows can be simplified through this platform. Instead of coding or reconciling each workflow, solutions can be managed as they should. We do this by creating one workflow to manage the business process from the beginning to the end that is able to orchestrate multiple internal system workflows.
* Rich adapter support—Jump-starting development is critical to organizations. BizTalk has a wide range of adapters to support your integration needs. In the insurance space, there is an ACORD adapter that can jump-start your integrations. In conjunction with the ACORD adapter, the Web Services Adapter and File-Based Adapters are available for BizTalk.
* Message routing and transformation—Message routing can be very complex when the messages must be transformed so that other systems understand the message. BizTalk can provide a platform to reduce the complexity and still align with open standards.

The Insurance Agent Policy System

Currently, the technology trends in the insurance industry vary from portals, thick clients, 3270 mainframe terminal-emulation screens, and smart clients. Given the diverse number of applications and vendors in this space, we choose a smart-client user interface (UI) to provide the optimal experience for the agent for the following reasons:

* Offline and online modes
* No dependencies on network connectivity
* Rich user experience with much greater functionality

A disconnected model for agents makes sense in many situations, as brokers can often be mobile or have limited connectivity to network resources. However, because we will be using Web services as the core of our messaging strategy when architecting this solution, the manner in which the end broker submits policies should be trivial.

For the client-side architecture, we used Windows Forms as the user interface, which provides the user interface needed for the agents. There will be several controls, such as data grids, text boxes, and command buttons. A data grid on the Windows form will serve as the window into the policy pipeline for the broker. We use Web services to update this data grid to ensure real-time updates.

Because this is a smart client, returning data can be cached for offline viewing and updating. This provides significant benefits to the brokers. In addition to the data, a small layer of business logic would reside on the client application. The majority of the application logic will reside on the insurance company side. The rationale here is that we will have light rules to drive UI functionality.

Figure 4. Client logical architecture

To make the calls from the client to the messaging tier, we will use Windows Communication Foundation (WCF). WCF will send SOAP 1.2 Web services messages using the ACORD messaging schemas. The WCF layer will provide a unified development model for our developers when coding communications. From the protocol perspective, we will use a series of WS-* standards. However, this is not enough to ensure interoperability. Usage of the ACORD industry standards is key, too. We should be able to interoperate seamlessly between "homegrown" applications, COTS applications, and third-party services.
Messaging Architecture

The use of Web services enables this broad variety of channels to leverage a common Web service that receives new business applications into the underwriting process in the form of an ACORD 103 message that includes a policy number that has been assigned, and that will be used for tracking/correlation purposes throughout this demonstration. This ACORD 103 New Business Submission message will be based on a SOAP Message Transmission Optimization Mechanism (MTOM/XOP) attachment containing the binary representation of Robert's signature to authorize release of medical information, as required by HIPAA. It is absolutely critical that the ACORD standards are incorporated in our messaging. This will ensure portability of the architecture.

It is also essential that communications are secure and reliable. To achieve, this we will use WS-Secure Conversation (WS-SC) for personal information that might pass through an undetermined number of intermediaries. We also use WS-SC for high-volume, frequent requests (such as credit checks) that will be required for all new policy applications. We use WS-Security for less frequent requests, such as an Attending Physician's Statement (APS), where the overhead of session establishment is not justified by the request volume. We also use TLS/SSL (also known as HTTPS) in rare cases where a service is directly processing requests without any intermediate routing.

For messaging where tracking reception is important, such as ensuring receipt of a new policy to claim a commission, we use WS- Reliable Messaging (WS-RM). We also use WS RM for data requests that are expensive to process (typically involving human workflow, such as APS queries). This ensures that requests are only delivered once, and avoids expensive duplicate requests.

For long-running messages, we use WS-Secure Conversation (WS-SC) (See Resources.)

Figure 5. Client message-exchange patterns
Transaction Business processes WS-* protocols Architecture decision
Submission of new policy (103 Request) Broker-client

Underwriting process
WS-Security (WS-S)

WS-Reliable Messaging (WS-RM)
WS-S used for personal information that might pass through an undetermined number of intermediaries.

WS-RM used to track message receipt.

Because of infrequent transactions, no need for session-oriented security mechanisms, such as WS-Secure Conversation.
Status queries (122 Request/Response) Broker-client

Underwriting process

Fulfillment process
WS-Secure Conversation (WS-SC) Noncritical and individual request or response messages that can be retried easily, but still contain personal information.
Underwriting requirement order request (121)

Underwriting requirement order response (1122)
Underwriting process

Fulfillment process
WS-Secure Conversation (WS-SC) or WS-Security (WSS) or Transport-level security (TLS/SSL)

WS-Reliable Messaging (WS-RM)
These messages contain personal information.

WS-SC will be used for high-volume, frequent requests (such as credit checks).

Use WS-Security for less frequent requests, where the overhead of session establishment is not justified by the request volume.

Use TLS/SSL where a service is directly processing requests without any intermediate routing.

Use WS RM for data requests that are expensive to process.

Table 1: Business-process messaging design decision matrix

You might ask yourself, after making a submission: Why is the status returned in a separate transaction? Well, the reasons are twofold. Firstly, it is important for this to be asynchronous, and the ACORD standard does not allow an implementation without separating the status from the submission. Secondly, the Broker will be getting status returns periodically through the course of the application process by querying the Status Service.
The Insurance Carrier Systems

When architecting the server side of the solution, there were particular aspects and assumptions considered:

* This architecture accounts for fragmented systems.
* Functional areas are self-contained and need to be managed.
* Operating systems and development environments differ.

As a result, there are a significant number of point-to-point integrations with very specific applications, thus causing proprietary implementations. In this solution, the façade layer will be created around these existing applications.

Figure 6. Insurance message bus

Here, you can see how we are using the enterprise service bus (ESB) as a message bus. This layer will serve as the centralized messaging layer that will manage our internal and external messages. Management and orchestration are key benefits of this architecture.

An infrastructure like this can bring order to the chaos of disparate point-to-point integrations by putting intelligent, long-running orchestrations and policies around transactions in one layer instead of many. It would be common to have several distinct COTS-based applications in upwards of five or six systems to accomplish an end-to-end transaction. We are reducing these systems significantly by consolidating the redundant functions, such as workflow and messaging, leaving infrastructure-level functionality where it belongs and keeping the business logic in applicable applications.

It is important to remember that this message bus is a logical representation. The implementation view can look very different. For example, the message bus could be several BizTalk servers, or there could be servers in different DMZ environments to manage both internal and external communications.

Click here for larger image

Figure 7. Workflow designer (Click on the picture for a larger image)

The next tier down, which is where specific business functions are performed, contains two different legacy systems wrapped with an interface: the Order system and the Fulfillment system. The reason that we are keeping these as separate systems instead of consolidating them is that, the majority of the time, these would be two separate COTS-based systems.

A Status system was added for the following reasons:

* To provide a centralized way to report status to the agents.
* To reduce the number of interfaces and control logic needed to query multiple systems.
* It fits nicely with the orchestration capabilities of our ESB for our long-running workflow.

The Ordering system and the Fulfillment system have been converted into course-grained services. By doing so, we have removed the dependencies of independent implementations. All communication that occurs to these systems now goes through our message hub. The exposed Web service endpoints that are managed from the message bus can then be managed with orchestration technologies built in BizTalk.

Figure 8. End-to-end message-exchange pattern

Now that these applications are exposed as Web services, any technology that can accept Web services XML can integrate with these applications. This removes the tight coupling of other technology protocols that would limit interoperability. For example, you could just as easily use existing Java-based systems, if those were your legacy systems.

SQL Server is used here to store application data in the database layer. Because the core focus of this paper is integration and composite applications, we will not highlight this.

The third-party services referenced are external services that are called by the Fulfillment service. These services have varying protocol needs. However, this paper will show how WS-* standards can provide increased functionality for your services. It is important to note that many of the real-world insurance third-party services only support XML-based communications, not the more advanced SOAP-based Web services. The messaging-architecture sections that follow will have more on the third-party services.
Insurance Carrier Messaging Architecture

This section walks through a basic life policy that is processed by the insurance carrier. Based on the information Robert supplied, the business rules/heuristic logic defined in the underwriting process decides that an Attending Physician Statement APS (that is, a physical) is also required.

Because another provider must fulfill this request, the Order system builds an ACORD XML TransType 121 General Requirements Order Request transaction (TXLifeRequest) and transmits it to a secondary external ordering system for Robert's physician (the APS system). This message also contains the MTOM/XOP attachment of Robert's signature that was originally carried on the ACORD 103 New Business Submission, authorizing his physician to release his medical information to the insurance company.

At some point, Robert's physician will process the APS order by verifying that Robert's signature matches the one he has on file, and then examining Robert's medical history, filling in the necessary information required on the APS report.

After the physician has completed the APS report, an 1122 General Requirements Status/Results Transmittal message is generated and transmitted back to the endpoint reference specified in the WS-Addressing ReplyTo specified in the previous ACORD 121 request. This message will also be delivered reliably using WS-Reliable Messaging.

The rest of the business process runs, including any automated-actuary decision. However, in this case, because there is an APS and possibly some additional information that cannot be processed automatically, the case is flagged for an underwriter's review and approval.

Figure 9. Underwriting process message-exchange pattern
Fulfillment Service

In the insurance industry, a fulfillment system or service is very different from the process of fulfillment:

* Fulfillment system: A system or service that receives a request and fulfills it. Think of a fulfillment service as an integration component for gathering data. In this scenario, the fulfillment system is responsible for pulling the various reports from third-party providers.
* Fulfillment process: The process in which a policy is issued by the insurance carrier.

You might ask why we kept the fulfillment service. For this scenario, we are assuming that systems such as these are purchased as black-box solutions. This is not to say that you could not remove these layers and incorporate them into a message bus, instead.

Choosing a messaging pattern is not as clear-cut as choosing one set of standards. When designing this part of the solution, we had to take a step back and look at the business and legacy aspects of each individual transaction.

Some transactions, such as receiving a credit report, were easier decisions. However, other transactions, such as pulling an APS report, required the ability to contain attachments.

Here are some aspects to consider when designing you messaging:

* Understand the business process. It is critical to understand how the business uses these messages (for example, securing data). If the data being sent is not sensitive, you do not need to take extensive security precautions for the message.
* Understand how the transactions are consumed from the service providers. This can be both internal and external. Many times, when relying on service providers, there are technical limitations. These can range from standards support to hours of operation.
* Give proper attention to security. This is often overlooked. Protocol-level security, such as SSL/TLS, often is sufficient, but not always. Make sure you evaluate the sensitivity of the data and review the message paths to determine how many endpoints there are before the ultimate consumer.
* Be realistic and pragmatic. When designing these services, do not go overboard trying to use every standard. Do not force a standard into a message, if it does not belong. This will only introduce unneeded complexity.

Figure 10. Fulfillment system message-exchange pattern
What Is the Value?

We talked quite a bit about the Microsoft platform and development technologies by walking through the scenario. We also highlighted architecture decisions. But what we did not do is highlight the features of these Microsoft technologies.

The following are the core benefits using Microsoft technologies in the insurance industry:

* Business-process automation—Business processes are complex and specific to each carrier. With the orchestration tools provided in BizTalk, orchestrations can be developed by business analysts, removing the developer from this process and enabling the business.
* Reduction of integration code—With the custom adapters in BizTalk and the unified programming model of WCF, the code required to integrate systems is drastically reduced.
* Alignment with standards—WCF and BizTalk are based on Open XML standards out of the box. No more custom coding to incorporate Web services standards.
* Productivity—With an integrated Visual Studio IDE and .NET 3.0 technologies, both the tools and the development language provide substantial productivity gains over other languages.

Conclusion

As demonstrated in this white paper, using protocol-level standards alone is not enough; capturing the business side of the messaging transactions is key to making interoperability work for your business. This is true across all industries, not just insurance.

We have Web services standards, but that is not enough. There is still a level of due diligence that is required to make the optimal technology decisions for your organization. With this specific reference implementation, we go through a real-world scenario and determine the optimum messaging with this scenario's business forces. This can serve as a guide to help you choose message-exchange patterns in your enterprise. With all architectures, there are trade-offs when choosing specific standards. It is important to understand these trade-offs and be willing to take on the resulting implications.

Microsoft is committed to making the job of architecting and developing service-oriented solutions easier for its customers. And we show here that Microsoft has removed many of the industry barriers and complexities that daunt customers today. These range from providing thought leadership in the industry standards to automating and building out-of-the box Web services support.

Read More / Source