Tools & Automation

The recent explosive growth of transactional information and applications over the Web has led to a very real concern for IT managers - how to address the processing bottleneck in Web and application servers. For service-oriented architectures (SOAs) that use XML to bridge the transfer of information across disparate technologies, this processing bottleneck may hinder the deployment and adoption of XML standards and Web services.

The fundamental basis for Web services starts at the lowest level, standardization of protocols and data formats. The XML standard is the chosen data standard for many good reasons including being human readable, hierarchical, and extensible, but because of it's verbosity it is inefficient for machine-to-machine interaction. In many cases, the computer resources needed to process XML datasets in an enterprise can consume as much as 80% of the available host CPU cycles. As the amount of XML traffic increases so does the demand on the available CPU processing capability. Some models project that IT departments may have to double the number of application servers because of the impact of XML/XSLT processing. Today, IT managers are faced with a difficult choice, to delay or slow the development of Web services or spend scarce IT capital dollars to increase the number of application or Web servers in the data center.

An alternative solution is to integrate specialized hardware platforms in the datacenter to offload the XML processing necessary to keep up with demand. While this approach offers an immediate benefit in terms of performance, it is imperative that the solutions considered satisfy the strict requirements for datacenter systems. These include specific requirements for ease of integration, scalability, robustness, and remote management. Once these requirements are addressed, the price and performance advantages of true hardware-based platforms can be realized.

Performance, Performance, Performance
Custom XML processing devices allow the separation of XML data processing from the business logic on application servers. It is anticipated that devices on the market in 2004 will accelerate the processing of XML up to 20x that of standard general purpose server devices.

Although traditional Intel-based server architectures can be used to address this problem (when combined with software algorithms for XML processing), this approach will not offer the performance density and price performance of an XML hardware device that includes XML-specific processors.

Purpose-built, XML hardware platforms achieve the highest levels of performance by combining a full range of technologies including:

• Parallel processing hardware and the associated software compiler technology to exploit it. This hardware could involve multiprocessors that independently parse, validate, and transform (parallel processing engines); the support of multiple threads; the creation of highly efficiently parse data structures, and the ability to couple the parse to portions of the transformation stylesheet to exploit the parallelism of the template match patterns and Spath expressions.
• Caching of static documents and instructions and structures for fast retrieval.
• Segmented functional units in a highly pipelined network architecture to extract, process, and distribute data to achieve maximum performance (such as SSL, TCP offload).

It is anticipated that devices built on such an architecture will offer XML/XSLT processing at more than 1Gbps throughput in 2004. An example of the data processing flow in such a device is shown in Figure 1.

Act Like an Enterprise System
Performance is not the only criteria in delivering enterprise-class Web services acceleration platforms. Enterprise datacenters demand adherence to standards, seamless integration into the existing environment, ease of deployment and management, and the highest availability.

Ease of Integration and Deployment
Integration with standard Web services, data from databases, and data from other applications that run on other platforms are all critical in today's data center environment. This integration is best done using standard APIs that are being supported in the marketplace. Hardware-based platforms will be required to interface to these APIs and provide key features and functions to take advantage of the price performance advantages offered by hardware that includes configurable parsing, validation with schemas or DTDs, and transformation using XSLTs. All functions should be available using standard APIs.

Configurations for applications residing on the hardware-based platform interface through standard APIs such as the Java API for XML Processing (JAXP) that allows applications to parse and transform XML documents using an API that is independent of any particular XML processor implementation.

Another integration approach in the enterprise environment is via the proxy mode. The processing of the data is initiated when a message is sent over the network via a messaging service such as SOAP over HTTP.

Control and Management
The deployment of "black box" machines within a data center is a daunting task to a network administrator who must maintain, manage, and configure each device to ensure high availability during periods of peak processing. The reality is that a specialized hardware platform must provide the network administrator with the means to remotely monitor and configure any system within the datacenter. This network management port should be a secure and nonintrusive entry used to retrieve log files, error reports, usage modes, and statistical analysis. A management information base (MIB) is required and available over SNMP so that network administrators can monitor and manage the device to check for statistics such as bytes sent/received, fragment packets, dropped packets, and other statistical information.

As a high-availability, high-performance platform, the XML hardware device should provide the highest levels of robustness possible. The data path, Web services, and documents being processed should therefore be kept separate from the control path, configuration information, monitoring processes, and device logging.

The separate data path and management ports should include:

• Non-intrusive secure management port to retrieve log files, error reports, statistical analysis, etc.
• Separate processor subsystems for data and control paths
• Encryption support for both paths

Robustness
In addition to being able to observe system behavior, error logging, and even predictive error heuristics, the system should be able to operate with transitory faults or even with failing hardware with degraded, but reasonable performance. Important hardware system attributes here include:

• ECC protected memory for detecting and correcting errors
• Fail-over mechanisms to detect when a memory module(s) fails with an ability to redirect data to another memory module and run with limited number memory banks
• Redundant architecture that upon failure allows de-rated performance and/or fail-over mechanism to redirect the data to another network port to keep processing data
• Software monitor to determine health of server and report or alert of hard failures and soft failure trends that suggest impending hard failures

Conclusion
A well-architected, hardware-based XML processing device must offer the flexibility of deployment required by the APIs established in the industry, and it must provide all the expected quality of service and reliability features of high-end server systems. These features and qualities enable enterprise adoption and integration that fulfills the promise and potential of high-performance, low-cost Web services and XML message processing.

Conformative Systems <CSXi>
Conformative Systems' <CSXi> server appliance is one example of a hardware-based processing device specifically designed to process XML transactions. Each appliance can deliver the throughput of up to 20x that of equivalent of software-based solutions running on general-purpose servers. Custom ASICs and parallel processing technologies built into the appliance perform the parsing, validation and data transformation on all XML data at greater than wire speeds. Each Conformative Systems CSXi server appliance is built on a fully redundant architecture offering the highest single-device availability for business-critical enterprise applications.