With the recent launch of its Financial Markets Framework, IBM is positioning itself as a provider of pre-integrated solutions to trading and investment firms, leveraging its own portfolio of hardware and software offerings, and augmenting them with consulting expertise and partner products. The move follows a direction that the IT giant has already taken in verticals such as banking, insurance, and outside of financial services.
While IBM is not pitching itself as a vendor of complete trading applications, it is looking to provide all of the building blocks with which to create them, including server and storage products, messaging middleware, information management and analytics offerings. Plus - vitally - it is also providing architectures and designs to integrate these components for optimum functionality and performance, as provided by IBM's consulting business.
It is this integration expertise that underpins what IBM is referring to as "frameworks" - and which are designed to make it easier and quicker for customers to deploy applications. This reduces both cost and time of implementation - two perennial issues for trading firms looking to move rapidly into new business areas, and increase operational efficiency.
Moving beyond a current focus on the front office - as exemplified by its WebSphere Front Office and WebSphere MQ Low Latency Messaging products, and solidDB in-memory and Informix time-series databases - IBM is pushing into the middle and back office, leveraging products such as InfoSphere Streams for stream (complex event) processing, Ilog for rules processing, and Cognos and SPSS for business intelligence and analytics.
Integration of partner offerings also forms part of the frameworks approach. For trade matching, Chi-Tech's software can be integrated - as has been proven at the trading venues that parent Chi-X operates globally. And for settlement risk, R2 Financial Technologies provides risk calculation software. IBM has also partnered with Fixnetix to provide co-location and market connectivity.
As well as integration of its own and partner products, the need to embrace existing legacy infrastructures within customers is also understood, with the frameworks' architecture being based on open standards and being agnostic to the use of specific components. For example, a customer with an existing database would not be required to adopt DB2 or Informix.
IBM has identified six broad application areas to target with frameworks:
* Low-latency market data delivery.
* Algorithmic trading.
* Trade execution and matching.
* Post-trade securities processing.
* Market surveillance and trade monitoring.
* Settlement risk management.
In order to demonstrate how frameworks are implemented in the real world, IBM has created the Securities Exchange Reference Architecture (SXRA) for trade execution and matching, and from it has built a working implementation, and subjected it to various performance benchmark tests.
For its SXRA benchmark implementation, IBM designed a system to include a FIX gateway, a matching engine, a data logger and an in-memory database, and also implemented an order management system to test against the core trading system.
Specific software components used in the SXRA implementation are WebSphere MQ LLM and solidDB, as well as custom integration code for data logging, running mostly on Red Hat Linux (which IBM distributes). The hardware environment comprises: IBM HS22 Blades (for the gateways and trade matching) and IBM 3850 servers (for data logging), all comprising Intel Xeon 5500, 5600 and 7500 series microprocessors; an IBM Power 7500 server (for solidDB); and an IBM XIV storage system.
Both Voltaire InfiniBand and Mellanox 10 gigabit Ethernet (supporting RDMA) is included in the implementation as a comparison and to demonstrate real-world implementation scenarios.
Among the many benchmark figures derived, at a rate of five million orders per second, the round trip latency from the FIX gateway to the matching engine was as low as 12 microseconds.
Benchmarks for Blade-to-Blade communications show that at 1.5 million messages per second (using 45 byte messages), round trip latency is an average of 2.5 microseconds for InfiniBand and eight microseconds for 10gE, with InfiniBand providing the lower jitter.
As well as demonstrating raw latency and message throughput performance, the SXRA implementation also illustrates high availability features, such as hot fail-over, and load balancing to scale up and down to meet capacity.
In being able to integrate its own hardware and software products, and augment them with its consulting knowledge and best practices drawn from real-world engagements, IBM has carved out a unique position as a provider of technology to financial markets firms.
While the likes of HP and Dell (and increasingly Cisco Sytems) compete with IBM on hardware and integration, most trading markets-focused competitors - some of which, like NYSE Technologies, are also close IBM partners - provide just software, and sometimes integration.
Perhaps the closest competitor to IBM is Oracle, which - as a result of its acquisition of Sun Microsystems - now plays in both the hardware and software spaces, and is looking to provide strong integrations between them. But while Oracle has an array of data management software, including in-memory and relational databases and a CEP engine, it lacks some key components, such as low-latency messaging, and the many data feed handlers that IBM has developed. If Oracle wants to compete fully against IBM, it will most likely need to make an acquisition.
For IBM, one concern is that its pre-integrated solution approach will drive away key software partners, who might feel it is becoming too competitive. And while IBM executives say the company is not looking to become full application vendor, that could happen as the by-product of a future acquisition, as Sybase found when it acquired Aleri and thus became a vendor of applications, such as for algorithmic trading, liquidity risk management.
For sure, IBM's direction adds a new dimension to the buy versus build choice, offering a buy most, build some scenario that could prove attractive to financial markets firms, and should also provide good business for its high growth, high margin consulting operation.