E*TRADE Migrates to Red Hat JBoss A-MQ From Sonic MQ and IBM WebSphere

Logo - No Image

July 12, 2013

E*TRADE used open source offering, Red Hat’s JBoss A-MQ, to build a risk-computing platform for its high-volume trading products, achieving higher performance, greater scalability, and improved reliability at significantly lower cost.

Customer: E*TRADE

"In addition to reduced licensing costs, Red Hat JBoss A-MQ delivered stability, scalability, and performance improvement." - Duc Doan, Director, System Engineering

Geography: North America
Country: United States


Business Challenge:

Online trading firm E*TRADE was dissatisfied with the commercial message buses it was using in its mission-critical risk-computing platform.

Migration Path:

Migration from SonicMQ and IBM WebSphere MQ to Red Hat JBoss A_MQ

Software:

Red Hat JBoss A-MQ

Hardware:

Dell PowerEdge R610 multicore servers with four gigabytes of memory

Benefits:

E*TRADE used open source offering, Red Hat’s JBoss A-MQ, to build a risk-computing platform for its high-volume trading products, achieving higher performance, greater scalability, and improved reliability at significantly lower cost.

More
Background:

E*TRADE is an innovative financial services company offering a full suite of easy-to-use online brokerage, investing, and related banking solutions, delivered at a competitive price. The company empowers individuals to take control of their financial futures by providing the products, tools, and services they need to meet their near- and long-term investing goals. These services are provided to customers online and through the E*TRADE network of customer service representatives, investment professionals, and investment advisors—over the phone and-in person at 30 E*TRADE branches.

Business Challenge:

Commercial message bus products too costly

Margin loans are a significant source of revenue for E*TRADE Margin requirements—or the ratio of money borrowed versus asset value—are constantly changing due to regular fluctuation of the asset based on market conditions. This ratio must be monitored closely to reduce risks, preferably in real time. For brokerage firms in general, the revenue from margin loan income has manageable risk—if the brokerage firm has adequate resources and modern technology to monitor and manage those risks.

E*TRADE had a risk-computing platform that it built in-house to monitor margin ratios and mitigate margin risks using SonicMQ and IBM WebSphere MQ as the message buses. This platform collected and analyzed market data and trading events using complex calculations in real time for E*TRADE’s popular Market Trader, E*TRADEPro, and myE*TRADE products. But the cost of the licenses and the constant painful upgrades made Director of Systems Engineering, Duc Doan, look for a message bus that could provide better performance and scalability at a lower cost. He was also looking for a solution that would be easier to manage and deploy.

E*TRADE also had to maintain separate teams to manage the two different messaging buses.

“We’re very cost-sensitive, but it wasn’t strictly the licensing cost that made us dissatisfied with our current stack,” said Doan. “There are a lot of different message technologies out there—all commercialized—and we had to pay multiple licenses and to keep upgrading them to the most recent version. That was very high cost and high maintenance. We wanted to consolidate our messaging stack so one team could manage it.”

An open source proponent, Doan decided to investigate Red Hat JBoss A-MQ message bus.

Solution:

Building a risk-computing platform with Red Hat JBoss A-MQ

With its small footprint, flexible configuration, and high-performance messaging platform, Red Hat JBoss A-MQ helps enterprises like E*TRADE to build integration solutions more rapidly. Organizations can respond to business events quickly, with intelligence, and with high-quality answers, products, and services.

The platform consists of a risk console, a complex events processing (CEP) platform, and data collection agents interconnected with the Red Hat JBoss A-MQ message bus in a distributed environment. With Red Hat® JBoss® A-MQ, E*TRADE was able to design a highly available, scalable, and high-performance architecture with multiple partitions, where each partition can accommodate up to 5,000 customer accounts with built-in redundancy for message persistence and high-performance caching. “We started building up the platform to support a small group of high-net worth accounts, and gradually ramped it up to thousands of accounts at very low cost,” said Doan.

Red Hat JBoss A-MQ enables integration that goes beyond the datacenter. Its elastic IT footprint means organizations like E*TRADE can deploy it as much or as little as they need. In addition to being able to put a broker anywhere, Red Hat JBoss A-MQ’s small footprint makes it easy to implement with inexpensive hardware and limited IT staffing. Because it has centralized management, it is easy to configure, deploy, and manage customized brokers. Finally, it is cost effective. As an open source product, it is less expensive than traditional stacks, making it practical to deploy broadly.

Benefits:

Greater scalability, reliability, and performance with Red Hat JBoss A-MQ

First and foremost, Doan saw a significant jump in performance.

“For us, transaction throughput is very, very important,” he said. “The faster you can confirm a trade, the faster our customers get their money, which they can use to make even more money. The performance and stability of the message bus is critical to our line of business.”

Next, Doan appreciated the stability of Red Hat JBoss A-MQ. “We had to keep upgrading Sonic MQ and IBM WebSphere MQ, and upgrading commercial products is not an easy thing to do,” he said. “Now we no longer see the error alerts we were getting with the other message buses.”

Another advantage was that Doan could run Red Hat JBoss A-MQ on the latest hardware to get the greatest throughput, capacity, and performance gains.

Because E*TRADE used Red Hat JBoss A-MQ to consolidate the formerly disparate message brokers into a single broker deployment, it can run multiple products on it, with the ability to scale as needed.

Doan also switched out the service-oriented architecture he was using to move to an event-driven architecture. By doing this, he’ll be able to put all services on the same message bus so they can send the same event to other services. Doan expects to use this to extend greater capabilities to E*TRADE customers, such as the ability to use market data in their portfolio margining.

“We want to use a consolidated message bus that can use the same protocol, and in the process, create a Red Hat JBoss A-MQ client library where all of our service code stack can work together with the broker,” said Doan. “That way, any service can be registered to or subscribed to by any of the other services. In effect, we’re creating a giant social network where all the services can talk to each other.”

“We typically do 300,000 transactions every day,” said Doan. “If something goes wrong with the message bus, there are a lot of costs associated with fixing it.

Contact Sales

Less