Customer Solution Case Study
/ / Reuters Research Enjoys High Availability for 10 Terabytes of Financial Information
Overview
Country or Region:United States
Industry:Financial Services
Customer Profile
Based in New York, Reuters provides information, news, analytics, and systems to financial services, media, and corporate markets. Reuters employs some 16,000 people in 94 countries and serves more than 52,000 client locations.
Business Situation
Reuters Research needed flawless replication, high availability, and easy administration to support its more than 10 terabytes of financial information.
Solution
Reuters Research deployed more than 700 database instances of Microsoft® SQL Server™ 2000 on some 150 computers to support its financial information delivery system.
Benefits
Flawless replication
High availability and disaster recoverability
Easy administration
Upgradeability
Scalability / “Replication is at the foundation of our entire Reuters Research solution, and is one of the most important elements of our design. SQL Server transactional replication has worked flawlessly for us.”
Alex Meikson, Vice President, Database Administration, Reuters Research
Reuters Research uses more than 700 database instances of Microsoft® SQL Server™ 2000, hosted on 150 computers running the Microsoft Windows Server™ 2003 Standard Edition operating system, to support customer access to the more than 10 terabytes of financial information in its flagship Reuters Knowledge service. Reuters Knowledge, a global information and analytics service for research and investment banking professionals, provides a centralized source of information including fundamentals, earnings estimates, and financial reporting from industry analysts. Using SQL Server 2000 as a foundation for the Reuters Knowledgeapplication has provided the company with efficient data replication used for populating a secondary datacenter, backup copies, and special-purpose clusters; high availability; easy administration using SQL Server Enterprise Manager; and an easy upgrade path.
Situation
Founded by Paul Julius Reuter in 1851, his namesake company has grown into a global information organization with operations in 91 countries. Reuters, which in the early days used a flock of homing pigeons to deliver financial news faster than it could travel by train, is perhaps best known as the world’s largest international multimedia news agency. Yet more than 90 percent of its revenues come from its broad range of offerings for the financial services industry.
Reuters information is trusted and drives decision making across the globe. The company has a reputation for speed, accuracy, and freedom from bias. Some 330,000 financial market professionals working in the equities, fixed income, foreign exchange, money, commodities, and energy markets around the world use Reuters products. They rely on Reuters services to provide them with the information and tools they need to help them be more productive.
In 2003 Reuters purchased Multex, a financial research and information service with extensive content assets and technology, and renamed it Reuters Research. Reuters Research launched Reuters Knowledge, a Web-based financial information delivery system, which supplies real-time and historical equity, fixed income, and economic research from the world's leading analysts to investors all over the globe. The Reuters Research financial research database contains morning notes and research reports from all the major global brokerage firms and hundreds of regional specialists. In addition to the financial research data, which is a flagship of Reuters Research financial information offerings, company fundamental data, earnings estimates, significant developments, business intelligence, and other financial data is collected and provided for thousands of companies around the world.
As Reuters Knowledge was launched, the company needed to ensure:
Efficient data replication
High availability and disaster recoverability
Easy administration to support the dynamic nature of the industry
Upgradeability
Scalability
Solution
The Reuters Knowledgeservice provides comprehensive content coupled with powerful search capabilities to help users better understand the changing marketplace. The financial research data gathered by Reuters Knowledge includes documents from more than 900 sources including the world’s leading brokerage firms, investment banks, market research, and independent research providers. The Financial Research historical database contains more than 9 million full-text reports covering over 25,000 companies that span 100 countries and 23 languages. The Company Fundamentals database contains information on about 50,000 companies, and the Earnings Estimates database has information on more than 25,000 companies.
All of this financial data is stored on 231 unique databases with more than 500 instances of replicated databases, including backup and working copies. All these databases are located on some 150 computers utilizing Microsoft® SQL Server™2000 Standard Edition. Locally attached storage is about 25 terabytes in size. For the financial research documents the SQL Server databases include pointers to the documents stored in PDF and other formats on file servers holding about 10 terabytes of disk space. Reuters Knowledge runs on SQL Server on the Microsoft Windows Server™ 2003 Standard Edition operating system. Both Windows Server 2003 and SQL Server 2000 are part of the Microsoft Windows Server System™ integrated server software.
A multi-tier architecture is used for Reuters Knowledge, including:
Presentation Tier. A Web browser is used to access information from the presentation tier that is hosted on a number of load balanced computer clusters running Internet Information Services version 6.0, the Web server that comes with Windows Server, and Windows Server 2003 Standard Edition. The presentation tier is hosted on Hewlett-Packard (HP) ProLiant DL380 2-way computers with 2 gigabytes (GB) of RAM.
Application Tier. The application tier is built as a standard Web service that provides a generic interface between the presentation tier and the backend. The application tier is hosted on HP ProLiant DL380 2-way computers with 2 GB of RAM.
Data Tier. All this financial information is stored in SQL Server 2000 relational databases. For the financial research information, SQL Server stores document description data, including pointers to where the actual documents are stored on the file servers. Data is stored on read-only queryable clusters, and on non-queryable reporting clusters for internal use. The clusters making up the data tier are hosted on HP ProLiant DL380 2-way computers and HP ProLiant DL560 4-way computers with 2 GB of RAM.
SQL Server provides the foundation of the highly available Web services-based environment that supports the Reuters Knowledge service. Reuters uses SQL Server transactional replication to forward incremental data changes from the publishing database, through a distributor, to subscribers, limiting the demand on the publisher to forwarding changes from a previous data state already present at the subscriber. Distributors run SQL Server 2000 and Windows Server 2003 on an HP ProLiant DL380 2-way computers and HP ProLiant DL560 4-way computer with 2 GB of RAM.
The basic multi-tier architecture is replicated symmetrically (with specific exceptions) at the company’s main data center at 75 Park Place, New York, with its secondary data center in Nutley, New Jersey.
Benefits
SQL Server is providing Reuters Research and its Reuters Knowledge service with a number of benefits, including flawless replication, high availability and disaster recoverability, easy administration, upgradeability, and scalability.
Flawless Replication
Reuters Research uses transactional replication across its operations, including for populating its secondary data center which provides a live/live environment and serves as a warm backup for recovery. Replication also is used to create backup copies within sites, and for updating multi-node query and reporting clusters.
“Replication is a very big deal to us,” says Alex Meikson, Vice President, Database Administration at Reuters. “Replication is at the foundation of our entire Reuters Knowledge solution, and is one of the most important elements of our design. SQL Server transactional replication has worked flawlessly for us.”
Flawless replication wasn’t what Meikson was expecting after his experiences with other database products. “SQL Server transactional replication is very stable,” says Meikson. “It doesn’t break. This kind of replication stability isn’t what I’ve experienced working with other databases.”
High Availability and Disaster Recoverability
SQL Server transactional replication helps Reuters Research meet its service level agreements (SLAs) for providing Reuters Knowledge customers with high availability. “We guarantee up to 99.9 percent availability for the interactive decision support systems, which is something considering the complexity of our environment with 150 servers running more than 700 instances of SQL Server databases, and terabytes of information,” Meikson says. “SQL Server 2000 has never let us down.”
All components of the Reuters Knowledge service production infrastructure are designed with disaster recovery in mind. Reuters Knowledge SQL Server production architecture is an integral part of this design. “We operate live/live from two symmetrical data centers,” Meikson says. “One data center is designated as the primary [75 Park Place], as this is where data originates. Each of 231 master databases in the primary data center has at least one backup copy on a warm standby server in the secondary data center in Nutley. The near real-time synchronization between the master database and its backups is maintained through Microsoft transactional replication.”
Prior to opening its data center in Nutley, both of the company’s data centers were in the financial district of Manhattan, with the primary data center located a block away from the twin towers of the WorldTradeCenter. When the WorldTradeCenter was destroyed on September 11, 2001, Reuters Research lost power to both of its data centers.
“We couldn’t even get access to the buildings for three days,” says Meikson. “When we finally got a generator into our secondary data center on Williams Street, all we had to do on the SQL Server side was re-configure the backup SQL Servers as master SQL Servers and we were up and running again. It was unbelievable, and I say that because I have 20 years of experience as a database administrator working with DB/2, Sybase, and other products, and I’ve experienced so many lesser disaster recovery efforts where things wouldn’t come back up as they were supposed to. So we were all very pleased with the recovery capability of SQL Server.”
Two years later a major power outage affecting much of the Northeast and Midwest, from New York to Canada, and as far east as Detroit, made headlines around the world. “Again we saw the excellent recoverability of SQL Server,” says Meikson. “It is extremely resilient.”
Easy Administration
Reuters Research is at the hub of a dynamic industry where agility is required to quickly respond to new market opportunities. Meikson’s long experience as a database administrator made him keenly aware of the need for a database management system that was easy to administer—something he has found with SQL Server 2000 and the SQL Server Enterprise Manager, which provides a Microsoft Management Console (MMC)-compliant user interface.
“Enterprise Manager provides a centralized workspace for managing all of our databases,” Meikson says. “The drag-and-drop functionality of the graphical user interface is extremely efficient and easy to work with. With other database products I’ve worked with I had to memorize long command line sequences. With Enterprise Manager you can just click away without memorizing commands. It provides a nice and easy way to accomplish what you need to do.”
Meikson credits the ease with which SQL Server can be managed, along with its reliability, for the ability of his small group to manage so many database servers. The group’s four database administrators are responsible for:
Capacity planning.
Installing and configuring SQL Server.
Migrating new database structures, such as tables, indexes, and stored procedures, from the development to the production environment.
Day-to-day database upgrades according to developer requests (such as adding new columns, stored procedures, tables, indexes, and such).
Database backup and recovery procedures.
Disaster recovery procedures.
Replication.
SQL Server security management, performance monitoring and tuning, and troubleshooting.
SQL Server monitoring.
24x7 SQL Server production DBA support.
Meikson, a former database consultant stays in touch with colleagues and informally gauges efficiency of operations using other database products.
“A colleague has a staff of 12 to manage a mixed database environment that includes mainframe DB2, and Sybase running on about 100 UNIX computers” says Meikson. “I have just four direct reports, and they manage 150 servers. I’ve found that much of the custom code for utilities we used to have to write for other databases isn’t required for SQL Server because the functionality is already built into the product. When I talk with my friends who are working with other database products, they think I’m joking when I tell them of how easy and trouble free our operations are.”
Upgradeability
Reuters Research values the enhancements and improvements it gains whenever Microsoft releases a new version of SQL Server. “We immediately upgraded from SQL Server 7.0 when SQL Server 2000 was released,” says Meikson. “We saw a 30 percent improvement in query processing with the upgrade, and found that replication had been made even more stable. Because of our experiences like these, and the ease of upgrading to the next version, we are looking forward to beginning an upgrade to SQL Server 2005 as soon as the product is released.”
Scalability
Scalability is a very important feature to Reuters Research because the organization works with such large databases with its Reuters Knowledge service and other projects. “We need to be proactive to keep up with our rapid growth and to avoid running out of server space or overloading CPUs, or allowing performance to suffer,” says Meikson. “We can scale up by upgrading hardware to bigger and faster computers, and we can scale out by deploying additional servers. We use Windows® Network Load Balancing and Cisco Layer 4 load balancing clustering technology for scaling out. When we need additional muscle in our SQL Server environment, we add a new SQL Server to a cluster. The largest family of identical SQL Servers consists of 14 machines, 7 SQL Servers in each of two clusters—one at our main data center at 75 Park Place, the other at our secondary data center in Nutley.”
Microsoft Windows Server System
Microsoft Windows Server System integrated server infrastructure software is designed to support end-to-end solutions built on Windows Server 2003. It creates an infrastructure based on integrated innovation, Microsoft’s holistic approach to building products and solutions that are intrinsically designed to work together and interact seamlessly with other data and applications across your IT environment. This helps you reduce the costs of ongoing operations, deliver a more secure and reliable IT infrastructure, and drive valuable new capabilities for the future growth of your business.
For more information about Windows Server System, go to: