Hidden inside your Bloomberg is an impressive suite of trading and risk management applications.

By Andrew Webb

Although I have been aware of Bloomberg since its inception, I've spent much of my trading career using proprietary systems. From a derivatives viewpoint, I have always regarded Bloomberg as a product for dilettantes-fine for getting an overall picture (and including some neat features), but nothing you could really trade with.

I have to admit that most of my preconceptions were wrong. In the last three or four years, the company's derivatives service has come of age. Its data and analytics service incorporates derivatives functionality for all types of underlying securities. This functionality is not a stand-alone module, but is fully integrated with all other security pricing and analytics, news, and messaging.

In the past, the company was infuriatingly selfish with its data, and made it difficult or impossible for users to export its data to their own spreadsheets for analysis. But the company's Open Bloomberg system now allows derivatives aficionados to manipulate Bloomberg's proprietary database.

Bloomberg's screens had also been hampered by limited hardware capacity. A large investment in mainframes has given the company the horsepower to handle the data proliferation caused by the huge range of contract specifications in derivatives. At the same time, a large number of programmers have been added to the staff, improving the range of derivatives functionality and analysis.

You won't, however, find screens for the latest hot pricing model. Bloomberg tends to be cautious about road-testing such features as new pricing models before putting them on the system. But I was surprised by just how sophisticated the derivatives coverage was. As one user put it, "You'd never say that Bloomberg is going to be the first to do something, but when they do it you know it'll be solid."

Looking at the fixed-income screen (GOTK) gives one a clear impression of the system's scope. Users can value bonds against swap curves, both historically and on an asset-swap basis. Although Bloomberg was not designed as a dedicated derivatives feed and trading system, the swaps coverage comes pretty close. The range of contributors is impressive and it has the analytics to match-fixed-versus-floating, currency swaps, callables, whatever. Complete swap portfolios as well as individual swaps can be entered, tracked and revalued at will. Equity derivatives coverage is equally comprehensive, with every conceivable market covered in detail.

Value-at-Risk, curve volatilities and other risk management numbers are also available, but I don't think Bloomberg is (or pretends to be) a complete risk management solution. That's fair enough, as the Bloomberg approach is to combine respectable depth with enormous breadth. Getting involved in the ramifications of developing an on-line risk management application would divert resources needed for media, data and other analytics.

Those clients who opt for Open Bloomberg have the ability to download both raw and Bloomberg-calculated data into Excel and manipulate it, so if they want to use your own proprietary pricing models, volatilities and curves, they can. On the other hand, they can't (as yet) export the formulas for Bloomberg analytics, which is a shame. Given that some of the pricing models have become virtually the industry standard, this shortcoming could allow the competition to steal a march in the longer term.

Being a prominent member of the paper-based flat earth society, I feel uneasy without a systems manual. Bloomberg's view is that since virtually every week brings a new release or modification to the system, any printed manual would be out of date as soon as it was released. Users can instead print out any relevant sections of the online manual that they require, to ensure timeliness.

Support was outstanding, and the staff was both knowledgeable and helpful. There was always someone available; no hanging on listening to canned music or being shunted around voice mail hell. Nothing was too much trouble. When a flaky dial-up connection on my workstation acted up (not a Bloomberg problem), a member of the support team immediately dispatched a Bloomberg Traveler laptop to my office, two hours from Bloomberg's London office.

Bloomberg includes in its price on-site training for new clients, yet the system is fairly intuitive in its own right. It took me about an hour without assistance to get up to speed on the basics. Within four hours, I was happily constructing improbable swaps. I found both the range and depth of Bloomberg's derivatives coverage impressive. Granted, if you're an ambitious quant on the cutting edge of derivatives innovation, you might find an odd gap to complain about, but for mere mortals this is the right stuff.

Jose Ohayon
Managing director, OTC

Overall, Bloomberg's derivatives offering is quite comprehensive, combining superior market information and securities descriptive data with a set of powerful analytics and risk management calculations.

Market information

Thanks to the breadth and depth of its information, Bloomberg uniquely provides the ability to analyze domestic and international fixed-income and derivatives markets to identify trading and arbitrage opportunities. World markets swaps, money market rates, and generic and customized pages are available to compare historical spread between countries, track live rates, or perform basis spread analyses. Bloomberg also offers a broad selection of international swap curves (Eurodollar, international and non-LIBOR) as well as user-definable swap curves that can be manipulated and stored for scenario analysis and what-if valuations for swaps, caps, floors, swaptions, bonds and repo instruments. These are complemented by a wide range of international yield curve volatilities that provide the foundation for pricing swap instruments with embedded options. Finally, a multitude of spot and forward yield curves (governments, corporates, agencies, swaps, Euros) allow users to perform multiple curves comparisons, intermarket spread analyses, and historical and future analyses of the bond and swap markets.

Securities descriptive data

The ability to create, store and retrieve derivative securities descriptive and pricing information via predefined templates for swaps, swaptions and other complex swaps structures, combined with the capability to distribute these "deals" to customers, make the Bloomberg product a powerful yet simple "offering" system. Securities coverage includes bonds, caps, floors, collars and FRAs, as well as vanilla, amortizing, cross-currency, callable/puttable, O/N indexed swaps and LIBOR/non-LIBOR basis swaps. Securities can be marked-to-market and what-if analyses can be performed and saved on the user's private database. Market conventions are also stored, including the curves selected for valuation purposes.

Calculators

Bloomberg offers many calculators with predefined market and data conventions relying on both standard and nonstandard calculation methodologies. These calculators can be combined with user-defined yield and volatility yield curves to allow for exhaustive what-if analyses.

These calculators include swap calculators for all swap structures; cap/floor/collar calculators for premium, implied volatility or strikes values using Black-Scholes FRA calculators (including broken dates); calculations using cash/LIBOR and Eurocurrency rates; swaption calculators for market value and sensitivities calculations for European-, Bermuda- and American-style options using Black, Black-Derman-Toy, log-normal and normal nonreversion calculation methods; and cash flow analyzers for valuing swaps cash flows with user-defined curves.

Analytics

Bloomberg analytics enable users to perform a wide variety of analyses on securities stored in both the user's and Bloomberg's private database. They include swap horizon analyses; swap cash flow generation and NPV calculations; rich/cheap analyses and current/future values for FRAs; swap hedge ratio analysis; TED and asset swaps spread analysis; swap futures duration hedging; matrix pricing; futures strip analyses; and bond-versus-swap yields analysis.

Options analytics include premium, volatility and sensitivities calculations for equities, indices, bonds, swaps and currencies, including exotic options (Asian, barrier, digital, compound, chooser and executive options). All these analyses can be performed on previously stored securities using a variety of standard and user-defined yield curve scenarios and benchmarks.

Portfolio analysis and risk management

Bloomberg offers both portfolio analysis and tactical risk management capabilities. Users can create, store and retrieve individual securities or portfolios, and perform position, P&L and Value-at-Risk calculations.

Portfolio historical, present and horizon analyses offer position, P&L, cash-flows, price sensitivity and return calculations under different currency, yield and volatility curve scenarios. Yield, duration and convexity scenario analyses allow users to perform strategy adjustments on their portfolios. Stress-testing of portfolios is made easy thanks to the available market and securities information and their complete integration with analytics such as the Portfolio Rebalancing and Trade Entry Idea capabilities. Several formats for portfolio pricing, structuring and risk display are available. The product gives users up to 10 risk reports for each portfolio.

Bloomberg's PVAR enables users to assess risk across a group of securities using both historical and option analysis. Bloomberg performs VAR calculations using the Variance/Covariance methodology of JP Morgan RiskMetrics with either JP Morgan's data or its own. Both portfolio summary and individual securities VAR calculations are available, and allow for intracurrency, intercurrency and intermarket risk reduction. Finally, a PCAD (portfolio capital adequacy) report is available.

Once more, the strength of this analysis stems from the availability of a comprehensive security database and cash-flows for all types of instruments, historical prices to track volatilities and correlations (correlation matrices), and option analytics to handle nonlinear assets. For the latter, Bloomberg calculates the delta, vega, gamma and theta using a modified Black-Scholes model for European options and a binomial price tree for American options.

Overall, Bloomberg has delivered strong derivatives analytics and tactical risk management capability by seamlessly integrating Bloomberg market and securities data with powerful analytical, modeling and archiving capabilities. The ability to create curves, swap structures, analyses and risk reports and to distribute them in a simple way is one of the most powerful features of the system. The product provides a standard, easy-to-use interface but lacks flexibility, particularly with respect to capturing transactional data from external systems. In conclusion, the product meets the needs of a broad audience of risk practitioners and managers.

Jose Ohayon is a managing director of OTC, a consulting and software development organization. Ohayon has more than 12 years' experience in the strategic analysis and development of electronic order routing and execution, trading and analytical systems in the foreign exchange, fixed-income and derivatives markets. He holds a master of science degree in computer science and an MBA from the Massachusetts Institute of Technology.

Bloomberg

One system: $1640/month
More than one: $1140/month per system; slightly higher in continental Europe and Indonesia
Terms: two-year lease; 1.5 months try and return.
Headquarters: Bloomberg Financial Markets, 499 Park Avenue, New York, NY 10022
Contact: Leslie Van Orsdel, derivatives product manager
Phone: (212) 318-2244
Fax: (212) 318-2080
email: orsdel@bny1.bloomberg.com
London: City Gate House, 39-45 Finsbury Square, London EC2A1PQ, 44-171-330-7500

Clients' Opinions

Dan Golla
LIFFE floor trader (Bund and Euromark)

Although I don't use Bloomberg now (it isn't really appropriate when you are looking at only a few markets), I used it extensively when I was a member of the OTC bond option team at UBS. For that it is excellent, because you are looking at curves, repo, volatility and history at the same time. It's OK for daily charting, but not ideal for really fast markets like these. A great advantage is the extensive database for bonds, options, volatilities and so on. There is really nothing like Bloomberg, and for a prop trader in a bank it's ideal.

Gavin Gilbert
Associate director, Bear Stearns, London

It's the best and cheapest thing out there for people who can't or won't build their own system. Though I trade using models that I built myself, I use all the Bloomberg functions as well, because clients and other traders also use them. Data delivery is timely-it may not always be the fastest, but if you want it much faster you'll have to go and work on the floor.

Bloomberg is careful to test new functions thoroughly before general release. I do quite a lot of this for them, and after an initial trial period we sit down and discuss any problems. Even after extensive additional testing, it will only go on conditional release with a disclaimer attached, warning that it's still under test.

Jim Liu
Vice president, Union Bank, California

We've been using Bloomberg for just over a year. We use it for interest rate swaps, currency swaps and pricing interest rate options such as caps, collars and floors.

It's also used for such non-vanilla things as swaps in arrears, amortizers and roller coasters. I'm quite happy with it; if there is ever a problem I send Bloomberg a message and it is corrected in a few days. Once, for example, the amortization schedule was too short. We asked them to lengthen it and they did. As a result of all this feedback, the software is getting better all the time.

We're thinking of moving over to the Open Bloomberg system. Now, if I create mark-to-market values for 40 swaps I have to manually plug those numbers into a spreadsheet, which is hard work.

I particularly like the portfolio feature, which allows users to put everything together and run the entire portfolio in a few seconds. That saves a lot of time. Users can even schedule it to do that automatically.

The system also has the ability to analyze multiple scenarios, so users can apply shocks to the yield curve and recalculate mark-to-market values instantaneously, for example.

William Hugel
Assistant vice president, Fuji Bank, New York

We use the trading system for our capital markets and securities trading groups. I've personally used Bloomberg since 1988 in both the open and closed versions of the product. I think its big strength is the ability to benchmark models against something that many other users and clients are running their systems off-or trading with. Because there is a wide acceptance of Bloomberg's models, they have credibility. The models offer a huge range of analysis, and Bloomberg is the only company that can provide you with all of this. In order to reproduce it yourself, you'd have to hire your own quant.

It's still early, but Informix's latest offering seems to serve up just what the doctorates ordered.

Universal Server is Informix's object-relational database management system (DBMS), which allows complex data types to be defined and handled efficiently through the use of purpose-built installable software modules. The driving force behind this innovation has been the fact that real-world data queries not only involve multiple data types, but often have complex internal structures. Though the product has numerous applications in the general business world, it has a number of specific features that make it particularly appropriate for financial markets, especially in risk management and derivatives. Universal Server runs on both UNIX and Windows NT.

History and background

Universal Server evolved from parallel developments at Informix and Illustra Information Technologies. In the early 1990s Informix made the decision to invest significant resources in "re-architecting" its existing OnLine Dynamic database server. The objective was to enable the server to use a multithreaded multiserver architecture called dynamic scaleable architecture (DSA), which uses lightweight threads for user tasks rather than full OS processes. Apart from allowing the database management system to use system resources more efficiently, this would also provide support for symmetric multiprocessing platforms and parallel database operations. In particular, the ability to partition data across multiple disks and to parallel-process such things as index creation and backups meant that the OnLine Dynamic server was capable of efficiently handling extremely large databases.

At the same time, Informix took the precaution of building hooks into core DBMS components so that the OnLine Dynamic server would be able to accommodate such things as user-defined data types and functions in the future. Specifically, Informix chose to adopt the object-relational model from the Postgres database research project at the University of California at Berkeley.

Postgres subsequently evolved into the Illustra Server, an extensible object-relational DBMS that could query and store complex data types as first class types as well as mix them with traditional data types within a relational database. A prominent part of the technology was the ability to use installable software modules (called DataBlades) that could introduce the server to new and complex data types, which could be user- or third-party defined. Informix, wanting to be first to market with an extensible DBMS, decided to acquire Illustra Information Technologies in early 1996 to shorten the development time span.

Product structure and principal features

The principal feature that sets Universal Server apart is its extensibility. A major factor in that extensibility is the DataBlade, which contains a collection of code and database objects that enhance the database server's performance by allowing it to handle specific and complex data types. The result is that server speed performance on these new data types is indistinguishable from that on existing built-in formats. Universal Server can thus enable the user to query the database to return the value of a portfolio of swaps and options. Currently no other database has the ability to directly represent, manage and query complex financial objects.

Informix has already provided some 20 different DataBlade modules which can handle data that includes anything from video to time-series information. In addition, the company provides the DataBlade Developers Kit, which clients and third-party developers can use to produce their own specialist data types. The DataBlade API allows any DataBlade to be snapped directly into the Universal Server. Informix proposes to license the API to third parties including other DBMS developers in the expectation of it becoming a common standard. Tandem has already announced its intention to add the DataBlade API to its NonStop SQL database manager, which will allow it to use any DataBlade that works with Universal Server.

The time-series DataBlade that Informix has developed is of particular relevance to financial markets. The number of individual records that this solution will produce for a financial instrument database is a fraction of that produced by a conventional RDBMS.

My view

The product is still in its early days. No investment banks have completed an implementation of Universal Server for risk management or derivatives as yet. Having said that, there are a number of banks already using either Illustra or OnLine Server who are likely to switch over to the combined technology of Universal Server in the next six to 12 months. There is no doubt that Universal Server is conceptually an attractive proposition; the ability to create one's own specialist data types with no performance penalty attached (the reverse, in fact) is significant. Even more significant are the architectural implications for risk management solutions when complex financial queries can be executed directly on the server. This is technology that has the potential to change the workflow from front to back office for derivatives trading. Informix's marketing claim that Universal Server's technology effectively provides a hardware upgrade to your data server is probably going to prove correct. The amount of redundant data processing that DataBlade technology should eliminate is justification enough for that.

The other obvious benefit will be a substantial reduction in network loading. The current situation in many banks is to have increasingly powerful user workstations with trading or RM models installed on them. The snag is that this involves calls to the server for binary large objects (BLOBs), which are then manipulated on the workstation before being returned to the server. Given that these may easily exceed five megabytes apiece, it doesn't take many power users to gridlock a network. Universal Server allows this data processing and manipulation (if desired) to be carried out on the server, with only the results of the analysis being sent to the workstation. With Sun's heavy promotion of its "thin client" network concept and the large number of Sun servers already in use in banks, the timing of Universal Server's release is, to say the least, fortuitous.

Independent Reviewers

Doug Crawford

With the addition of the Universal Server product line to its existing DSA server architecture, Informix offers intriguing new possibilities for designing and managing complex financial and risk management systems. The Informix Universal Server (IUS) greatly expands the range of traditional relational database systems by supporting complex data types and functions at the server level. The IUS combines the historical strengths of relational systems (speed, query facilities and so on) with the advantages of object-oriented systems in handling complex data. Complex data types can be manipulated as objects defined to the server rather than as BLOBs. Thus developers can use a proven query standard against complex data types.

Processing efficiencies

Informix has long used its dynamic scaleable architecture (DSA) as the basis for providing scaleable high performance in a transaction-processing environment involving traditional data types-numerical, decimal, floating point and so on. DSA uses server-optimized parallel, concurrent scan, join and sort operations to reduce processing times. This feature increases efficiency since the major components of query operations are broken down and processed in parallel, with the results being passed to the next operation as early as possible. Thus, the initial record scan is broken down across multiple processors and, as soon as results start to become available, join operations begin, followed by sort operations. In many systems the operations are performed consecutively. With the addition of IUS, applications can benefit by utilizing client or third-party-defined complex data types stored at the server rather than application level. One result will be applications with a naturally thinner client layer.

Development efficiencies

Developers can optimize their productivity through increased use of standards and reusable code. Since IUS supports complex data types and related functions defined as natural objects, developers can now use SQL to query complex data. Consider the query "Give me stocks whose 30-day moving average is higher than their 60-day moving average and recent articles on those stocks written by the analysts at Goldman Sachs and Morgan Stanley." In a traditional relational system, this query would be both a user's and a developer's nightmare. The time it takes the programmer simply to understand the bounds of the request would take several times longer than the user expects to get the completed answer back. Not to mention that such a query is not static; that is, the programmer would probably have to tweak his or her code if the same query ran a month later. If, on the other hand, the developer can define time series and text as data types and moving average as a function within the server, the query utilizes standard SQL with no sets of complex mapping required to retrieve and manipulate the data. By supporting complex data as objects, developers can take advantage of such object-oriented benefits as inheritance and overloading. Thus developers might define a simple swap object by defining a particular time-series data type for cash flows along with functions to calculate fixed flows, floating flows and price. To add a currency swap, object developers can create a subclass of swap and need only add a function for cross rates; everything else is inherited from the swap object.

Development flexibility and extensibility

The merger of Informix and Illustra and the resulting DataBlade product offers developers the ability to choose among a variety of solutions to a given problem involving complex data types. DataBlades are packages that contain definitions of complex data types along with all the functions for manipulating the new types. DataBlades "snap" into IUS through a DataBlade API that allows the new data types to take advantage of IUS/DSA features-SQL3 parser, query optimization, access methods and so on.

Architectural flexibility

IUS allows system architects greater flexibility when designing database structures and related applications as well as when converting from existing systems. While IUS naturally supports a movement toward a thinner client/fatter server architecture, users can choose to move anywhere along the spectrum. A good example would be an existing client/server system using a standard SQL2 database engine moving to an Informix SQL3 database engine. The easiest approach would be to simply swap the SQL3 engine for the SQL2 engine and leave the client and mapping layers virtually untouched. Most work would involve the differences between SQL3 and SQL2 (differences in stored procedure handing, updating, client API and so on). Moving along the spectrum, users could start to define new data types to the server, thus reducing the client/mapping layer and providing more efficient handling of complex data.

Implications

The features described earlier have a number of implications for system designers in the derivatives area by effectively blurring the traditional distinctions between applications and database. Many problems in typical derivatives environments are caused by the necessity of (and subsequent errors in) reconciling multiple versions of data. In many areas the front-office pricing system and models are completely separate from the back-office systems, creating an often difficult reconciliation between P/L numbers. Another problem often occurs in the confirmation process. The documentation area must typically prepare a confirm, send it to the counterparty and then track its progress until signed and returned. IUS offers developers an opportunity to reduce the number of data sources (and thus the number of reconciliation processes) without sacrificing flexibility. Developers can define swap objects with different price functions for different users-multiple sets for a trader who is trying to understand a market, and another defined by the risk managers as the standard curve for end-of-day valuation. Once the price functions are signed off by risk management, the source data can be reduced to a single point. By supporting complex data such as text, IUS allows greater flexibility in designing systems allowing personnel to compare different versions of a document to track changes or to alert a trader that a certain counterparty has documents outstanding such that trading restrictions exist. In addition to increasing systems productivity, if properly conceived and implemented, the IUS model could be used to greatly reduce such risk factors as operational risk, credit risk and fraud.

Doug Crawford is an independent consultant based in Norwalk, Conn. His practice focuses on database issues in the development and management of derivatives pricing, operations and risk management systems. He has been involved with database consulting, derivatives sales and trading, running derivatives operations and developing derivatives systems for the last 18 years. He holds a bachelor of arts and an MBA from Yale University.

Bryan Johnson

The traditional Relational Database Management System (RDBMS) was developed to handle information stored using a limited number of simple data types, notably integers, scientific floating-point, character strings, date/time and money. Elaborate but reliable mechanisms have been developed to allow for the data in the database to be accessed and updated by many users simultaneously while ensuring that the consistency of the data is maintained.

The largely industry standard structured query language has provided the means to make relatively complex queries on this sort of data. But a simple SQL query alone cannot answer the question, "What are the 13-week average sales for our top-five profitable products?" The profitability of all products must first be calculated before they can be ranked, and, for the top five, the calculation of the 13-week average sales must be performed. In order to keep processing of the query within the database engine, special functions, which have knowledge of the data, need to be invoked by the engine. Currently some of this processing ends up being performed in application code outside of the database engine, breaking up what could have been a single query into several steps and making the whole total query more difficult to maintain.

Organizations are now increasingly requiring their information systems to process complex data types such as images, audio, time series or dynamic web pages. Complex data types are often referred to as "objects" because they represent complex internal structures and usually require special functions or methods to create and manipulate the data. To date, RDBMS products have not handled complex data types well, relegating them to second-class status, by storing them as BLOBs. Although the database engine can store the complex data, as it has no understanding of the structure of the data in a BLOB, it cannot perform content-based queries on them. Queries on the complex data or objects largely have to be performed outside of the database engine by applications that do have knowledge of them.

It has become one of the holy grails of the database industry to provide SQL-based object-relational database systems that provide unlimited support for complex data types such as video, spatial, geographic and web, while retaining the performance, security and transactional features now expected of a RDBMS.

Universal Server can have any number of DataBlade modules installed. Customers will be expected to mix and match modules from Informix and third-party vendors or to write their own to meet unique business requirements. Informix has invested in a DataBlade Developers Kit to make it as easy as possible for third parties and customers to create, build and deploy their own DataBlade modules. Informix plans to have 250 available by the end of 1997.

While Informix sees the DataBlade architecture as a key benefit of its Universal Server, others have claimed that the approach is flawed. Grafting a third-party plug-in onto the kernel is seen as risky, because if the plug-in malfunctions, the database could crash. Oracle has announced that it will only embed in the kernel of its relational database engine the support for complex data types it develops in-house. Extensibility provided by third-party vendors, meanwhile, will reside outside of the database engine, in what are to be known as data cartridges.

Integrating the DataBlade modules within the database engine should allow for existing DBMS functionality such as transactional management, backup and recovery to be applied to the data stored in the new data types. This should increase the safety and reliability of the data. The stance taken by Oracle would seem to increase the problems of maintaining data consistency, as it will be stored across a number of independent databases, with the ensuing problems of multiphase commits.

In order to minimize the chance of reliability problems with DataBlades, Informix has established a certification process to ensure that a DataBlade module works with the Universal Server and all other certified modules, and that the DataBlade code does not affect the integrity of the DBMS. It is said that the certification process evaluates the DataBlade module in a number of areas, including code quality, test coverage, documentation and conformance to guidelines for safe execution.

Object-oriented programming is becoming the mainstream manner of application development. It is important that the underlying database should store data in a form that is as close as possible to the way it appears in the application. DataBlades provide the means to do this. The fact that it was shipped on schedule indicates the strength of the underlying object-orientated architecture the two original products shared. I look forward to having the opportunity to build an application using Universal Server.

Bryan Johnson, a British citizen, is a senior technical consultant at JCC (America). He holds a bachelor of science degree form Lancaster University in Great Britain, and has worked for the past 17 years as a database specialist on several large computer projects for clients including MCI, the government of the Netherlands, Digital Equipment Corp., ARAMCO and Price Waterhouse.

Client Opinion

Alexander Pastron
Assistant vice president of front-office technical services, Credit Suisse First Boston, New York

I think the Universal Server is a powerful concept. We are able to combine the standard relational data and the nonstandard object-oriented data in the same queries. This saves a lot of work on optimizing queries and gluing different systems together. Without Universal Server, the standard method would involve storing descriptive and time-series data in two separate databases. This causes a lot of problems for application developers; they have to know two different interfaces. It also causes problems when maintaining data for data integrity: How can you have the price for the security if you do not have a description of the security?

It's important that we are not tied to existing DataBlades, but that we have the option of writing our own code that will run in Universal Server. The extensibility of the server is the key feature, because while some existing DataBlades may not be as strong as existing Sybase functions, we can modify and enhance them ourselves, which you can't do on Sybase. You also have future proofing; if a previously untaught piece of technology suddenly appears, the company that specializes in it can simply write a DataBlade that can be ported to Universal Server far more quickly than to a standard closed-relational database.

Contacts

U.S.: James Maldonado, Director, financial services industry marketing, Informix Software Inc., 805 Third Avenue, New York, NY 10022
Phone: 212-546-0645
Fax: 212-753-4210
email: jem@informix.com
U.K.: John McKee, Sales manager, finance and banking division, Informix Software Ltd.
Phone: +44-171-395-5950
Mobile: +44-385-233674
Fax: +44-171-395-5901