Real-time Analytical Processing

Introduction: strategic advantage
"A well kept secret"
Why has it taken so long to discover TM1?
TM1 has a 13-year headstart
How TM1 works
The Signature Group signs on for ease-of-use
TM1 Technology Leadership in the OLAP market

What it is, and how it gives companies a strategic advantage

As aggressive companies seek more efficient, more powerful ways to implement strategic decision-making, many of them have discovered TM1 Real-time Analytical Processing. TM1 is extraordinarily powerful, scalable and performs so well that real-time operational and strategic decision-making is possible with complex data models.

While these companies are just discovering the benefits of the TM1 RAM-based of OLAP, we have been developing and offering it since 1984. TM1 Software has been one of the lone advocates of RAM-based OLAP, pioneering its use for time-critical and closed-loop, real-time, read/write applications. We quietly attracted a growing and loyal customer base of customers world-wide (now numbering more than 2,000 companies). However, the market did not realize or believe that it was possible to perform complex calculations of large data sets in memory (in real-time). It seems counter-intuitive, and, in fact, it is not easy to do. If it were easy, someone else besides Applix TM1 would have been doing it. No one was.

"A well kept secret . . ."

Richard Creeth and Nigel Pendse, the authors of The OLAP Report, were the first "industry experts" to reveal TM1 to the world, calling TM1 "a well-kept secret," in their first edition (August 1995). "Few other OLAP products can compete with TM1's ability to dynamically recalculate large models in a multi-user environment," the report said. "Thus, in contrast to most OLAP products, the TM1 engine allows true multi-user read-write, with an audit trail and instant recalculation of results if data changes."

In November 1996, the "secret" made headlines when Hyperion Software signed an agreement with TM1 Software to use TM1 technology for Hyperion OLAP. Now with the February 24th agreement between Comshare and Applix TM1, both of the major players in the financial decision support application marketplace are using TM1 technology.

Why has it taken so long for TM1 to be discovered?

Perhaps because TM1 Software had long been content to be a technology company and placed little emphasis on marketing and product distribution channels. That changed in 1995 when founder Manny Perez realized that marketing companies with inferior products - Arbor in particular - were positioned to dominate the OLAP market. So he brought in marketing and sales experts who could spread the word about the benefits of TM1.

TM1 has a 13-year headstart

The basic design of TM1 has remained intact since 1984. Earlier, Perez, a mathematician, tried the same pre-calculation designs that are widely used today, before switching to the RAM-based, real-time analytical processing approach. Perez, now TM1 chief technology officer, has worked closely with global customers since 1984 to evolve and enhance TM1. Early customers included Exxon, Equitable, Bristol-Myers Squibb and Occidental Chemical.

"The TM1 architecture has withstood and benefited from the test of time," Perez says. "Its strength comes in no small measure from our having started early. We've been listening to the market and responding with a series of products that deliver greater and greater scalability and flexibility. We have a cycle of ever-increasing power, based on demand from clients, that just keeps pushing the envelope. Our challenge all along has been to keep up with our customers' needs, and we have - for 13 years. As a result, we have technology that no else has."

Perez describes developing RAM-based real-time OLAP as "far from trivial. But while we have done a lot of clever stuff, our biggest asset is the experience of so many years of practical application of the technology. It is our sustainable, technological, competitive advantage."

How TM1 works

The TM1 real-time analytical processing engine departs radically from batch-oriented OLAP products that achieve fast response time by pre-calculating consolidations and derived values, and storing results on disk. TM1only stores the lowest level input elements in the model, and does so very efficiently, about 10 bytes per record. When a user asks for calculated numbers, TM1 processes them on demand, using patented algorithms that The OLAP Report described as "unusually efficient." The result is Real-time Analytical Processing.

"The technology is based on keeping all the active cubes (logically, if not physically) in RAM," The OLAP Report says. "Although this sounds ambitious, the algorithm succeeds in keeping a remarkably consistent 100,000 numbers per megabyte of RAM. The largest database that Applix is aware of has a size of about 500Mb, roughly 50 million actual cells. As TM1's approach is not to precalculate if possible, or only to do so to a minimal extent, this probably corresponds to roughly a billion plus cell database in an OLAP product with a pre-calculation approach, which would occupy several hundred GIGABYTES of disk space, a size that few of the disk-based products reach in practice."

One of the biggest problems for OLAP users is data explosion - the huge expansion of data caused by batch pre-calculations. The TM1 real-time processing approach eliminates this problem. Our "just-in-time" processing makes more sense, since 99% of the numbers in an analytical application are never needed. So why waste computer resource and time waiting for a batch process?

In addition, incremental updates are available as soon as they are loaded in TM1. The pre-calculated approach can take hours or days if large data sets are being exploded, not an uncommon practice.

From a user perspective, the TM1 design facilitates collaborative, workgroup computing whether across the room or across the world. When a change is made, everyone sees the result. There is no interruption. No waiting for an answer. Everyone stays on task, focused and moving toward the goal of making the right decision and executing the right plan.

The TM1 multi-cube architecture provides two important benefits for large organizations - scalablity and a distributed network architecture, both of which make TM1 applications deployable more widely and more easily. Which would you rather distribute: a huge cube or logical cubes that are distributed and accessed across many servers?

The Signature Group signs on for ease-of-use

In addition to flexibility, TM1 is truly easy to use for anyone with spreadsheet skills, unlike approaches that purport to be easy, but require a steep learning curve. If you build applications using Excel or Lotus, you can build models in TM1. If you use spreadsheets, you can use TM1.

"You can write a formula and get a number in TM1," said Greg Shireman, Director of Financial and Business Information Systems, at The Signature Group in Chicago. "You can't do that in Essbase. The strength of TM1's integration to Excel and Lotus …may be the strongest feature as far as an accountant is concerned, especially for budgeting and forecasting. The seamless spreadsheet integration is a very important part of the product."

TM1 is such a natural extension of Excel that it's easy to automate spreadsheet reporting, format reports and write macros that save many man hours, Shireman said.

"I can take educated spreadsheet users, give them a half a day of education on TM1, and they can produce reports immediately."

TM1 Technology Leadership in the OLAP market:

We have been "first" to the market with many innovations including:

First product that met E.F. Codd's original 12 rules of OLAP
First "intelligent calculation in real-time" with no scripting
First memory-centric, RAM-based engine
First multi-cube
First multi-server
The only vendor besides Arbor (who commissioned the study) to subject itself to a full appraisal by E.F. Codd & Associates, and to pass
Pioneered the use of multidimensional OLAP databases linked to spreadsheets, even before Windows was available
Delivered the first multi-user, multidimensional database linked to standard Windows spreadsheets (Excel and Lotus)
First OLAP vendor to provide an external API
First OLAP tools aimed at user-builders (knowledge workers and power analysts)
First true client-server OLAP
First OLAP product to calculate real-time feeds
First intelligent integration to Web spreadsheet