The Magic Cube
Select the right database for your next generation high performance application
What is Magic Cube?
Magic Cube is a simple, visual tool created by Versant to help you compare database technologies. It looks at the principle application requirements along the three axis that impact the overall system performance: data volume, data complexity and concurrency. Each axis in the Magic Cube represents one criteria. You can map the requirements on a continuous scale from 0 (minimal requirements) to 1 (maximum requirements).
.jpg)
Certainly, there are additional requirements that may be as important as volume, complexity, and concurrent access. For example, support for your development environment, availability of administrative tools, support for 24/7 production, backup, disaster recovery, and business continuity. The Magic Cube leaves these additional considerations out as they will not make a difference if the principle performance requirements are not met.
The Magic Cube dimensions:
| DATA VOLUME | The DATA VOLUME that must be managed is a moving target, as we are facing an unprecedented data tsunami and exponential growth of data. What’s Big today will be average tomorrow. Look ahead; extrapolate your requirements for the years to come. | 
ESA Case Study - 6 to 7 Terabytes of raw telemetry data, decompressed, stored, replicated worldwide. |
| DATA/MODEL COMPLEXITY | The DATA COMPLEXITY, including the relationships between data objects, continuous to grow. A recent IBM study states, that the “Majority of CEOs identify complexity as their organizations greatest challenge”. Face this reality. | 
EidosMedia Case Study - complex, unstructured needs maximum flexibility. |
| CONCURRENCY | The amount of CONCURRENT DATA transactions (the number of users and the level of concurrent access to the database especially involving conflict resolutions) is growing exponentially driven by Mobility and the Consumerism of IT. How much latency can you afford and at what cost? | 
China Telecom Case Study - 480,000 queries and 1,000 update transaction per second. |
How the Magic Works
We have evaluated 4 popular data management categories and populated the Magic Cube for you (see image above). These categories are:
| Data Management Category | Data/Model Complexity | Concurrency | Data Volume |
| File Systems | Minimal | Minimal | Marginal |
| Key Value Stores | Marginal | Fair Amount | High |
| Relational Database Systems | Fair Amount | High | Fair Amount |
| Object Database Systems | High | High | High |
The Magic Cube assumes a product from any of these categories will become an excessive burden to deliver reliably or will outright fail if the product requirements evolve outside of the boundaries for which it was designed.
For example, a flat file may work reasonably well as long as it can be loaded into memory (low volume requirements) and there is no concurrent access to any of the elements stored in it (no concurrency requirements). However, as soon as one of these requirements changes, the flat file technology will fail and must be replaced.
Each category represents several varying products. For example, our object database products VOD and db4o both belong to the Object Database category. However, db4o is designed to be an embedded, small footprint system, whereas VOD is designed to run carrier grade enterprise databases. Those design requirements heavily influenced the architecture, implementation and tooling of these systems.
Work The Magic Cube Yourself
Step 1: Find out what type of database category has the best fit for your application profile today.
Step 2: Look ahead at your growth scenarios. For example, what are the minimum, and what are the maximum requirements three, five or even ten years ahead? We believe it is fair to assume that while hardware technologies continue to advance, they may not advance as fast as your needs grow for data management.
These first two assessments give you a spectrum of categories and products to evaluate.
Step 3: Now you can further narrow down the selection of products to evaluate by looking at other important criteria, such as tooling, etc.
Take a look at successful applications powered by Versant.