AlgoQuest System is now owned by Celinvest
Celinvest bought all rights and property of the invention of Elie Ouzi Koskas: a super fast relational database management system, named AlgoQuest System.
After several years knowing Elie Ouzi Koskas, and initiating several discussions regarding a potential collaboration between the inventor and Celinvest, we decided to invest in the potential of his invention.
The technology is a jaw-breaking fast database management system, answering the needs for large volume combinatory queries to get answers that are not based on statistics and that can be delivered in acceptable timeframes. The tipping point for us was when we compared the technology with an IBM-Netezza installation in one of the largest private banks in France, on a query that took the bank’s optimized platform over 11 hours to deliver. On exactly the same database, with no further preparation of the dataset, and with a simple laptop, we managed to deliver the exact answer with 100% accuracy in less than 12 minutes.
This is the moment we decided to invest massively in the technology and the development of its potential.
Key Milestones
First encounter with Elie Ouzi Koskas
First financial investment from Celinvest and creation of AlgoQuest System SA to market the technology
Celinvest delegates Gilles Celimene as CEO and Administrator of AlgoQuest System
Following liquidation of AlgoQuest System SA, Celinvest legally buys all assets of the company including the technology, intellectual property and all applications (present and future)
Details and organization of the project between Celinvest and inventor Elie Ouzi Koskas
Back in 2010, Elie Ouzi Koskas had already spent several years conceiving the database engine that would then become the core of the technology. Yet, at this moment, the technology was efficient but not mature enough to be marketed directly to customers. This is the reason why it was decided to create the company AlgoQuest System SA, to finance the development of the tool, and to market it and sell it to potential customers.
The technology is built around two main parts: the engine (called AlgoQuest Engine) and the tools (called AlgoQuest Tools). Roughly speaking, the Engine is the part where all calculations happen, and in which queries are handled. The Tools is the user interface of the system, allowing all interactions with the Engine: adding a new database, submitting new queries, updating content, collecting answers from the Engine, and making them readable for the end-user.
The main logic behind the Engine, and the key invention from Elie Ouzi Koskas, is a new way of managing data through a new approach of mathematics: the Engine only manages unique iterations of the data in databases, allowing to avoid repetitions in the calculation while still taking them into account for the final result. As a consequence, the Engine proves itself to be extremely efficient and creates a huge difference with standard database management systems for the management of combinatorial queries with high volumetry, answering SQL queries much faster, with 100% accuracy and no statistical approximation.
This engine relies on a fully new mathematical approach to be so efficient. Because it was built from the ground up with this new approach, many constraints known by all other database management systems (such as data indexing for instance) are not relevant anymore with this solution, making it much more agile and efficient.
This is one of the main discoveries behind the technology: the team proved that looking for results in a relational database is equivalent to looking for paths under constraints in a graph. Usually, finding the optimal path in a graph is a very complex operation subject to the combinatorial explosion – Elie Ouzi Koskas created an algorithm that calculates the optimal path in a graph, in a timeframe linearly proportional to the size of the graph representing the database, and therefore not subject to combinatorial explosion.
If we consider the database as a graph, this technology is then easily adapted to find results in a relational database. This technology has been the main driver for our investment in this project, and this is what was fully sold to AlgoQuest System SA by Elie Ouzi Koskas, and later fully acquired by Celinvest, including all intellectual property and applications of it.
The main challenge was then to make this revolutionarily powerful Engine usable by users, as this Engine would require SQL queries to be transformed and filled in entry parameters, and would not export a human-readable format for the results. This is why Tools was conceived and would become the interface and management part of the overall system: we hired then freelance developers in Romania to support the development of the Tools part of the software.
In the attached example here, the Tools interface was made to compare the performance of the technology with Oracle Database 19c. The data shown for Oracle is simply taking the results communicated by the Oracle tool, with no transformation, no constraint added, and allowing all optimization preparations by the Oracle tool (index).
These results were achieved on a simple Microsoft Surface and would show that our technology was able to provide the same answer as Oracle Database, generating 54’314’281 results in 3.196 seconds without any cache or any trick, while it took Oracle’s optimized calculation almost one hour on the same computer.
The technology implies the combination of both Engine and Tools to be usable by a human being directly with no other software. This is why the AlgoQuest System overall is the combination of both AlgoQuest Engine and AlgoQuest Tools. Elie Ouzi Koskas is the inventor behind the Engine, while the Tools software has been fully developed by the development team hired after Celinvest joined and invested massively in the project.
The technology’s power truly shows itself through large volume and highly combinatorial queries: the more results are generated, the higher the volume, and the greatest the difference in query processing time. As you may see in below’s achieved results on a single laptop, by multiplying the volume of rows in answer by over 155, we only multiplied our processing time by 1.6, while Oracle Database multiplied its processing time by more than 126. This difference only keeps increasing when we increase the volumetry of the database.
Rows in answer | 349’542 | 54’314’281 |
Our technology’s processing time | 1.97s | 3.19s |
Oracle Database’s processing time | 27.42s | 57min 55.78s |
The largest the volume, the greater the difference: this is the proof that this technology solved the challenge of combinatorial explosion, making it a linear challenge.
The verification of the combinatorial explosion is done by analyzing the run time of traditional relational database management systems, even the ones relying on extremely large calculation power and expensive hardware. Increasing the calculation power only delays the feasibility limit, with decreasing efficiency.
Our approach not only is faster to handle complex queries; it also enables a completely new field of possibilities that were not even considered before.
We better understand how AlgoQuest can be so strong and powerful when we compare the processor usage between the different existing database architectures.
By removing the need for indexes, and getting direct access to data without the need for matching indexes with actual data, we are able to focus the largest part of the processor power towards the actual calculation.
As a result, the technology initially conceived by Elie Ouzi Koskas allows close to 15 times more usage of the processor time for actual calculation: this is a major game-changer. And this is why we invested in it, and later bought all rights of the technology and its applications.
Current situation of the project
Since 2010, Celinvest has been by very far the main financial support for AlgoQuest System (a few other minor investors joined the project as well, yet all other financial efforts cumulated did not equal what Celinvest injected in the technology). Yet, after over 9 years of financing for all the teams to develop the platform, no client purchased any license of the solution.
This is the reason why in July 2019, Celinvest decided to involve one of its founders with deeper technical knowledge as a trusted expert to not only audit the solution, but also review with the teams the development strategy and overall the organization of the company. After a few months of using the solution, confronting it in real market situations, and understanding the potential and limitations of the tool, he came to a question-raising conclusion. The Engine was indeed extremely powerful, yet is lacking so many necessary functions provided by all major actors today on the market (AWS, Oracle, Microsoft just to name a few) that it would take over another 10 years of development just to get closer to the “must-have” features that these solutions are providing today.
As powerful as the solution was at this moment, its true added-value did not compensate for the missing capabilities that were required by potential clients.
This is how we concluded in early 2020 that the way of auto-financing and self-developing the company was not a reliable way to go, and another path was needed.
Celinvest, as the primary financial investor of the company, suggested that the way to further develop the technology would be through a partnership with one of the major actors of the market. Such a partnership would have allowed the AlgoQuest System technology to get access to much larger capabilities, and then develop all the missing features quickly by leveraging the already existing technologies of such a partner. Yet, the discussion on this topic led to a disagreement with the technology inventor and blocked all the development opportunities for the company.
Following this, it was decided that Celinvest would not finance anymore a strategy we did not believe in; yet our investments in the technology remained fully valid. After a few months, it was then decided that AlgoQuest System SA would be liquidated, and following legal procedure Celinvest SA legally bought all assets of the company including the technology, all intellectual properties attached, and all applications of the technology (existing and future) that Elie Ouzi Koskas fully sold to the company. Consequently, all potential future developments of this technology are Celinvest’s property.
Today, and since July 2021, Celinvest is the full owner of the technology from Elie Ouzi Koskas and all its applications following legal decision.
Want to discuss the technology? Get more information? Discuss a partnership?
Do not hesitate to contact us
You may contact us using the attached form, or through the following means
contact@celinvest.com
Address
Celinvest SA
2 rue Astrid
L1143 Luxembourg-City
Luxembourg