New Paradigm


Hello,

Lisa Park would like to present to you Octopus CIP technology and some ideas that lie in its  foundation.The detailed description of   this system is obviously outside the scope of this introduction and would be the subject of further discussions, but here we are going to make a short overview of Octopus concepts. 

The theoretical foundation of Octopus Platform is rooted to multidisciplinary researches and development performed in the course of more than 20 years. The main theoretical results are presented in the Alex Mylnikov PhD dissertation that was completed in 1984 in the Science and Research Institute of Central Statistical Agency of the former Soviet Union. In the following after that years the initial scope of this work was substantially extended with new results in Information Technology, Artificial Intelligence, Data modeling and Data Analysis, Operational Researches, Graph Databases and Semantic Web.

Over the years, as hardware and operating systems matured, the system's design was crystallizing into a more definitive shape.  With the advent of cloud computing, it became clear that the system can finally be realized.  So it was built and dubbed Octopus CIP, which stands for Octopus Cloud Interactive Platform.

The fact that Octopus is a platform rather than a program or a product is important one.  Several attempts to market this system in the form of an application with limited functionality increased our belief in the fact that the main strength of Octopus CIP is in its ability to serve as a platform, into which many various types of applications can be incorporated.  So far, we have successfully used this system for real-time interactive statistical analysis, data integration and enterprise resource planning applications. 

To sum up: we believe that the moment is ripe for a system like Octopus CIP.  Globalization processes are transcending all known thresholds. Big Data is approaching fast on all fronts - business, social, security, military, etc. Octopus methodologies can be independently applied in almost any sector. As long as some data relationship can be formalized, Octopus models can be put to work. 

To lighten things up: you might remember an old story about six blind man, trying to describe an elephant by examining different parts of his body and each one  imagining a completely different animal.  Well, Octopus could have helped the six blind men.  It could establish the shareable knowledge base for them.   It could connect all individual discoveries of each blind man into a graph of meaningful relations.  It could process all models processing elephant part's data on cloud servers.  And in the end, it could predict the true shape of an elephant with a very high degree of statistical certainty. And that's Octopus for you.

Our vision is firmly rooted in this simple idea: it's not the raw data that needs to be collected in analytical databases, but rather metadata and processing algorithms (in our case - models). For us, data semantics lies in a way we use or process data. Since we are using models to handle data processing, models are the formal representation of data semantics.  This approach is a two-pronged strategy.  

First, we can concentrate on detecting trends a lot more rapidly, practically in real-time.  By abandoning querying, we can concentrate resources on processing thousands, perhaps millions of data streams simultaneously. 

Second, collected models can be easily searched by many parameters (source, destination, type of calculation, etc.), in effect creating analytical knowledge base. As a result, the process of extracting analytical insight shrinks down to hours and minutes. ​

With Octopus technology we are going to return a common sense to the data analysis:

  • perform observation only a handful set of important indicators (KPI - Key Performance Indicators);
  • pay a special attention to the historical and geographical  data compatibility (do not compare apples and oranges); 
  • combine data gathering and real-time data processing and analysis wherever it possible;
  • perform most of data processing in the place where data are originated;
  • do these 24x7;
  • go for additional data only in the case if absolutely have to;
  • use adaptive models and analytical methods to reflect natural changes in the dynamic of KPI behavior (changes are the only constants in the World);
  • use multiple models to assess paramenetrs and dynamics of each indicator (choice needs more than one) ;
  • use forecasting to predict a future behavior and assess each of redundant models fitness (use model's forecast to judge it's performance) ;
  • calculate prediction values as an integral assessment of multiple models (optimal decision is always compromise);
  • distribute calculations over multiple nodes and perform them in parallel;
  • raise to the higher level only analytical results with fully specified references to the source data and avoid to send raw data wherever it's possible. 

Our site www.lisa-park.net may prove to be a source of additional information. Also, we have a large volume of technical information, which we would be happy to provide for you, should you request.

We are positive that if we set up a meeting and discuss things in more details, we could shed more light on the Octopus CIP's benefits.

Sincerely yours, iSAS and Lisa Park team

e-mail:  info@lisa-park.com ; mobile: +1 215 696-4383



Octopus CIP is a Cloud Interactive Platform that provides support for building, deploying and managing Model based applications on the Cloud.  The true strength of Octopus CIP lies in large scale information environments for the following reasons:

  1. An Octopus based application, consists of models.  These models can be as simple or as complex as they need to be. Models are completely independent, without any hard links to each other.  Only some external entity can be aware of models and the ways they can be interconnected.  This design makes the system very lightweight and yet, robust and resilient.
  2. Octopus is cloud based.  When cloud advantages are applied to modular designs like Octopus, we beget the elasticity, necessary to run very large scale applications.  Thousands, or hundreds of thousands of models can run in parallel.  Cloud also solves the issue of start-up costs, since it is consumed as a utility (when the cooking process is completed, the stove can be turned off).
  3. Octopus is perfectly suited for utilizing the power of graph databases.  Graph databases bring us the ability to perform lightning fast searches on semantically related data. Graphs are the pathfinders in the age of Big Data. We are currently closely working with the biggest graph database maker in the world - Neo4j.

Octopus CIP is a SOA by design and right now it provides support for the following applications that are available as services:

  • iSaaS is an interactive Statistical Analysis as a Service;
  • iOaaS is an interactive Optimization as a Service;
  • iNaaS is an interactive Network management as a Service;
  • iIaaS is an interactive Integration as a Service.

As any of new and innovative developments Octopus CIP affects many other related areas of science, researches and technologies. The following is a very short list of these domains.

Statistics.
Most of the statistical tools and software are designed to work with already collected data (R, SAS, SSPS, JMP etc.).  Octopus models can perform data analysis in real time, combining data gathering and data processing in one continuous process. All mentioned packages are using script language to describe the actual data processing. Script is tightly coupled with data that needed to be processed and can not be easily reused to process different set of data. Octopus processing models are decoupled from the data and can be reused with any semantically compatible data sets.

Octopus processing models challenge statisticians to research and to develop a set of new methods and algorithms that can support sequential, real time, perpetual data modeling and data analysis.

Modeling and Simulation.
Octopus models naturally support inter model collaborations and parallel executions. Set of simultaneously running models and their interactions can be changed in run-time without stopping a simulation process. The challenge here is to develop recommendations and methodology  to perform data modeling and simulations with such continuously changing and evolving  collection of interconnected models. Some developments in this area were performed as a part of a cell-automata's theory – Octopus platform moves those researches to the  realm of the practical utilization.

Information technology.
All Octopus CIP applications are developed and deployed as a separate VM (virtual machine) on the Octopus Cloud. VM has a minimal configuration to only support the needs of a given application. As a consequence, the deployment of a new version of the application is, in the essence, creation and deployment of the new VM. This approach is based on the following principles:
  • Applications and their VM images are not patchable, but rather disposable. All changes are implemented by replacing the image of an old application’s version with a new virtual machine that contains the new version of an application;
  • Each application is executed on multiple devices simultaneously. This ensures that failure of one instance will not cause failure of the whole application;
  • There are no direct tightly coupled connections between applications or even  components of the same application;
  • Current application status, wherever is possible, persists on the separate physical device. This  allows lossless transition of the application from a broken device on the new one.
Being technically doable this approach however requires additional R&D to make experimental implementations ready for the production deployments.


The slide-show bellow demonstrates the simplicity and power of the Octopus CIP when it is applied to the interactive, real time, collaborative statistical analysis.