icCube Review – Production Quality, Embedded BI

364

Overview

icCube is a high performance, production oriented BI platform. In essence icCube is a single product with a Java server and JavaScript client. It comes with modest but adequate ETL capabilities, a sophisticated OLAP server and the standard language MDX (which is enhanced by over 250 additional functions), easy-to-use reporting, charting and dashboard tools, and the user interface is wholly web based. The OEM market is particularly keen on icCube, because of its modest cost, high performance, and excellent embedding capabilities. In reality icCube is an OEM’s dream, and a solid platform for businesses to implement production oriented analytics.

The company is headquartered in Switzerland, and as such we might expect a pragmatic, well engineered product. This is precisely what icCube is. On the plus side it means that icCube is performant, reliable and flexible. However, while the company offers advanced statistical capability it will have to embrace integration of algorithmic analytics and provide tools to automate some of the more tedious forms of analysis, although other products might be added to do these things.

Reporting and Dashboard Tools

Because the reporting and dashboard tools are JavaScript based and the deployment is HTML5, the resulting visuals run on any platform in the browser. The usual chart widgets are supported (bar, funnel, pie column etc.), and mapping is supported via Google GeoChart. Other widgets include pivot tables, filters and SVG/HTML. Power users can get low level JavaScript access, and all relevant charts support drill-down to detailed data. Particularly relevant to the OEM is the theming of visuals, which can be defined once and applied everywhere.

The user interface architecture is simple, but very powerful. It allows developers to create advanced visualization widgets (using D3), is pure HTML5 based, multi-tenant ready, and embeds into existing applications (Apache, IIS, Servlet Engines, and WordPress). In a word it is an OEM’s dream and accounts for the large number of OEMs using the product.

icCube

The Data Engine

All common data sources are supported by icCube, including databases, Excel, NoSQL (MongoDB) and Google data. The engine itself is a MOLAP server which can be queried using an extended version of MDX called MDX++. This adds over 250 additional functions for complex querying and calculation (vector and matrix syntax, real-time creation of classifications and hierarchies and debugger). It runs on Windows, Linux and MacOS, and is easily configured in the cloud.

A semantic layer can be created using a thin client IDE, and many-to-many relations are accommodated within the architecture of the product. The server is a single Java process, supporting hot restart, remote maintenance and sophisticated authentication and authorization.

The R statistical language is integrated directly into the MDX query language, enhancing statistical analysis particularly and the ability to define R functions which can be reused.

Embedding

The relatively low cost (compared with many competitors), flexible architecture, extensibility (functions, widgets), scalability, multi-tenant architecture, white labeling, on-the-fly authentication and authorization, and high performance, have all proved to be attractive to OEMs. And as many businesses move away from ad-hoc analytics into production analytics (integrated with production environments), so the embedding capabilities of icCube will be very attractive.