Introduction to R Service Bus

Having the right algorithm is a first big step to using advanced analytics to solve your problem and inform your decisions. The next one is to integrate the algorithm into your workflows and business processes.
The R Service Bus is a swiss army knife that allows you to plug R into your processes independently of the technology used by other software applications involved in the workflow. It has been in development for over two years and released under open source license to the R community in April 2011.
In this presentation I will provide an overview of the features of the R Service Bus and show how it can be used to integrate R into your workflow. The R Service Bus ensures reliability, flexibility, high availability and scalability of R-based analytics applications

Transcript

1. R Service Bus Integrating R into business processes Willem Ligtenberg - willem.ligtenberg@openanalytics.eu October 14, 2015
2. Willem Ligtenberg Company: Interests: Social Media: Open Analytics · parallel computing (ROpenCL) network theory databases performance improvement · · · · @wligtenberg +wligtenberg · · 2/24
3. Open Analytics 3/24
4. Enough about me
5. Data scientists have a cool job play with data from all kinds of exciting studies and experiments new technologies ask for new methodology development using R package development process in place (and easy) · · · lingua franca of statistics goldmine of algorithms tools for reproducible research (Sweave/knitr/rmarkdown) feel the pulse of statistical innovation - - - - · 5/24
6. The boring stuff your method works and becomes popular! (of course it did!) standardized reporting or QC of data sets (repetitive work) ad hoc integration with existing infrastructure, IT struggles doing less cool data science :( · · · · 6/24
7. R Service Bus
8. Demo 8/24
9. Take a step back
10. High-level diagram 10/24
11. Application layer business analyst (mail) scientist (web client) partner (dumping of files on sftp server e.g.) software application (web services e.g. REST) · · · · 11/24
12. Statistical layer always serve the same results using a common statistical engine under full control of the data scientist using R scripts / packages · · · 12/24
13. Integration layer R Service Bus copes with the tower of Babel jungle of proprietary software plethora of protocols and payloads · · spaghetti of files horrible spreadsheet formats databases SOAP web services (s)ftp servers RESTful web services e-mail protocols - - - - - - - 13/24
14. Architecture diagram 14/24
15. Example deployment dose response over SOAP web services, web client and RESTful web services; analysis results stored in a company wide data warehouse 20 other statistical applications on a common default RPooli pool separate RPooli pool for NGS data analysis application (big iron) adding a new application is a matter of hours (and not of months) · · · · 15/24
16. Security for Web (and Other) Clients Central Authentication Service (CAS) and OpenID client-side X.509 certificates, LDAP and ActiveDirectory HTTP BASIC, HTTP Digest custom file- and database backed authentication backends · · · · 16/24
17. Administration and monitoring administration API advanced monitoring of RSB and RServi using JMX (RMI and HTTP) mobile clients (work in progress) · configuration restart pool management R package installation - - - - · · 17/24
18. Next steps
19. Where to get it https://hub.docker.com/r/openanalytics/rsb/ 19/24
20. Can I use it too? Of course! free as in beer and speech RSB developed in a GxP context (to automate toxicological analyses) in production use since 2010 scales very well (8500 requests per second 24/7) · http://www.r-project.org (GPLv3) http://www.openanalytics.eu/r-service-bus (AGPLv3) http://www.openanalytics.eu/architect (EPL) - - - · · · 20/24
21. Dockerfiles View in Architect 21/24
22. Acknowledgements R Core Development Team (in particular Martin Maechler, Duncan Murdoch, Prof. Ripley, Simon Urbanek) Philippe Lamote · · 22/24
23. willem.ligtenberg@openanalytics.eu Questions?
24. We are hiring! http://www.openanalytics.eu/jobs Statistician for a MSc / PhD in Statistics, Applied Mathematics or Computer Science Data Scientist for a MSc / PhD in Computer Science, Engineering or Statistics Java Developer for a MSc in Computer Science · · · 24/24