Les 1er et 2 avril derniers se tenait la 3ème édition du salon Big Data Paris, avec nombre de retours d’expérience et de conférences adressant les non-techniciens d’entre nous, et même une série d’interventions dédiée aux usages en marketing.
Que les « big data » soient une vraie nouveauté ou un bon concept marketing qui permette au sujet d’émerger différemment dans l’espace de discussion et l’intérêt qu’il suscite auprès des entreprises, une chose est sûre néanmoins: c’est un sujet désormais démocratisé, à la portée d’un grand nombre d’entreprises même petites, avec des applications très variées où l’imagination risque plus souvent d’être la ressource manquante que la donnée ou la technologie…
En voici pour preuve les différents retours d’expérience recueillis au cours de la première journée à laquelle j’ai participé :
1. Présentation Tata Consulting Services
2. Présentation Cap Gemini
3. Intervention Doug Cutting, Cloudera (co-fondateur d’Hadoop): ci-dessous
4. Intervention de Christophe Benavent (Université Paris Ouest)
5. Amadeus Leisure Group
6. Renault et Access France
7. Vega France et Coheris
8. SFR et Sinequa
——————————————————————-
3. The Future of Data
by Doug Cutting, Chief Software Architect, Cloudera, and co-founder of Hadoop
This is a transcription of Doug Cutting’s presentation with his own comments.
Some facts make good predictions.
Fact : Hardware gets cheaper.
> Associated prediction : we’ll be able to store and process even more data than we do today ; we need to think about it. What do we want to do with it ?
Fact : Data is valuable.
> If we want to improve, we need to measure what we’re doing first. We cannot afford not to do that to be successful.
Fact : Open source wins.
In particular in platform technologies, in the foundation of our systems. Also in databases. Also in mobile (Android).
Users of software, in particular in enterprises, learned to fear software that was not open source.
Open source wins on the long term, and is a good attractor for adoption.
Fact : Hadoop gets better.
Fact : Hadoop is general.
Fact : Hadoop dominates Big Data.
> Oracle, Microsoft… all have adopted Hadoop as the standard big data platform.
> Associated prediction : siloed systems will diminish as more people will move on a common platform and have specific applications for each of their line of businesses. Which makes it a lot easier to experiment and find new applications.
Fact : Even transactions possible on Hadoop.
Transactions are hard to implement in distributed systems, but this is also starting to happen (ref. to a Google paper on this subject published a while ago).
Conclusion : All this leads to the emergence of the Enterprise Data Hub = the set of functions that lets institutions bring all their data to one place and build their applications upon it and experiment with their data.
This is a revolution in data processing.
And Hadoop is a safe path to this future.