20. Google Cloud Dataproc & The network behind the elephant

Agenda

• 17.45: drink, socialize

• 18.00: first talk: Google Cloud Dataproc

Speaker:Guillaume Leygues is Sales Engineering Lead EMEA for Google Cloud Platform Infrastructure services.

http://photos1.meetupstatic.com/photos/event/2/e/2/b/600_444671819.jpeg

Guillaume Leygues is Sales Engineering Lead EMEA for Google Cloud Platform Infrastructure services, based in Stockholm.He is french, has been living in Stockholm for almost 20 years, likes heavy metal and will go to Judas Priest's concert at Ericsson Globe on Saturday night.

He will be joined by Angus Davis, Software Engineer from Google working on the Dataproc team:

Angus Davis is a Software Engineer at Google working on all aspects of Cloud Dataproc. Before Dataproc, Angus developed the Cloud Bigtable Java client, maintained the GCS and BigQuery Connectors for Hadoop, and improved and maintained bdutil. Before joining Google, Angus developed big data applications with Hadoop, HBase. and ElasticSearch.

Also joining will be Giuseppa Reina, "Technical Solutions Engineer" working within Cloud Technical Support, specialized on the Hadoop ecosystem and Dataproc.

Abstract: Open source data processing frameworks - with great power comes great complexity. While the Spark and Hadoop ecosystems make it possible to process vast amounts of data, managing clusters and workloads can be complicated, time-consuming, and costly. As a solution to this problem, Google has combined its experience in infrastructure and software to create Cloud Dataproc. Powered by Google Cloud Platform, Cloud Dataproc makes Spark and Hadoop easy, fast, and cost-effective. This means anyone interested in taking advantage of Spark and Hadoop, from advanced users to novices, for a range of use cases, from MapReduce to machine learning, can have a cluster with hundreds of nodes at their fingertips in less than 90 seconds.

• 18.45: eat, drink, socialize (more)

• 19.00: second talk: The network behind the elephant

Speaker: Edward Zambrano Network Engineer at Spotify

http://photos3.meetupstatic.com/photos/event/3/4/8/0/600_444493440.jpeg

Edward is a part of the XNET team inside Spotify, a bunch of network engineers with different backgrounds in ISPs, Enterprise networks, NOC and coding. This team handles the network of Spotify Datacenters around the world. He has been working in Spotify for a year and half, recently worked as the road manager for deploying the new network architecture inside the San Jose Data Center. In his free time, Edward enjoys playing the piano, boxing and climbing.

Abstract: Networks are fun when they work, and a huge pain when they don't. Spotify's datacenter network is all about allowing developers the continuous deployment of servers without worrying about things like bandwidth, routing and rack awareness. For the case of Hadoop, having a cluster of about 1600 servers requires having a flexible, highly scalable and automated network. In this talk we will be talking about the story and reasons behind our current network design, and how we did use coding to solve problems like IPs assignations, firewall automation and deployment of new racks using the CLI.

• 19.45: drink, socialize (even more)

Follow SHUG on twitter (https://twitter.com/shug_meetup)!