MongoDB and Neo4J from an Operations Point of view
von Thomas Fricke (Endocode AG)
Introducing NoSQL Databases normally is driven by development. The talk shows what this means for operations. If software is not mature enough for the data center, and you cannot stop it, you have to devops it, until it is a first class database. This means a lot of details: implementation of a backup-restore cycle, monitoring and a concept how to run a database like MongoDB on the same level of reliability as any classical SQL DB.
It is shown, how MongoDB replaces a large proprietary SQL database and what has to be done to implement the missing features from the operations point of view. MongoDB has a build in replication working out of the box. However, replication does not replace backups, which can be added with a few scripts. It is shown which parts are missing, and howto add them with a handful of scripts.
Another examples shows how Neo4J can be used to split the big picture of any entitity in the datacenter to a set of relations effectively being handled by cipher queries. The general idea analyses configuration files and maps each server group into a graph model represented by graphviz dot file. The dot files can be imported into the Neo4J database for further analysis by the using the powerful query language.
Ãœber den Autor Thomas Fricke:
Thomas Fricke graduated with a Diplom 1989, made his PhD in 1994 in theoretical Physics at RWTH Aachen and worked as post-doc at Humboldt Universität Berlin. He is involved in Linux since kernel 0.95 and is aware of system administration. Since 1992 he planned and implemented large compute and workstation-clusters. Since 1996 he works as a C/C++/Java developer for several small and big companies.
Since 2005 works as a freelancer with main topic development of Java applications in a Linux environment (application servers, public key infrastructure, web services, messaging systems), since 2006 developed software for Xen and VMWare virtualisation, and is since 2009 developing Android and JS for eeVentmap.
In 2013 he joined Endocode as Head of Data Center Automation and Principal DevOps Wizard.