The DEVIEW 2014 developer conference was held at the Lotte Hotel World in Seoul, South Korea, from September 30-31, with Gruter’s dev team actively involved in the event. Attracting upwards of 2000 attendees each year, DEVIEW is now one of the biggest IT developer conferences in the country, with this year’s installment no exception.
Joining experts from domestic tech powerhouses such as Naver, SK Telecom and Samsung Electronics, and global web-tech giants including Linkedin, Netflix, OpenStack Foundation and Github, the Gruter team made the most of the occasion, presenting, learning and making new connections at what was a major success for the South Korean tech community.
Hyoung-jun Kim, Gruter CTO and respected data platform architect, gave a technical overview of the Vectorized Processing Model currently being developed by the Apache Tajo community. Designed to overcome execution inefficiencies in widely-used query logics such as the “tuple-at-a-time” approach, Kim explained how vectorization could overcome many of the processing bottlenecks currently impeding massive data query processing.
Gruter Senior Developer, Jae-hwa Jung, looking at home on the podium
Jae-hwa Jung, Gruter Senior Developer, Apache Tajo committer and Hadoop textbook author, gave a timely talk on Hadoop security. After reviewing the four main Hadoop security layers—authentication, authorization, auditing and data encryption—Jung went on to detail the use of Kerberos authentication in the Hadoop security design flow. Given Kerberos often requires more resources than initially expected, Jung emphasized the need for careful design and testing before releasing it into production. His talk concluded with practical recommendations for using LDAP and Active Directory in conjunction with a Hadoop security gateway such as Sentry and Knox.
Gruter Senior Developer, Ho-wook Jeong, speaks to a packed audience.
Ho-wook Jeong, Gruter Senior Developer and Search Specialist, gave a talk on Elasticsearch performance optimization. Discussing the various factors affecting search engine performance and optimization—from the hardware layer through to the indexing and query layers—Jeong then drilled down into the finer points of running Elasticsearch on Hadoop. Jeong concluded his talk with a closer look at “SQL on Elasticsearch”, an approach enabling the querying of Elasticsearch with standard SQL syntax via a JDBC driver.
Meanwhile, the Gruter team also ran a booth showcasing Apache Tajo and Cloumon, and gave attendees an early look at two new Tajo-based products under development: Tajo Cloud, a SaaS-style solution, and Tajo Enterprise, an integrated, on-premises Tajo stack. Stayed tuned for more on these!
* Session deck: Elasticsearch Performance Optimization (in Korean)
Elastic Search Performance Optimization – DEVIEW 2014 from Gruter Corp