Distributed Data Show 72
Patrick welcomes Shogo Hoshii to the show to learn why Yahoo Japan is running over 200 clusters of Apache Cassandra and why NoSQL databases are starting to gain momentum in Japan.
0:15 - Patrick welcomes Shogo to the episode and we learn that there are over 200 Cassandra clusters at Yahoo Japan
1:25 - 90% of internet users in Japan are on Yahoo
2:00 - The main use case is the advertisement platform - tracking which users watch ads, for how long, and this information is used to drive which are ads are shown most frequently
3:19 - Yahoo is unique in Japan for the scale of their use of Cassandra and other NoSQL databases. There is still a big preference for relational databases
4:29 - Hiring for NoSQL skills in developers and operators can be a challenge in Japan
5:10 - Shogo’s team is solving this problem with automation, especially metrics monitoring. In some cases they are sending actions to instances based on alarm conditions such as running compactions.
6:13 - Another struggle is finding space in data centers - they can’t increase the number of clusters indefinitely. Shogo’s team is looking into running Cassandra clusters on Kubernetes to utilize resources more efficiently.
Developer Relations at DataStax