Katie has spent her career working on cloud technology. As a Site Reliability Engineer at Google, she worked first with BigTable, a highly scalable distributed storage system which was around before people started using the term “NoSQL”. Then moving to Google Switzerland, she was tech-lead of SRE for Google’s storage of social data, working with systems based on Megastore and Spanner, both of which take very different approaches to maintaining consistency across a distributed system.
She is now a full-stack developer at Grok Learning, which provides online courses and IDE for teaching coding in high schools. There she gets to play with Amazon’s cloud offerings instead of Google’s.
YOW! West 2017 Perth
Consistent Storage or Scalable Storage – Why Not Both?
TALK – VIEW SLIDES
Relational databases are the product of decades of research and optimisation, but when your business grows so big that a database reaches breaking point, where can you turn? On the other side of the fence are a variety of noSQL databases which sell on scalability and cheaper hardware but come with their own set of issues around schemas, atomicity and consistency.
Since most developers are more familiar with relational databases, in this talk I’ll focus more on a couple of NoSQL options and how they work rather than how to use them. The fun parts are how eventual-consistency is maintained at scale and how they avoid single points of failure and bottlenecks.
In the last section, I’ll go into depth on Google’s Spanner database, which is only just recently available for public use. This follows on well from the NoSQL implementation since it was developed as a direct response to the problems of noSQL storage. Through some cool innovations, Spanner manages to achieve both global consistency and scalability.