Amazon Aurora: Design Considerations for High Troughput Cloud-Native Relational Databases

Tomorrow night, 18:00 Moscow We're going to have a Code of Architecture release about an interesting database. Amazon Aurora. It is interesting both in terms of marketing and technical: This technical product within AWS was highlighted by Jeff Bezos in his annual letters to shareholders. 2014 and 2015 year This technical product is the implementation of the concept "the log is a database".

The document itself consists of the following parts: The I/O bottleneck faced by traditional database systems changes in the cloud environment. Durability at cloud scale - an approach to considering beech D from the familiar acronym ACID, how to think about quorums, correlated errors and reduce MTTR (mean time to recovery) by separating data into segments The log is the database - how to turn the redo log into the main element of the database and materialize the remaining entities asychronously The log marches forward - the main part about the logic of the log sequence number in normal condition and in case of failures Putting it all together - a section with a beautiful picture, which shows the overall architecture, gathering together thoughts from the previous sections Performance results - the results of benchmarks, which show why the whole cheese buff was started Lessons learned - a story about why Aurora is good for SaaS applications, about the evolution of the scheme inside databases and how to patch the Aurora itself without downtime Related Work – This section describes related work

In general, this white paper is a great example of an interesting design document. It will be interesting to discuss it with our guests:

• Igor Maslov - Head of Basic Technologies and Data Processing Department at Tinkoff Vitaly Kondratov - architect in the management of basic technologies

#Databases #CoA #Architecture #SoftwareArchitecture #SystemDesign #DistributedSystems #Cloud