- Published on
We introduced a memory database named scuba
- Authors
- Name
- Hao Chen
- @haoel
Content
Hi all in the previous share, we introduced a memory database named scuba.
Today we continue to focus on scuba, paper download link:
https://research.fb.com/wp-content/uploads/2016/11/fast-database-restarts-at-facebook.pdf
In this paper, the author introduces a new memory storage layout to use column-oriented storage for better compression and faster query execution strategies. they reference lots of database design include Dremel Powerdrill etc. (Describe in Section 2.1)
Highlight you need to know in this paper:
- General query performance of Scuba in real facebook business. (section 1)
- Why they use heap not directory use Memory Mapping. (before section2)
- Deploy Architecture. How data load to cluster and why so fast. (at start of section2)
- How they decide which storage layout to use. (before section4)
- we advise read the conclusion after reading section1(introduce) to find out which aspect of the database design you interested in.
- 5.1 how scuba copy data
- 5.2 how they rollout database version
- 5.3 describe overhead in recovery from the disk.
Conclude:
This paper describes how scuba in action. Besides the knowledge of how scuba designed, we can also learn about how facebook engineers study and make decisions.