{[ promptMessage ]}

Bookmark it

{[ promptMessage ]}


2008 test and set write requiredversion can be

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: Figure 1 from (Johnson, 2009) 27 28 SQL Execution Plan MapReduce Dataflow Diagram •  Tablets are the input. •  Map function have different execution path per type. •  Storage format. Reduce- 2 System sort Reduce- 1 Remove duplicates 4 Filtering ê༎ Parallel maps Figure 3 from (Johnson, 2009) 29 Figure 4 from (Johnson, 2009) 30 5 10/29/13 Query Processing Algorithm Ê༌  Can be used to process a wide range of SQL queries (select- group- aggregate). Ê༌  Scalable for large DBs. Ê༌  Possible Problem: flow of large fraction of the DB à༎ high network traffic + slow system sorts. Ê༌  The author’s conclusion: SQL evaluation can be done in the PNUTS clouds, but its performance is still an open question. (Cooper et al., 2008) 31 PNUTS Per- Record Consistency vs. Eventual Consistency Ê༌  Designed for Yahoo’s web applications. Ê༌  In eventual consistency, updates will eventually be applied Ê༌  Geographically distributed database system. Ê༌  Requirements: high availability, scalability, and response time. Ê༌  Per- record consistency (relaxed consistency). to all replicas but they can be out of order. Ê༌  Too weak for some applications. Ê༌  Example: Photo sharing application (photos with access privileges). (1) Change access (2) Upload Photos R1 (1) Upload Photos … (2) Change access Rn Ê༌  In Per- record consistency, updates will eventually be appli...
View Full Document

{[ snackBarMessage ]}

Ask a homework question - tutors are online