Skip to main content
Fig. 15 | Journal of the Brazilian Computer Society

Fig. 15

From: Using weaker consistency models with monitoring and recovery for improving performance of key-value stores

Fig. 15

Comparing the completion time of sequential consistency (R1W3) vs. eventual consistency with rollback and adaptive consistency (R1W1 + adaptive) in Social Media Analysis application. On a power-law clustering graph, before 90% of the nodes are processed, R1W1 + adaptive progresses about 18% faster than R1W3. Overall, R1W1 + adaptive is 9.5% faster than R1W3. On a regular random graph, the benefit before 90% of the nodes are processed is 26% and the overall benefit is 20.8%. a Power-law clustering graph with 50000 nodes. b Random regular graph with 50000 nodes

Back to article page