PRISM: PRecision-Integrated Scalable Monitoring
msra(2006)
摘要
This paper describes PRISM, a scalable monitoring ser- vice that makes imprecision a first-class abstraction. Ex- posing imprecision is essential for both correctness in the face of network and node failures and for scalability to large systems. PRISM quantifies imprecision along a three- dimensional vector: arithmetic imprecision (AI) and tem- poral imprecision (TI) balance precision against monitor- ing overhead while network imprecision(NI) addresses the challenge of providing consistency guarantees despite fail- ures. Our implementation provides these metrics in a scal- able way via (1) self-tuning of AI budgets to shift impre- cision to where it is useful, (2) pipelining of TI delays to maximize batching of updates, and (3) dual-tree prefix ag- gregation which exploits regularities in our DHT's topolog y to drastically reduce the cost of the active probing needed to maintain NI. PRISM's careful management of impreci- sion qualitatively improves its capabilities. For example , by introducing a 10% AI, PRISM's PlanetLab monitoring ser- vice reduces network overheads by an order of magnitude compared to PlanetLab's CoMon service, and by using NI metrics to automatically select the best aggregation resul ts, PRISM reduces the observed worst-case inaccuracy of our measurements by nearly a factor of five.
更多查看译文
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络