Shiding Lin

RepStore: a self-managing and self-tuning storage backend with smart bricks

With the continuously improving price-performance ratio, building large, smart-brick based distributed storage system becomes increasingly attractive. The challenges, however, include not only reliability, adequate cost-performance ratio, online upgrades and so on, but also the systems ability to achieve these goals in as self-managing and self-adaptive a manner as possible. In this paper, we describe RepStore, a system that fulfills these goals. RepStore unites the self-organizing capability of P2P DHT and the completely autonomous, per-brick tuning mechanism to derive a scalable and cost-effective architecture. RepStore employs replication for active write-intensive data and erasure-coding for the rest, strives to achieve the best cost-performance balance automatically and transparent to application, and does so in a completely distributed manner. Our preliminary evaluations reveal that the system performs much as expected, achieving performance and reliability closer to a 3-way fully replicated system with only 60% of the cost.

Sigma: a fault-tolerant mutual exclusion algorithm in dynamic distributed systems subject to process crashes and memory losses

This paper introduces the Sigma algorithm that solves fault-tolerant mutual exclusion problem in dynamic systems where the set of processes may be large and change dynamically, processes may crash, and the recovery or replacement of crashed processes may lose all state information (memory losses). Sigma algorithm includes new messaging mechanisms to tolerate process crashes and memory losses. It does not require any extra cost for process recovery. The paper also shows that the threshold used by the Sigma algorithm is necessary for systems with process crashes and memory losses.

P2P resource pool and its application to optimize wide-area application level multicasting

The concept of resource pool has a very long history. Propelled by the need to share CPU cycles of supercomputers for highthroughput computing jobs from the scientific community, the vision is most recently explored by the advocates of Grid. On the other hand, the advent of P2P researches has demonstrated the feasibility of integrating potentially unlimited amount of less powerful machines around the world. Organizing a P2P resource pool thus becomes an interesting research topic. This paper attempts to address two problems. The first is how to organize a P2P resource pool, and our answer is to combine the self-organizing strength of P2P DHT with an in-system, selfscaling monitoring infrastructure that is layered on top of DHT. The second question is the utility of the P2P resource pool for interesting applications. And we choose to showcase its power by optimizing wide-area application level multicasting (ALM), a problem far more challenging and interesting than conventional tasks such as massively parallel computation. We show that utilizing spare resources in the pool results in significant savings for single ALM session. Furthermore, we adopt a purely market-driven approach to optimize multiple concurrent sessions. As expected, sessions of higher priority are given higher share of resources.