Hakim Weatherspoon, Tal Moscovitz, and John Kubiatowicz. Introspective failure analysis: Avoiding correlated failures in peerto -peer systems. In Proc. International Workshop on Reliable Peer-to-Peer Distributed Systems, Oct 2002.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....to assign identifiers to 4 nodes, and claim that this way nearby nodes (in the id space) are likely to be diverse and have independent failures. Unfortunately, when applied to many objects replicated on such assumptions, the independence is unlikely to hold for a substantial number of objects. [19] suggests a technique to find independent sets of nodes. Such techniques would be easily exploited in SelectCast and Astrolabe to select failure independent routers and contacts. 7 Conclusion A P2P system should be aware of heterogeneity in the network. The reasons may be functional, for ....

H. Weatherspoon, T. Moscovitz, and J. Kubiatowicz. Introspective failure analysis: Avoiding correlated failures in peer-to-peer systems. In SRDS02 [16].

....using a model based on disk mean time to failure. In a separate paper, the authors also address the issue of independence between host failures. They mention the need to quantify dependence between failures at a coarse grained level, i.e. failures due to network disconnectivity, OS version, etc. [17] However, these efforts do not capture the complexity of the peer topeer environment particularly the user controlled transient outages that dominate host availability in most real systems. The dependence of host availabilities on time and whether there is any interdependence between host ....

H. Weatherspoon, T. Moscovitz, and J. Kubiatowicz. Introspective failure analysis: Avoiding correlated failures in peerto -peer systems. In Proceedings of the International Workshop on Reliable Peer-to-peer Distributed Systems, October 2002. 6

No context found.

H. Weatherspoon, T. Moscovitz, and J. Kubiatowicz. Introspective failure analysis: Avoiding correlated failures in peerto -peer systems. In Proc. of Intl. Workshop on Reliable Peerto -Peer Distributed Systems, October 2002.

....is called the dissemination tree for that object. Every time the primary While using Tapestry in this manner yields some degree of failure independence between fragments encoding the same block, it is preferable to achieve this independence more explicitly. We have a proposal for doing so [37], but it is not yet implemented. App App T archive (Inner Ring) Primary Replica Replica Secondary Replica Secondary Replica Secondary T hop T req T agree T disseminate Time Archive Figure 2: The path of an OceanStore update. An update proceeds from the client to the primary replica ....

....the first place To solve this problem, we rely on an entity known as the responsible party, so named for its responsibility to choose the hosts that make up inner rings. This entity is a server which publishes sets of failure independent nodes discovered through offline measurement and analysis [37]. Currently, we access this server through Tapestry, but simply publishing such sets on a secure web site would also suffice. An inner ring is created by selecting one node from each of 3f 1 independent sets. Superficially, the responsible party seems to introduce a single point of failure into ....

H. Weatherspoon, T. Moscovitz, and J. Kubiatowicz. Introspective failure analysis: Avoiding correlated failures in peer-to-peer systems. In Proc. of International Workshop on Reliable Peer-to-Peer Distributed Systems, October 2002.

No context found.

Hakim Weatherspoon, Tal Moscovitz, and John Kubiatowicz. Introspective failure analysis: Avoiding correlated failures in peerto -peer systems. In Proc. International Workshop on Reliable Peer-to-Peer Distributed Systems, Oct 2002.

No context found.

H. Weatherspoon, T. Moscovitz, and J. Kubiatowicz. Introspective failure analysis: Avoiding correlated failures in peer-to-peer systems. In Proc. of the International Workshop on Reliable Peer-to-Peer Distributed Systems, October 2002.

No context found.

H. Weatherspoon, T. Moscovitz, and J. Kubiatowicz. Introspective Failure Analysis: Avoiding Correlated Failures in Peer-to-Peer Systems. In Intl. Workshop on Reliable Peer-to-Peer Distributed Systems, Osaka, Japan, October 2002.

No context found.

H. Weatherspoon, T. Moscovitz, and J. Kubiatowicz. Introspective Failure Analysis: Avoiding Correlated Failures in Peer-to-peer Systems. In Intl. Workshop on Reliable Peer-to-Peer Distributed Systems, 2002.

No context found.

H. Weatherspoon, T. Moscovitz, and J. Kubiatowicz. Introspective failure analysis: Avoiding correlated failures in peer-to-peer systems. In Proc. of International Workshop on Reliable Peer-to-Peer Distributed Systems, Oct. 2002.

No context found.

Hakim Weatherspoon, Tal Moscovitz, and John Kubiatowicz. Introspective Failure Analysis: Avoiding Correlated Failures in Peer-to-Peer Systems. In Proceedings of International Workshop on Reliable Peer-to-Peer Distributed Systems, October 2002.

No context found.

Hakim Weatherspoon, Tal Moscovitz, and John Kubiatowicz. Introspective Failure Analysis: Avoiding Correlated Failures in Peer-to-Peer Systems. In Proceedings of International Workshop on Reliable Peer-to-Peer Distributed Systems, October 2002.

No context found.

Hakim Weatherspoon, Tal Moscovitz, and John Kubiatowicz. Introspective Failure Analysis: Avoiding Correlated Failures in Peer-to-Peer Systems. In Proceedings of International Workshop on Reliable Peer-to-Peer Distributed Systems, October 2002.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC