We study the problem of k-anonymization of mail messages in the realistic scenario of auditing mail traffic in a major commercial Web mail service. Mail auditing is necessary in various Web mail debugging and quality assurance activities, such as anti-spam or the qualitative evaluation of novel mail features. It is conducted by trained professionals, often referred to as "auditors", who are shown messages that could expose personally identifiable information. We address here the challenge of k-anonymizing such messages, focusing on machine generated mail messages that represent more than 90% of today's mail traffic. We introduce a novel message signature Mail-Hash, specifically tailored to identifying structurally-similar messages, which allows us to put such messages in a same equivalence class. We then define a process that generates, for each class, masked mail samples that can be shown to auditors, while guaranteeing the k-anonymity of users. The productivity of auditors is measured by the amount of non-hidden mail content they can see every day, while considering normal working conditions, which set a limit to the number of mail samples they can review. In addition, we consider k-anonymity over time since, by definition of k-anonymity, every new release places additional constraints on the assignment of samples. We describe in details the results we obtained over actual Yahoo mail traffic, and thus demonstrate that our methods are feasible at Web mail scale. Given the constantly growing concern of users over their email being scanned by others, we argue that it is critical to devise such algorithms that guarantee k-anonymity, and implement associated processes in order to restore the trust of mail users.

Yahoo mail servers have been receiving an enormous number of messages each day for the past 17 years. The vast majority of today's messages are machine-generated (about 90% of the messages), based on a boilerplate with a small number of specific per-recipient changes. We show that the popular Zlib compression to gzip format fails to fully utilize the high similarity between these machine-generated messages. In this paper we analyze the data redundancy in Yahoo mail, and present methods to reduce its space requirements while using the standard Zlib library. Our results show we can further reduce the compressed data size by a factor of almost 2.5, compared to traditional gzip compression.

This paper proposes a general optimization framework to allocate computing resources to the compression of massive and heterogeneous data sets incident upon a communication or storage system. The framework is formulated using abstract parameters, and builds on rigorous tools from optimization theory. The outcome is a set of algorithms that together can reach optimal compression allocation in a realistic scenario involving a multitude of content types and compression tools. This claim is demonstrated by running the optimization algorithms on publicly available data sets, and showing up to 25% size reduction, with equal compute-time budget using standard compression tools.

In this paper, we present PACK (Predictive ACKs), a novel end-to-end traffic redundancy elimination (TRE) system, designed for cloud computing customers. Cloud-based TRE needs to apply a judicious use of cloud resources so that the bandwidth cost reduction combined with the additional cost of TRE computation and storage would be optimized. PACK's main advantage is its capability of offloading the cloud-server TRE effort to end-clients, thus minimizing the processing costs induced by the TRE algorithm. Unlike previous solutions, PACK does not require the server to continuously maintain clients' status. This makes PACK very suitable for pervasive computation environments that combine client mobility and server migration to maintain cloud elasticity. PACK is based on a novel TRE technique, which allows the client to use newly received chunks to identify previously received chunk chains, which in turn can be used as reliable predictors to future transmitted chunks. We present a fully functional PACK implementation, transparent to all TCP-based applications and network devices. Finally, we analyze PACK benefits for cloud users, using traffic traces from various sources.

In this paper we present PACK (Predictive ACKs), a novel end-to-end Traffic Redundancy Elimination (TRE) system, designed for cloud computing customers.

Peer-to-peer (P2P) networks are in the spotlight due to the wide-spreading file-sharing applications. Many file-distributing algorithms have been suggested and implemented. The various solutions need to cope with a heterogeneous and unstable environment, where peers can arrive and depart at a high rate (churn). Sometimes cooperation cannot be assumed. These issues make the structured attitude less practical. Even some of the algorithms that are considered as unstructured try to maintain long-term parent-child relationships. Existing unstructured (mesh-only) algorithms for file-distribution work well on the Internet on the average. But some of the participating peers may suffer from a slow start or high latency because of the randomness of the peer and piece selection for upload and download. In this paper we propose a fair unstructured system for file-distribution from a single source, with no central authority. The proposed protocol is fair both with respect to load balancing and with respect to the latency in each peer. It is based on a novel weights-algorithm that helps peers to determine what piece to ask from which peer, in a manner that increases their chance to get served. In this way it also lowers the overhead. The proposed algorithm welcomes newcomers while being resilient to churn, being resilient to free-riders, and adaptive to heterogeneous bandwidth.