Group Newsletter-Winter 2018

Over the last decade, there has been a tremendous growth in data-intensive applications and services in the cloud. Data is created on a variety of edge sources, e.g., devices, browsers, and servers, and processed by applications in the cloud to gain insights or take decisions. Applications and services either work on collected data or monitor and process data in real time. These applications are typically update-intensive and involve a large amount of state beyond what can fit in main memory. However, they display significant temporal locality in their access pattern. We developed FASTER, a persistent key-value store for state management. FASTER combines a highly cache-optimized concurrent hash index with a hybrid log: a concurrent log-structured record store that spans main memory and tiered storage while supporting fast in-place updates in memory. The hybrid log offers a self-tuning data organization capability to support a potentially drifting hot set, without requiring any finegrained statistics or meta-data. FASTER extends the standard key-value store interface to handle readmodify-writes, blind and CRDT-based updates by leveraging dynamic code generation to provide native support for advanced user-defined update types. Experiments show that FASTER achieves orders-ofmagnitude better throughput – up to 150M operations per second on a single machine – than alternative systems deployed widely today, and reaches baremetal performance when the workload fits in memory. This was part of Guna's Summer 2017 internship with Microsoft Research Database Group where he was mentored by Badrish Chandromouli. Project website: https://www.microsoft.com/enus/research/project/faster/ Title: Automatically Leveraging MapReduce Frameworks for Data-Intensive Applications. SIGMOD 2018 Authors: Maaz Bin Safeer Ahmad, Alvin Cheung Abstract: MapReduce is a popular programming paradigm for running MapReduce is a popular programming paradigm for running large-scale data-intensive computation. Recently, many frameworks that implement that paradigm have been developed. To leverage such frameworks, however, developers need to familiarize with each framework’s API and rewrite their code. We present Casper, a new tool that automatically translates sequential Java programs to the MapReduce paradigm. Rather than building a compiler by tediously designing pattern-matching rules to identify code fragments to translate from the input, Casper translates the input program in two steps: first, Casper uses program synthesis to identify input code fragments and search for a functional specification of each fragment. The specification is expressed using a highlevel intermediate language resembling the MapReduce paradigm. Next, each found specification is verified to be semantically equivalent to the original using a theorem prover. Casper then generates executable code from the specification, using either the Hadoop, Spark, or Flink API.We have evaluated Casper by automatically converting real-world sequential Java benchmarks to MapReduce. The resulting bench-marks perform up to 32.2x faster compared to the original, and are all translated without designing any patternmatching rules Project Website: http://casper.uwplse.org/ Title: Bias in OLAP Queries: Detection, Explanation, and Removal (Or Think Twice About Your AVG-Query). SIGMOD 2018 Authors: Babak Salimi , Johannes Gehrke (Microsoft) and Dan Suciu (UW).