Showing papers by "Philip A. Bernstein published in 2019"
Proceedings Article•
01 Jan 2019TL;DR: It is argued that the virtual actor model is an ideal platform on which to build a serverless event-stream processing service with pay-for-use and SLA’s.
Abstract: A serverless event-stream processing service with pay-for-use and SLA’s must be scalable, highly available, and low cost. In this one-page abstract, we argue that the virtual actor model is an ideal platform on which to build such a stream processing system.
6 citations
Journal Article•
TL;DR: It is argued that distributed transactions on middle-tier servers need to return to the mainstream after a 15-year decline, especially for applications targeted for cloud computing.
Abstract: Over the years, platforms and application requirements change. As they do, technologies come, go, and return again as the preferred solution to certain system problems. In each of its incarnations, the technology’s details change but the principles remain the same. One such technology is distributed transactions on middle-tier servers. Here, we argue that after a 15-year decline, they need to return to the mainstream
1 citations
17 May 2019
TL;DR: The program committee process for SIGMOD 2018 is reported, including statistics, trends, and changes from previous years, including changes to reduce the number of parallel sessions and return to clustering industry presentations into separate industry sessions rather than grouping them with research papers on the same topic.
Abstract: This paper reports on the program committee process for SIGMOD 2018, including statistics, trends, and changes from previous years. Some highlights are: Submissions to SIGMOD 2018 were down 6% from 2017. The acceptance rate of research papers was 20%, which is in line with recent years. Reviewers showed a strong bias to reject borderline papers rather than giving authors the opportunity to revise. By being biased in the second round in favor of offering authors the opportunity to submit a revision, we increased the acceptance rate significantly from the first to second round. We strongly recommend that future PC chairs adopt this bias. To help ensure high quality reviews, we gave PC members a light reviewing load and ensured that 95% of review assignments went to PC members who bid Eager or Willing to review the paper. Nevertheless, approximately 20-25% of reviews are unacceptably shallow. We need to do better. The main changes in 2018 were (i) to reduce the number of parallel sessions, (ii) include tutorials during the main conference (Tuesday-Thursday), and (iii) return to clustering industry presentations into separate industry sessions rather than grouping them with research papers on the same topic. To enable (i) and (ii), we shortened the standard presentation time to 20 minutes and offered only 10 minutes to 40% of the papers. Anecdotal evidence is that attendees were happy with these changes. The paper closes with comments about my previous experiences as PC chair for SIGMOD 1979 and VLDB 2002 and with the evolution of PC processes.