Computing Platforms Seminar Series (COMPASS)

The Computing Platforms Seminar Series (COMPASS) is focused on talks by industry and academia around the general topic of computing platforms.

COMPASS is held on most Thursdays during the semester 10:00-11:00 (with some exceptions) in CAB E 72.


Upcoming Talks:

Thursday, 21. November 2019, 10:00-11:00 in CAB E 72

Speaker: Tamás Hauer, Technical Program Manager, SRE, Google Zurich

Title: Meaningful Availability

Abstract:

High availability is a critical requirement for cloud applications; having a metric that meaningfully captures it is useful for users and system developers. Commonly used benchmarks are either too complex to be actionable or fail to capture the true perception of users, which leads to miscommunication, and suboptimal engineering decisions. We propose two improvements to availability measurements. A novel metric, "user-uptime" directly models user-perceived availability and avoids the bias often found in alternatives. A presentation paradigm "windowed availability" supports a holistic view by integrating timescales from per-minute to monthly granularity and allows to distinguish between many short periods of unavailability or fewer longer ones. We demonstrate the benefits of windowed user-uptime on synthetic models and on production data from Google's G Suite. Today, all G Suite products are instrumented with this novel metric, it is used both to support engineering decisions and to communicate system health to enterprise customers.

Bio:

Tamás Hauer holds a PhD in theoretical physics. After a brief career as a physicist at the Max-Planck-Institut and CERN, he worked as a research associate at the department of Applied Computer Science of the University of West of England, leading a group in the area of semantic web, grid services and health informatics. He was work package leader of the European FP5 Mammogrid and FP6 Health-e-Child projects. As the CTO of the Swiss startup Prodema Medical / McMRI, he led the development of appMRI, a brain MRI image analysis platform to launch and certification as a class IIa medical device. He joined Site Reliability Engineering of Google Zürich in 2016 as a Technical Program Manager, his current interest is data analysis of service level indicators and service level objectives.


Thursday, 28. November 2019, 10:00-11:00 in CAB E 72

Speaker: Djordje Zegarac and Martin Marciniszyn (Tensor Technologies)

Title: High Frequency Trading and FPGAs 

Abstract:

High-Frequency Trading (HFT) platforms were typically implemented in software on traditional CPUs with high performance network adapters. However, the industry-wide race to "Zero Latency" has led the trading world to explore alternative system architectures that would minimize the internal latency. Field-Programmable Gate Arrays (FPGAs) offer the superior performance with deterministic execution while providing custom implementation flexibility. Due to these valuable architectural properties FPGAs became the integral part of HFT industry in accelerating trading solutions and reducing wire to wire latencies. In this talk we are going to outline the general architecture of our system and describe the main design challenges.

Bio:

Djordje Zegarac received the B.Sc. degree in Electrical Engineering from the University of Calgary, Canada in 2011, the M.Sc. degree in Electronics and Microelectronics from Ecole Polytechnique Fédérale de Lausanne and IBM Research, Switzerland in 2014. He worked as an IC Digital Design Engineer at u-blox, and as FPGA SoC Development Engineer at Enclustra, Switzerland. Currently, he is employed by Tensor Technologies as an FPGA Engineer. His main research interest is in the area of ASIC/FPGA design and hardware acceleration.

Martin Marciniszyn received a Ph.D. in Computer Science from ETH Zurich in 2007. Afterwards he spent the largest part of his professional career as a quant researcher at IMC Trading, one of the largest global HFT companies. Currently, he is the CTO of Tensor Technologies.


  

Past COMPASS Talks:  

Date Speaker Affiliation Talk
17.10.2019 René Müller
NVIDIA Simplifying NVIDIA GPU Access: A Polyglot Binding for GPUs with GraalVM
10.10.2019 Norman May SAP Research Exploiting modern hardware in SAP HANA
26.09.2019 Ben Zhao University of Chicago Hidden Backdoors in Deep Learning Systems
19.09.2019 Martin Hentschel/Max Heimel Snowflake File Metadata Management at Snowflake
11.07.2019 Boris Grot University of Edinburgh Scale-Out ccNUMA: Embracing Skew in Distributed Key-Value Stores
17.05.2019 Tim Kraska MIT Towards Learned Algorithms, Data Structures, and Systems
25.04.2019 Peter Pietzuch Imperial College London Scaling Deep Learning on Multi-GPU Servers
28.03.2019 Theo Rekatsinas
University of Wisconsin A Machine Learning Perspective on Managing Noisy Data
21.03.2019 Marko Vukolic IBM Research Hyperledger Fabric: a Distributed Operating System for Permissioned Blockchains
28.02.2019 Alberto Lerner University of Fribourg
The Case for Network-Accelerated Query Processing
21.02.2019 Thomas Würthinger Oracle Labs Bringing the Code to the Data with GraalVM
31.01.2019 Irene Zhang Microsoft Research, Redmond Demikernel: An Operating System Architecture for Hardware-Accelerated Datacenter Servers
25.10.2018 Mihnea Andrei SAP HANA Snapshot isolation in HANA - the evolution towards production-grade HTAP
04.10.2018 Philippe Bonnet IT University, Copenhagen, Denmark Near-Data Processing with Open-Channel SSDs
25.09.2018 Nandita Vijaykumar   Carnegie Mellon University Expressive Memory: Rethinking the Hardware-Software Contract with Rich Cross-Layer Abstractions
20.09.2018 Patrick Stüdi IBM Research Data processing at the speed of 100 Gbps using Apache Crail (Incubating)
15.08.2018 Leonid Yavits
Technion Resistive CAM based architectures: Resistive Associative In-Storage Processor and Resistive Address Decoder
06.07.2018 Martin Burtscher Texas State University Automatic Hierarchical Parallelization of Linear Recurrences
15.06.2018 Nitin Agrawal Samsung Research Low-Latency Analytics on Colossal Data Streams with SummaryStore
24.05.2018 Cagri Balkesen Oracle Labs RAPID: In-Memory Analytical Query Processing Engine with Extreme Performance per Watt
16.05.2018 Carsten Binnig TU Darmstadt Towards Interactive Data Exploration
09.05.2018 Bastian Hossbach Oracle Labs Modern programming languages and code generation in the Oracle Database
26.04.2018 Spyros Blanas Ohio State University Scaling database systems to high-performance computers
19.04.2018 Jane Hung MIT The Challenges and Promises of Large-Scale Biological Imaging
12.04.2018 Christoph Hagleitner IBM Research Heterogeneous Computing Systems for Datacenter and HPC Applications
14.03.2018  Eric Sedlar
 Oracle Labs
Why Systems Research Needs Social Science Added to the Computer Science
01.03.2018 Saughata Ghose Carnegie Mellon University How Safe Is Your Storage? A Look at the Reliability and Vulnerability of Modern Solid-State Drives
22.02.2018  Ioannis Koltsidas IBM Research Zurich System software for commodity solid-state storage