|
| Michal Szymaniak |
|
Contact:
The best way to reach me is by email: my_first_name @ my_last_name . net
Resume:
..can be found here. For my
current work, check out the Genesis primer.
I have been building globally distributed systems since 2007.
Google Spanner is known best across the industry, while Dropbox Magic
Pocket and Facebook fleet-wide overhauls were major milestones in their
respective environments.
I am currently working on Genesis, which is a low-level data store for
infrastructure and database projects. It combines replication, sharding,
and change notification in a single platform designed for high availability,
"infinite" scalability, and easy long-term maintenance.
Some examples of how Genesis can be used include:
- zero-dependency metadata store for low-level infrastructure,
- massively sharded NoSQL for online services,
- backend for a higher-level data store that needs replication and/or sharding,
- highly-available storage platform across multiple physical locations,
- replacement for FoundationDB, to adopt a multi-master architecture for higher scalability,
RocksDB/gRPC/C++ for easier maintenance, and a symmetric production setup for smooth operations.
You can find out more about the technology behind Genesis here.
Before 2007, it was all about research:
PhD Thesis:
Publications:
- Spanner: Google's Globally-Distributed Database,
ACM Transactions on Computer Systems, Vol. 31, No. 8, August 2013.
- Practical Large-Scale Latency Estimation,
Elsevier's Computer Networks, Vol. 52, No. 7, April 2008.
- Enabling Service Adaptability with Versatile Anycast,
Concurrency and Computation: Practice and Experience, Vol. 19, No. 13, September 2007.
- Latency-Driven Replica Placement (Extended Version),
Invited Article, IPSJ Journal, Vol. 47, No. 8, August 2006.
- Replication for Web Hosting Systems,
ACM Computing Surveys, Vol. 36, No. 3, September 2004.
- Versatile Anycasting with Mobile IPv6,
Proceedings of the 2006 ICST International Workshop on Advanced Architectures and Algorithms
for Internet Delivery and Applications (AAA-IDEA2006), Pisa, Italy, October 2006 [slides]
- Latency-Driven Replica Placement,
Proceedings of the 2005 IEEE International Symposium on Applications and the Internet (SAINT2005),
Trento, Italy, February 2005 (Best Paper Award) [slides]
- Scalable Cooperative Latency Estimation,
Proceedings of the 10th IEEE International Conference on Parallel and Distributed Systems (ICPADS2004),
Newport Beach, CA, USA, July 2004 [slides]
- NetAirt: A Flexible Redirection System for Apache,
Proceedings of the IADIS International Conference on WWW/Internet (ICWI2003),
Algarve, Portugal, November 2003 [slides]
Talks:
- Ad Hoc Distributed Servers,
Vrije Universiteit Amsterdam, February 2006.
- Stable Node Positioning,
Vrije Universiteit Amsterdam, November 2005.
- Deciding on Satellite Placement,
Google Inc., Mountain View, CA, USA, October 2005.
- Latency-Driven Replica Placement,
The 11th Annual ASCI Conference, Heijen, The Netherlands, June 2005.
- Wide-Area Handoffs in Ad Hoc Distributed Servers,
Vrije Universiteit Amsterdam, April 2005.
- SCoLE: Scalable Cooperative Latency Estimation,
The 10th Annual ASCI Conference, Port Zelande, The Netherlands, June 2004.
- Streaming Content Delivery in Peer-to-Peer Networks,
Technical University Delft, June 2004.
- Principles of Distributed Redirection,
Vrije Universiteit Amsterdam, February 2004.
- SCoLE: Scalable Cooperative Latency Estimation (RIPE TTM version),
RIPE-47 Meeting, Amsterdam, January 2004.
- NetAirt: a DNS-based Redirection System for Apache,
The 9th Annual ASCI Conference, Heijen, The Netherlands, June 2003.
- A Framework for Web Replica Hosting Systems,
Vrije Universiteit Amsterdam, February 2003.
- Communication Between Uncoupled Processes,
Vrije Universiteit Amsterdam, February 2002.
NetAirt:
- ..was my master's project;
it is the redirecting subsystem of Globule.
- The name NetAirt comes from the old Scottish word airt, which means direction, or to direct, to guide.
|