ChessBrain: a Linux-Based Distributed Computing Experiment
The SuperNode and PeerNode are multithreaded applications written in C++ and compiled using GCC under Red Hat Linux 7.1, 7.2 and 8.0. The primary SuperNode server runs under Slackware 8.0 at bteg network's colocation site in Northern California (Figure 2).
Because the applications are heavily multithreaded, I spent a fair amount of time resolving threading issues. I used GDB, DDD and custom logs to tackle debugging. Early in the development process, Perl scripts proved especially effective in helping test new functionality and stress test the software. I have 12 machines at home; these, plus an army of Perl scripts pounding on a local server, proved to be formidable testing tools.
Early in the project I realized the SuperNode server would need to communicate with other servers. During that time XML offered a viable approach, and later XMLRPC (www.xmlrpc.org) brought additional advantages. The Simple Object Access Protocol (SOAP) continued evolving to meet the needs of servers that speak to other servers. Encouraged by promises of improved interoperability, I adopted SOAP as the preferred method of communication for the SuperNode server and PeerNode client.
From the outside, the SuperNode acts like a Web server with SOAP-based interfaces. Although the SuperNode server handles HTTP GET and POST, the POST message is used most often. The SuperNode parses HTTPs and XML-based SOAP requests, processes those requests and returns HTTP packages with embedded SOAP payloads.
The SuperNode and PeerNode parse SOAP requests and route commands to an internal command dispatcher, which ensures that the correct command handlers process the requests. In the SuperNode, the most common requests come from PeerNode clients; a PeerNode must connect to request a job unit. A job unit is an XML block containing a game position and instructions on how to analyze the position. A PeerNode contains a complete chess engine component, compiled and linked as a static library. When the PeerNode receives a job unit, it processes the SOAP response, extracts the job-specific information and passes instructions to its internal chess component for analysis.
The SuperNode server then passes the current game position to the external BeoServer process. Interprocess communication between the SuperNode and BeoServer is accomplished using two pipes. In the near feature, we expect to move BeoServer to its own box and shift to UDP over 1000Base-T Ethernet.
Secure and tamper-free communication is a necessity for ChessBrain. An invalid result created by a malicious user could render the play ineffective and ultimately embarrassing. Sensitive communication is protected using the Advanced Encryption Standard, AES Rijndael (pronounced Rhine-doll). AES is a variable block symmetric encryption algorithm developed by Belgian cryptographers Joan Daemen and Vincent Rijmen as a replacement for the aging DES and Triple DES standards.
Before exploring Rijndael, the Blowfish symmetric cipher was used until the PeerNode client was ported to Mac OS X and problems surfaced involving endian issues with the implementation of Blowfish being used. AES is an endian-neutral algorithm and proved ideal for our situation.
The original design of the PeerNode involved having the client and its chess engine as two separate processes. The PeerNode started the chess engine process and redirected the standard I/O to establish a loose binding. Initially, we avoided directly linking chess code with the PeerNode client so the chess code could be replaced quickly and easily in future iterations of the software. We later moved to a static linking approach to deal with potential security issues. The problem was that it's entirely possible to write a chess engine proxy that sits between the PeerNode and the actual chess engine program. This would offer an easy way to alter results before sending them to the SuperNode server. We decided to link the engine component statically because of two key advantages, tighter security and function-based rather than I/O-based messaging.
The surge of interest from Slashdot soon made it necessary to reduce ChessBrain's bandwidth requirements. To this end, the use of SOAP offered many advantages, but its size left much to be desired. The Zlib data compression library (www.zlib.org) is now used prior to encryption to reduce the size of SOAP-based messaging. Naturally, adding compression and encryption reduces the potential for interoperability; however, the XML encryption specification (www.w3.org/TR/xmlenc-core) offers an alternative approach.
|Designing Electronics with Linux||May 22, 2013|
|Dynamic DNS—an Object Lesson in Problem Solving||May 21, 2013|
|Using Salt Stack and Vagrant for Drupal Development||May 20, 2013|
|Making Linux and Android Get Along (It's Not as Hard as It Sounds)||May 16, 2013|
|Drupal Is a Framework: Why Everyone Needs to Understand This||May 15, 2013|
|Home, My Backup Data Center||May 13, 2013|
- RSS Feeds
- Dynamic DNS—an Object Lesson in Problem Solving
- Making Linux and Android Get Along (It's Not as Hard as It Sounds)
- Using Salt Stack and Vagrant for Drupal Development
- New Products
- Designing Electronics with Linux
- A Topic for Discussion - Open Source Feature-Richness?
- Drupal Is a Framework: Why Everyone Needs to Understand This
- Validate an E-Mail Address with PHP, the Right Way
- What's the tweeting protocol?
- Kernel Problem
3 hours 4 min ago
- BASH script to log IPs on public web server
7 hours 31 min ago
11 hours 7 min ago
- Reply to comment | Linux Journal
11 hours 39 min ago
- All the articles you talked
14 hours 3 min ago
- All the articles you talked
14 hours 6 min ago
- All the articles you talked
14 hours 7 min ago
18 hours 32 min ago
- Keeping track of IP address
20 hours 23 min ago
- Roll your own dynamic dns
1 day 1 hour ago
Enter to Win an Adafruit Pi Cobbler Breakout Kit for Raspberry Pi
It's Raspberry Pi month at Linux Journal. Each week in May, Adafruit will be giving away a Pi-related prize to a lucky, randomly drawn LJ reader. Winners will be announced weekly.
Fill out the fields below to enter to win this week's prize-- a Pi Cobbler Breakout Kit for Raspberry Pi.
Congratulations to our winners so far:
- 5-8-13, Pi Starter Pack: Jack Davis
- 5-15-13, Pi Model B 512MB RAM: Patrick Dunn
- 5-21-13, Prototyping Pi Plate Kit: Philip Kirby
- Next winner announced on 5-27-13!
Free Webinar: Hadoop
How to Build an Optimal Hadoop Cluster to Store and Maintain Unlimited Amounts of Data Using Microservers
Realizing the promise of Apache® Hadoop® requires the effective deployment of compute, memory, storage and networking to achieve optimal results. With its flexibility and multitude of options, it is easy to over or under provision the server infrastructure, resulting in poor performance and high TCO. Join us for an in depth, technical discussion with industry experts from leading Hadoop and server companies who will provide insights into the key considerations for designing and deploying an optimal Hadoop cluster.
Some of key questions to be discussed are:
- What is the “typical” Hadoop cluster and what should be installed on the different machine types?
- Why should you consider the typical workload patterns when making your hardware decisions?
- Are all microservers created equal for Hadoop deployments?
- How do I plan for expansion if I require more compute, memory, storage or networking?