Deadlock Detection for Distributed Process Networks

1
Deadlock Detection for Distributed Process
Networks

• Alex Olson
• Embedded Software
• Spring 2004

2
Problem Statement

• Implement deadlock detection in a distributed
  process network (DPN)
• Deadlock detection required for PN scheduling
  algorithms
• Deadlock not statically predictable
• DPN allows for high performance computations
• Distributed Process Network
• Variable channel latency
• Limited bandwidth
• No central server

3
Goals

• Achieve high-performance
• Deadlock detection should have minimal impact on
  system performance, resource usage
• Deadlock detection should support future addition
  of dynamic process migration.

4
Implementation

• Distributed Computational Process Network
  (Allen00)
• Zero copy queues
• C, POSIX Threads
• Transmit and receive queues
• TCP socket for communication
• Simple Deadlock Detection Algorithm

5
Deadlock Detection

• Mitchell Merritts algorithm (Mitchell84)
• Single-resource model
• Detects local and global deadlocks
• Each Process has two labels
• Each label is (count, pid)
• Guarantees only one process detects deadlock
• Send tokens and control information on same
  socket, make use of FIFO guarantee
• No synchronization mechanism required

6
Mitchell-Merritt Example
pub
priv
7
Mitchell-Merritt Example
BLOCK
pub
priv
8
Mitchell-Merritt Example
BLOCK
pub
priv
9
Mitchell-Merritt Example
BLOCK
pub
priv
10
Mitchell-Merritt Example
Transmit
pub
priv
11
Mitchell-Merritt Example
Transmit
pub
priv
12
Mitchell-Merritt Example
DEADLOCK
pub
priv
13
Interface

• Processes inherit methods from a Process class
  and implement the following
• void Init()
• void Step()
• int GetPID()

14
Performance

• Up to 2 Million 1-Byte tokens per second
• 120 Mbytes/sec (1kb tokens)
• Verified deadlock detection for a a variety of
  cases

15
Performance Statistics Token Rate
16
Performance Statistics Data Rate
17
Conclusion

• Deadlock detection in DPNs can be fairly simple
• Method can be applied to regular PNs
• DPNs capable of high performance
• Future Work
• Process Migration
• Load Balancing

18
References
Allen00 G. Allen and B. Evans, Real-time sonar beamforming on workstations using process networks and POSIX threads, in IEEE Trans. Signal Processing, Mar. 2000, pp. 921926.
Mitchell84 D. P. Mitchell and M. J. Merritt, A distributed algorithm for deadlock detection and resolution, in ACM Symposium on Principles of Distributed Computing, 1984, pp. 282 284.