Fault Tolerance for Compound Web Services

1
Fault Tolerance for Compound Web Services

• Geoffrey Gamble
• CSE 294
• 2/15/08

2
References

• Thema Byzantine-fault-tolerant middleware for
  Web-service applicationsMerideth, M.G. Arun
  Iyengar Mikalsen, T. Tai, S. Rouvellou, I.
  Narasimhan, P., Reliable Distributed Systems,
  2005. SRDS 2005. 24th IEEE Symposium on , vol.,
  no., pp. 131-140, 26-28 Oct. 2005BASE Using
  abstraction to improve fault tolerance
• Castro, M., Rodrigues, R., and Liskov, B. 2003.
  ACM Trans. Comput. Syst. 21, 3 (Aug. 2003),
  236-269. DOI http//doi.acm.org/10.1145/859716.85
  9718Practical Byzantine fault toleranceMiguel
  Castro , Barbara Liskov, Proceedings of the third
  symposium on Operating systems design and
  implementation, p.173-186, February 1999, New
  Orleans, Louisiana, United States

3
Questions to be Addressed

• What is fault tolerance?
• What kind of faults are we concerned about?
• Why is fault tolerance important in distributed
  systems?
• How do we increase the fault tolerance of a
  distributed system?
• Can we adapt traditional fault tolerance
  solutions to web services based systems?

4
What Is Fault Tolerance?

• Fault Tolerance
• Property of a system which allows it to continue
  operating properly when some of its components
  fail

5
Properties of a Fault Tolerant System

• No Single Point of Failure
• Uninterrupted Repair
• Fault Containment

6
What Kinds of Faults are We Talking About?

• Crash
• - Servers O.S. crashes
• - Servers hardware fails
• - Program halts due to unexpected conditions
• Incorrect Execution
• - Program continues to execute but the results
  are undesirable
• - Malicious agent has hijacked a node

7
Importance of Fault Tolerance
THE MONOLITHIC APPLICATION
8
Importance of Fault Tolerance
9
Importance of Fault Tolerance
10
Importance of Fault Tolerance
11
Increasing the Fault Tolerance of a Distributed
System

• Byzantine Fault Tolerance (BFT)

12
Standard Byzantine Fault Tolerance Example
13
Standard Byzantine Fault Tolerance Example
14
Problems With Standard BFT Implementations
15
Applying BFT to Distributed Web Service Based
Applications

• A framework called THEMA brings BFT to web
  service/SOA based systems
• - Provides BFT to distributed applications
  composed of compound web services
• - Web services easily communicate across
  organizational bounds while maintaining BFT
• - Supports a 'mixed-fault' environment

16
THEMA Enables BFT For Web Services
17
How Does THEMA Work?
18
Performance Ramifications

• Results for TPC-W benchmark paint Thema in a
  favorable light, but are questionable upon
  examination

19
Questions That Have Been Addressed

• What is fault tolerance?
• What kind of faults are we concerned about?
• Why is fault tolerance important in distributed
  systems?
• How do we increase the fault tolerance of a
  distributed system?
• Can we adapt traditional fault tolerance
  solutions to web services based systems?

20
Conclusions

• Trade-offs, Thema BFT at what price?
• Costs/Drawbacks
• Computational overhead of extra network layer
• Need for external clients and services to
  install Thema
• Massive increase in necessary computational
  resources
• System management becomes much more difficult

21
Conclusions

• Trade-offs Thema BFT at what price?
• Benefits
• Theoretically, this can lead to a robust web
  service based system
• A good fit for large, distributed systems where
  availability/reliability are a primary concern
• Makes BFT more accessible by harmonizing with
  SOAP/WSDL
• Makes BFT transparent to developers who
  understand web service protocols
• Opens closed BFT systems to web services outside
  of organizational bounds