Fork Join
Download as PPTX, PDF•1 like•836 views
The document discusses the Fork/Join framework in Java for parallel programming. It describes how problems can be split into independent subtasks that are solved concurrently using multiple threads, and then joined together to compose the final result. As an example, it shows how the Fork/Join framework can be used to speed up a rainbow table attack to crack password hashes by parallelizing the generation of possible password combinations and hash computations across multiple CPU threads. Performance tests on Amazon EC2 show the parallel implementation using Fork/Join was able to crack 1000 MD5 password hashes much faster than a single-threaded approach.
Fork Join
- 1. Fork/Join frameworkRainbow attack examplePhilip [email protected]: philipsavkin
- 2. Fork Join FrameworkJava framework for supporting a style of parallel programming in which problems are solved by (recursively) splitting them into subtasks that are solved in parallel, waiting for them to complete, and then composing results.http://gee.cs.oswego.edu/dl/papers/fj.pdf
- 3. AlgorithmResult solve(Problem problem) { if (problem is small) directly solve problem else { split problem into independent parts fork new subtasks to solve each part join all subtasks compose result from subresults }}
- 4. Core classes
- 5. Differences from ThreadPoolExecutorSuited for recursive tasks and “divide and conquer” algorithmsWork stealing
- 6. Work StealingInitial stateWait for all threads to complete their tasks
- 7. Work StealingOther threads are idle until the first thread finishedWait for all threads to complete their tasks
- 8. Part II
- 9. Practical examplePasswords are rarely kept in cleartext nowadaysThe problem: restore passwords from a list of password hashes with Rainbow attack
- 10. Rainbow attackhttp://en.wikipedia.org/wiki/Rainbow_attackBuild a Rainbow table using the list of possible passwordsLookup passwords in the table
- 11. The algorithmLoad top 2000 english wordsAdd all case permutationsAdd numbers 0-9Results in 6 000 000 combinationsCompute hashesLookup hashes in the table
- 12. How ForkJoin can help?CPU intensive tasksGenerate the list of all possible passwordsCompute hashes
- 13. Let’s see some code!
- 14. Test resultsTested on Amazon EC2 Extra Large instance running 64 bit AMI Linux15 Gb RAM, 4 processorsRainbow table size: 6 041 508Input: list of 1000 MD5 hashesFound all 10 passwords
- 15. Test results
- 16. Bonus slide - OfftopicNever keep passwords in cleartext!MD5 is a bad choiceAlways add “salt” to passwordsThe right way: use Bcrypt!
- 17. More offtopicThe “pastebin” experimenthttp://pastebin.com/1iL2P0G5Found one password “a1”