230 likes | 357 Views
Abusing Cloud-Based Browsers for Fun and Profit. Vasant Tendulkar, Joe Pletcher , Ashwin Shashidharan , Ryan Snyder, Kevin Butler, William Enck. 2012 Annual Computer Security Applications Conference. Outline. Introduction Approach overview Browser resource limitations
E N D
Abusing Cloud-Based Browsers for Fun and Profit Vasant Tendulkar, Joe Pletcher, AshwinShashidharan, Ryan Snyder, Kevin Butler, William Enck 2012 Annual Computer Security Applications Conference
Outline • Introduction • Approach overview • Browser resource limitations • Designing and scheduling jobs • Evaluation • Discussion
Introduction • Pay-per-use computation • EC2 • Cloud-based Web browsers • JavaScript executes on the server • Arbitrary general-purpose computation within cloud-based browsers • Browser MapReduce(BMR) • Mapper.js • Reducer.js
Approach overview(2) • Cloud browsers have articiallimitations • Mappers cannot use local storage to communicate intermediate results • Bit.ly • Encode URLs up to 2022 characters in length • Rate-limits requests to 99 per IP address per minute • Job scheduling
Browser resource limitations • Benchmarks • Computation • Elapsed Time • Memory • Cloud browsers • Amazon Silk • Cloud Browse • Opera Mini • Puffin
Designing and scheduling jobs • Map and reduce abstraction • Scheduling jobs • Example applications • Word Count • Distributed Grep • Distributed Sort
Scheduling Jobs(1) • To effectively use cloud browser an URL shortening service resource • Assume • Input is divided into a large number of equally sized files • Mapper scheduling • Reducer scheduling
Mapper Scheduling(1) • Master determines • Mn, the number of mappers to spawn • Mf, the number of input files to pass to each mapper Assumes fs is several times smaller than bs
Mapper Scheduling(2) • Rate limiting of URL shortening services • Bit.ly • Encode URLs up to 2022 characters in length, Us = 2022 • Rate-limits requests to 99 per IP address per minute, Un = 99
Reducer Scheduling • Application spefic • bucket in bucket sort • Partition in word count
Example Applications • Word Count • http://foo.com/?word1=5&word2=7&... • Distributed Grep • http://foo.com/?bar1.txt=45&bar1.txt=48&bar2.txt=34 • Distributed Sort • TeraSort • http://foo.com/?key1=file1&key2=file2&key3=file3
Evaluation • Experiment setup • Word Count • 100 most downloaded books from www.gutenberg.com/ebooks/ • Distributed Grep • IRC logs for the #debian channel • Distributed Sort • Hadoopteragen • A 10 character key and an 88 character value
Discussion • Recommendations for Cloud Browser Providers • Rate limits on authenticated user • Enhancing BMR • Using multiple cloud browser • Combination of URL shortening services