term1 Definition1term2 Definition2term3 Definition3
Please sign in to your Google account to access your documents:
parallel programming challenges
1) race: two concurrent instruction sequences access the same address in memory and at least of of them writes to the address
2) starvation: processor needs a resource but never gets it
3) deadlock:
4) livelock: a processor keeps retrying an operation that always fails.
implement crawling under the constraint that in any given minute your crawlers do not request more than b bytes from any website
- how do we resolve a contention: FIFO? priority?
- maintain a permission server with which each crawler checks to see if it's okay to hit the particular host. Do we care about priorities? the server may enqueue the request and inform the crawler when is it right to download the file. IF the permission server becomes a bottleneck, use multiple permission server.
how would you sort a billion 1000 byte string? How about a trillion 1000 byte string?
1) billion 1000 byte string: to partition data into smaller blocks that fit in RAM, sort each block individually, and write the sorted block to disk and then combine the sorted blocks.
2) trillion 1000 byte string: distributed across a cluster of machines. organize the data so that lookups can be performed via binary search ( for each machine). - do we know how data is already distributed?
Need help typing ? See our FAQ (opens in new window)
Please sign in to create this set. We'll bring you back here when you are done.
Discard Changes Sign in
Please sign in to add to folders.
Sign in
Don't have an account? Sign Up »
You have created 2 folders. Please upgrade to Cram Premium to create hundreds of folders!