DOC PREVIEW
Duke CPS 210 - Midterm Exam

This preview shows page 1-2 out of 6 pages.

Save
View full document
View full document
Premium Document
Do you want full access? Go Premium and unlock all 6 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 6 pages.
Access to all documents
Download any document
Ad free experience
Premium Document
Do you want full access? Go Premium and unlock all 6 pages.
Access to all documents
Download any document
Ad free experience

Unformatted text preview:

CPS 210 Midterm Exam Spring 2010 This is a wordy exam asking you to explain things rather than to work out problems. Answers are graded on content and substance, not style. Please do not waste any words. I am looking for understanding and insights, rather than details. Please give the most complete answer that you can in the space provided for each question. You may attach additional pages if necessary. Watch the time: you have 75 minutes. This is a closed-everything exam, except for your one page (two sides) of notes. The five questions are each worth 40 points: 200 points total. Short Answers Q1. Answer the following questions in the space provided. Since there isn’t much space, you will have to gloss over some details, but that is OK. (a) Data cached at multiple clients can become stale if it is updated. Many distributed systems have taken a position that stale data is acceptable if it is not stale for “too long” (e.g., HTTP, DNS, and early NFS). Outline a technique to control staleness in such systems. (b) What support is needed from the operating system kernel to implement heap allocation primitives? (E.g., support for malloc and free, or new and delete).Short Answers on File Systems Q2. Answer the following questions in the space provided. Since there isn’t much space, you will have to gloss over the details, but that is OK. (a) In the Unix file system, data written to a file (i.e., in write system calls that complete with success) may be lost after recovery from a system crash. However, the original Network File System (NFS) requires a file server to commit each write before returning success. Why the discrepancy, i.e., why are weak assurances “good enough” in one case but not the other? (b) A read system call may cause a process to block (wait or sleep) in the kernel. Why? In each case, what causes the process to wake up? (c) Why does Unix designate specific file descriptors as “standard” for standard input (stdin), standard output (stdout), and standard error (stderr)?Short Answers on Cryptosystems Q3. Answer the following questions in the space provided. Since there isn’t much space, you will have to gloss over the details, but that is OK. (a) Are digital signatures on digital documents more or less secure than ink signatures on printed documents? Justify your answer. (b) Certificates enable the bearer to authenticate to another entity without the involvement of a third party (e.g., a certifying authority) at the time of authentication. How does this work? (c) In secure DNS (DNSSEC), each zone server holds an asymmetric keypair. Why use an asymmetric cryptosystem rather than symmetric crypto, which is less costly?More File Systems and WAFL Q4. To handle reads or writes on files, file systems must identify the physical disk locations where logical file blocks reside on disk. This involves lookups through layers of naming structures. WAFL uses similar data structures to classical Unix file systems. (Let’s ignore WAFL’s structures for aggregates and logical volumes or flexvols.) But WAFL makes some changes to support more flexible placement of blocks on disk. (a) Outline the steps to map a logical block to a physical block in WAFL. You do not need to describe the data structures themselves in detail, just the sequence of mapping steps. (b) What advantages does WAFL’s flexible block placement offer? Consider the handling of both reads and writes.(c) Outline the implementation choices that WAFL made to enable this flexible block placement, in terms of differences from the Unix file system implementation. Services and Binding Q5. Any client/server system (e.g., an RPC system) must define some method to establish a binding between the client and a server. How does the client determine who and where the server is (e.g., its IP address, port number)? How does the client verify that it is talking to the correct server? How does the server authenticate the client and determine if the client is authorized to invoke the service or operate on specific objects? Summarize and contrast the approaches to binding and authorization for three systems we have discussed: NFS, DNS, and Web/HTTP. I am asking about “plain vanilla” versions of these systems as discussed in class (e.g., plain NFSv3, DNS and not DNSSEC, HTTP including HTTPS). (Your answer may continue on the next


View Full Document
Download Midterm Exam
Our administrator received your request to download this document. We will send you the file to your email shortly.
Loading Unlocking...
Login

Join to view Midterm Exam and access 3M+ class-specific study document.

or
We will never post anything without your permission.
Don't have an account?
Sign Up

Join to view Midterm Exam 2 2 and access 3M+ class-specific study document.

or

By creating an account you agree to our Privacy Policy and Terms Of Use

Already a member?