Questions

Write down your answers in a piece of paper, and turn it in at the beginning of each class. You might be asked in class to exlain your answer. You can find the answers in the "readings" and "optional readings" sections.

feb 5: security analysis

  • what constitue of a cloud attack surface? (hint: consider a cloud service model)
  • why is it hard to make a system secure?
  • what is the vulnerability exploited in a "free shopping" paper?

feb 12: container security

  • what is the conceptual diff. between container/VM instance/sandbox/libOS (from the trust-model perspective)?
  • why do we need libOS?

feb 19: db optimizer

  • what is the idea behind db optimizer? what makes it useful/possible?
  • what is yet to be solved in the "DBO solved" paper?
  • what is "site selection" problem? and how is it solved?

feb 26: data federation

  • what is the typical work in data federation?
  • what makes data federation difficult?

mar 11: consistency attacks

  • what is consistency defined in "txos" paper?
  • what is consistency defined in "fork consistency" paper?
  • what is consistency defined in "certificate transparency" paper?

mar 25: TEE

  • what is the model of TEE?
  • How is SGX differentiated from TPM/TXT?
  • How is TEE related to cloud?

apr 15: TEE DB

  • what is the security properties addressed in CorrectDB?
  • what is the security properties addressed in TrustedDB?
  • what is the gap between those db-level security and TEE security?

apr 22: multi-party computing

  • what is the model (input and output) of MPC? what does it guarantee in security?
  • what applications would MPC be used for in real life?
  • what makes verifiable computing expensive?

apr 29: crypto-homomorphism

  • what is the interface of homomorphic signature? compare it with verifiable computing
  • what extra information is exposed by using OPE?
  • compare and contrast the system architecture of cipherbase and cryptodb?

may 3: tor & verifiable program

  • what is traffic analysis? and how to apply it to tor network?
  • why is it costly for verifiable program?
  • what are the ideas to make circuit succinct?