Techniques Recovery
thus far allow failure handling Recovery: operations that must be performed after a failure to recover to a correct state Techniques:
Checkpointing: Periodically checkpoint state Upon a crash roll back to a previous checkpoint with a consistent state
Computer Science
CS677: Distributed OS
Lecture 18, page 1
Independent Checkpointing
Each processes periodically checkpoints independently of other processes Upon a failure, work backwards to locate a consistent cut Problem: if most recent checkpoints form inconsistenct cut, will need to keep rolling back until a consistent cut is found Cascading rollbacks can lead to a domino effect.
Computer Science
CS677: Distributed OS Lecture 18, page 2
1
Coordinated Checkpointing
Take a distributed snapshot [discussed in Lec 11]
Upon a failure, roll back to the latest snapshot
All process restart from the latest snapshot
Computer Science
CS677: Distributed OS
Lecture 18, page 3
Message Logging
Checkpointing is expensive
All processes restart from previous consistent cut Taking a snapshot is expensive Infrequent snapshots => all computations after previous snapshot will need to be redone [wasteful]
Combine checkpointing (expensive) with message logging (cheap)
Take infrequent checkpoints Log all messages between checkpoints to local stable storage To recover: simply replay messages from previous checkpoint Avoids recomputations from previous checkpoint
Computer Science
CS677: Distributed OS Lecture 18, page 4
2
Recovery Oriented Computing
Cheaper to optimize for recover than to design the system to prevent faults Need to restart the system upon failure Nave case: reboot Reboot part of the system: modular system, where components can be restarted independently
Unix /etc/rc service
Stateful recovery
Database recovery Use of checpointing
Computer Science
CS677: Distributed OS Lecture 18, page 5
Security in Distributed Systems
Introduction Cryptography Authentication Key exchange Readings: Tannenbaum, chapter 9 Ross/Kurose, Ch 7
Computer Science
Lecture 18, page 6
3
Network Security
Intruder may
eavesdrop remove, modify, and/or insert messages read and playback messages Security threats
Interception, Interruption, Modification, Fabrication
Computer Science
Lecture 18, page 7
Issues
Important issues:
Encryption/ cryptography: secrecy of info being transmitted authentication: proving who you are and having correspondent prove his/her/its identity Authorization: verify you have rights to perform requested action Auditing: log actions and do post-facto analysis (forensics)
Computer Science
Lecture 18, page 8
4
Security in Computer Networks
User resources:
login passwords often transmitted unencrypted in TCP packets between applications (e.g., telnet, ftp) passwords provide little protection
Computer Science
Lecture 18, page 9
Security Issues
Network resources:
often completely unprotected from intruder eavesdropping, injection of false messages mail spoofs, router updates, ICMP messages, network management messages
Bottom line:
intruder attaching his/her machine (access to OS code, root privileges) onto network can override many systemprovided security measures users must take a more active role
Computer Science
Lecture 18, page 10
5
Encryption
plaintext: unencrypted message ciphertext: encrypted form of message Intruder may
intercept ciphertext transmission intercept plaintext/ciphertext pairs obtain encryption decryption algorithms
Lecture 18, page 11
Computer Science
A simple encryption algorithm
Substitution cipher:
abcdefghijklmnopqrstuvwxyz poiuytrewqasdfghjklmnbvczx replace each plaintext character in message with matching ciphertext character: plaintext: Charlotte, my dear ciphertext: iepksgmmy, dz uypk
Computer Science
Lecture 18, page 12
6
Encryption Algo key (contd)
is pairing between plaintext characters and ciphertext characters symmetric key: sender and receiver use same key 26! (approx 10^26) different possible keys: unlikely to be broken by random trials substitution cipher subject to decryption using observed frequency of letters
'e' most common letter, 'the' most common word
Computer Science
Lecture 18, page 13
DES: Data Encryption Standard
encrypts data in 64-bit chunks encryption/decryption algorithm is a published standard
everyone knows how to do it
substitution cipher over 64-bit chunks: 56-bit key determines which of 56! substitution ciphers used
substitution: 19 stages of transformations, 16 involving functions of key
Replacements: DES3 and now AES
Computer Science
Lecture 18, page 14
7
Symmetric Cryptosystems: DES (1)
a) b)
Computer Science
The principle of DES Outline of one encryption round
Lecture 18, page 15
Symmetric Cryptosystems: DES (2)
Details of per-round key generation in DES.
Computer Science
Lecture 18, page 16
8
Key Distribution Problem
Problem: how do communicant agree on symmetric key?
N communicants implies N keys
Trusted agent distribution:
keys distributed by centralized trusted agent any communicant need only know key to communicate with trusted agent for communication between i and j, trusted agent will provide a key
Computer Science
Lecture 18, page 17
Key Distribution
We will cover in more detail shortly
Computer Science
Lecture 18, page 18
9
Public Key Cryptography
separate encryption/decryption keys
receiver makes known (!) its encryption key receiver keeps its decryption key secret
to send to receiver B, encrypt message M using B's publicly available key, EB
send EB(M)
to decrypt, B applies its private decrypt key DB to receiver message:
computing DB( EB(M) ) gives M
Computer Science
Lecture 18, page 19
Public Key Cryptography
knowing encryption key does not help with decryption; decryption is a non-trivial inverse of encryption only receiver can decrypt message
Question: good encryption/decryption algorithms
Computer Science
Lecture 18, page 20
10
RSA: public key encryption/ decryption
RSA: a public key algorithm for encrypting/decrypting Entity wanting to receive encrypted messages: choose two prime numbers, p, q greater than 10^100...