1. Aspects of Distributed systems
Luk Stoops
VUB
4/13/2019

2. Organization of computing facilities
Distributed Systems
Organization of computing facilities
4/13/2019

3. Substantial Benefits Resource sharing Openness Concurrency Scalability
Transparency
Flexibility
Reliability
Performance

4. Resource sharing Fundamental characteristic Resources Data Software
catalog items
Software
Hardware
Managed by
server process (client server)
cooperating processes
Object based system is most flexible

5. Openness Technical Business Interfaces are published
World Wide Web is a typical example
HTML (XML)
HyperText Markup Language
HTTP
HyperText Transport Protocol

6. Concurrency Higher performance
Catalog can be consulted by several clients at the same time
selective publishing
selective pricing
promotions
ordering
delivery

7. Scalability... Replication Netscape DNS Server caching IP address !
Server caching IP address !
Works only worldwide
Statistically even distributed
Load distribution
Negotiating servers
Mobile agents

8. Scalability
No machine has complete information about the state of the system
Machines make decisions based only on locally available information
Failure of one machine does not ruin the algorithm
No implicit assumption that a global clock exists

9. Transparency Location transparent
user can not tell where resources are located
Migration transparent
resources can move at will without changing their names
Replication transparent
the user cannot tell how many copies exist
Concurrency transparent
multiple users can share resources automatically
Parallelism transparent
activities can happen in parallel without users knowing

10. Flexibility Adding and removing servers at runtime
Interoperability standards exist (CORBA)
Object based system
message B A
response

11. Reliability Boolean OR of components ?
5 servers 0.95 chance of being up
Chance all down = =
chance 1 ore more available
But mostly some servers are crucial for the system

12. Reliability Aspects Availability
no requirement simultaneous functioning
redundancy
file copies
consistency !
Security
prevent unauthorized usage
digital certificates instead of login procedure
Fault tolerance
server crash ?
stateless design
hardware redundancy

13. Performance Response time Troughput System utilization
Amount of network capacity consumed
Number of messages
small: not much protocol handling overhead
big: many activities running in parallel

14. Managing Distributed Systems
Central
+ No searching
- Server down ?
Distributed
+ Less vulnerable
- Where to search ?

15. Layered Architecture Applications Catalog Browser
Compression Encryption
Remote Procedure Call
Multimedia Support
Error - & Flow Control
Support
Communication

16. Non - Distributed System (single processor)

17. Client - Server Distribution Model

18. Forwarding Distribution Model
User Agent
asymmetric
Message Transfer Agent symmetric

19. Distributed Systems Models
Communication (distribution visible)
File Transfer
user is aware of the communication
Distribution (distribution invisible)
File Serving (Network File System)
user is not aware of the communication

20. PROCEDURE inc (VAR x,y);
Remote Procedure Call
Allowing programs to call procedures located on other machines
Parameter problems
VAR parameters
copy - restore
PROCEDURE inc (VAR x,y);
x := x + 1 ; y := y + 1
I := 0 ;
inc ( I, I );
print ( I ) 1

21. Interprocess Communication...
Race Conditions (sharing resources)
Mutual exclusion
Critical sections
No two processes simultaneously in their critical sections
No assumptions about speeds or the number of CPUs
No process running outside its critical section may block other processes
No process should have to wait forever to enter its critical session

22. Interprocess Communication...
Semaphores
UP : atomic increment by 1
DOWN : if 0 then sleep else atomic decrement by 1
Monitors
only one process can be active in a monitor
In distributed systems
Semaphores too low level
Monitors not usable

23. Interprocess Communication...
Message Passing
SEND
RECEIVE
Messages can be lost
No acknowledge then retransmit (TCP)
Ack can be lost
Each original message: sequence number
Authentication by digital certificates

24. Interprocess Communication...
The Dining Philosophers Problem

25. The Dining Philosophers Problem
think
Take left stick
Take right stick
yum-yum, rice
put left stick back
put right stick back
Suppose that all take their left stick simultaneously
Deadlock

26. The Dining Philosophers Problem
Think
Take left stick
Check to see if right fork is available
If not, put left stick back wait some time and repeat
Take right stick
Yum-yum, rice
Put left stick back
Put right stick back
Suppose that all start simultaneously
Starvation

27. The Dining Philosophers Problem
Think
Take left stick
Check to see if right fork is available
If not, put left stick back wait a random time and repeat
Take right stick
Yum-yum, rice
Put left stick back
Put right stick back
unlikely series or random numbers?

28. The Dining Philosophers Problem
Think
DOWN on mutex
Take left stick
Take right stick
Yum-yum, rice
Put left stick back
Put right stick back
UP on mutex
Only one philosopher can eat

29. Distributed Coördination
Safety
one at the time
Ordering
first comes first served
Liveness
acceptable wait time

30. Distributed Coordination Centralized Approach
Server gives token to first asker
Safety: OK
Ordering: OK
Liveness:
server down
client down
network fault
ACK
Time-out P2 P3
Token server

31. Distributed Coordination Distributed Approach
Process asks token (broadcast)
All other processes answer
Ti (counter) different for all Pi
Ti increases if token in possession
Multicast: Ti
Wait for (n-1) replies
If Ti < Tx then keep it , else grant
Liveness problem
ACK, time-out + retry P2 P3

32. Distributed Coordination Token Ring Distributed Approach
Logical token ring
No Ordering
Liveness
message lost
Process with token can crash
ACK lost
2 tokens
increase token# P2 P3

33. Election Algorithms Electing a coordinator Bully algorithm
P sends an ELECTION message to every process with a higher number
If no one responds, P wins and becomes coordinator
If one of the higher-ups answers, it takes over and start a new election

34. Bully Algorithm Example4 2 7 3 6 1 5 OK 4 2 7 3 6 1 5
election 4 2 7 3 6 1 5
election 4 2 7 3 6 1 5
coordinator 4 2 7 3 6 1 5 OK

35. Atomic Transactions Higher level abstraction Hiding mutual exclusion
critical region management
deadlock prevention
crash recovery
Object Transaction Service (CORBA)

36. Atomic Transactions Business transactions model New inventory Computer
updates
New
inventory
Withdraw (amount, account1)
Deposit(amount, account2)

37. Concurrence Control Locking shared read locks exclusive write locks
Distributed coordination
election for the coordinator
coordinator implements lock management