Distributed Systems (COMP90015_2020_SM1) 複習

Distributed Systems (COMP90015_2020_SM1)

該博客只限於該課程的複習，我只是在這裏記錄框架，並非全部內容。

1. introduction

1.1 define

1.2 computer network && ds

1.3 reason--6

• economy
• resource sharing
• functional separation
• reliability
• availability
• scalability

1.4 characteritics 5 

• no global state
• no global clock
• resource sharing
• communicaiton via message passing
• parallel activities

1.5 goals 5

• connecing users and resources
• transparncy
• openness
• scalability
• enhanced availability

1.6 consequences 4

• concurrency
• heterogeneity
• no global clock
• independent failures

1.7 challenges

• secruity
• openness
• concurrency
• transparency
• scalability
• heteogeneity
• fault tolerance

 

2. inter-process communication

2.1 layer - image

• physical layer --> internet layer --> transport layer -->application layer

2.2 TCP   -- image

• define: reliable / connection-oriented communicaiton 
• applicaiton: http/ftp/skype
•  data flows

2.3 UDP --image

• define: connectionless communicaiton / no guarantees order 

• applicaiotn :broadcasting / ping

• failure : order / omission failure / data corruption

2.4 port 

• positive 16-bit int
• post number >=1024
• Tcp / udp map data 

2.5 socket

• programme:client / server :udp+tcp
• define / introduction :language indepent  && interface for programming 

 

3. thread

3.1 middleware 

• define:layer between applicaiton / operting system

3.2 middleware / network operating system

3.3 threaded appliaiton

• multitasking / multithreading / bottom line

3.4 thread

• implement : thread class / runnable interface (program)
• define: a piece of code
• 生命週期 -- image
• priority ： set / get ：3 types

3.5 architecture

• worker pool -- image
• thread -per - request
• thread - per - connection 
• thread - per - objects

3.6 thread && process

• process-based:controlled by os
• thread-based:controlled by programmer and applicaiton program
• adv for thread:

1. share address space
2. communicaiton inexpensive
3. context swithcing inexpensive

 

4. ds models

4.1 . models 

• fundamental models / architectural models / physical model

4.2 fundamental models 

• intercation / failure / security model

4.3 architectural modesl

• clients and a single server
• multiple servers
• proxy servers with cache
• p2p

4.4 alternative client-server models

• mobile code 
• mobile agents
• network computers
• thin clients
• mobile devices
• spontaneous networking

 

 

 

5.Operating System Architecture

5.1 NOS && DOS

• nos（3） 
• dos（4）

5.2 core OS components

• processing manager
• thread manager
• memory model
• communication manager
• supervisor

5.3 supervisor && user model

5.4  popular kernel implementation methods

• Monolithic && Microkernel區別  -- image / advan / disadvan
• monolithic os : high performance + difficult to enhance
• layered  os ：easy to enhance / low performance
• Micro-kernel

5.5 json && xml

• easy program
• parser
• shorter string -- improve performance

 

6. Remote Invocation

6.1  3 types of protocols

• request
• request-reply (3 issue:timeout(client+server)  & duplicate requests 
• request - repley - acknowledgement

6.2 3 invocation semantics

• maybe 0/1
• at - least -once RPC : at least 1 time  &  idempotent operation
• at - most - once RMI : only once & non-idempotent operations.

6.3 RMI

• how to build---  image (processing)
• architecture and compoents -- image

1. remote reference module
2. proxy
3. communication model
4. dispatcher
5. skeleron

• describe：4 part：interface / implement / server / client 
• programming ：code

1. server : interface  / implement / main -- register & rebind
2. client :  lookup
3. security : eg:hellopoicy

• security manger -- implement

6.4 RPC

• define
• 4 compoents -- image

1. communication model 
2. client stub procedure
3. server stub procedure
4. dispatcher

 

7. security

7.1 threats 3 

• leakage
• tampering 
• vandalism  

7.2 attack 5

• Eavesdropping （leakage）
• Masquerading
• Message tampering
• Replaying
• Denial of service （ Vandalism）

7.3 worst-case assumptions （main4）

• interface are exposed
• networks are insecure
• node and algorithms are available to hackers
• attackers have large resources
• minimise the trusted base
• limit the lifetime and scope of each secret

7.4 encryption

• define
• shared secret keys :

1. define :use the same keys (kab)
2. process--image : 2 steps : send(kab)+get(kab)
3. drawbackes : key distributions / freshness of communicaiton

• public/private key pair

1. define: 2 keys(kpub + kpriv)
2. drawbacks: more computations
3. process : step3 -- image （get kpub --> creat kab --> get message(use kprivate))

7.5 applicaiton (major roles) encryotion 

• secrecy and integrity
• authenticaiton
• digitial signatures

7.6 PKI

• define
• goals
• certificaite(X.509)

7.7 certificates

7.8 kerberos && ssl

• 7.8.1 kerberos

1. system architecture: step 3 -- image
2. drawbacks: expensive to apply ( cannot multiple users && mounted all files)

• 7.8.2 ssl(tls)

1. two layers: handshake layer / tcl record protocol layer ( detail -- image)

 

8. Distributed file systems

8.1  issue

• recovery （after failure）
• high throughput ( reading and writing)
• scalability
• consistency
• cilent caching --> performance

8.2 transparencies

• access
• location
• mobility
• performance
• scaling 

8.3 file service architecture

• define :副本緩存在本地
• compoents：

1. a flat file service :operations
2. a directory service:mapping -- ufids
3. a client module :intefrated services

• architecture -- image

8.4 DFS (NFS)

• architecture -- image
• operariton -- functions(client/server) -programme
• NFS valid / invalid

1. valid : time ( fresh enough)  or (last modified recorded by the client and the server match)

8.4 cache / memory 

• disk : adv: safer dis:slower need a disk
• memory : adv: quick dis: limited cache size / not tolerate node failure well

8.5 write-through client cache update ( good / bad)

• define: write are propagated to the server, server's copy updated immediately
• adv: high reliability 
• dis 

1. reduced performance ( transmitted over the network)
2. no benefit for write requests ,but good for reads

8.6 absolute name(good / bad)

• define: a complete address : both server and path
• adv:

1. easy to find
2. no additional state
3. greater scalability
4. eaty to add and delete

• dis:

1. no location transparency
2. file cannot be moved
3. less resilient to failure

8.7 a global name (good / bad)

• define:
• adv:
• location transparency

1. name consistent across all clients
2. can move

• dis :

1. cilent(cached) -- difficult to consistent
2. performance problems

8.8 mount point (good / bad )

 

9. name services

9.1 URL

• define:
• different : URI/URL/URN 

9.2 name services

• define: consistent and uniform naming of resources
• important (why) 3 

1. recource localizations
2. uniform naming
3. device independent address

• role

9.3 process of name resolution

• interative process (get attributes or name)
• never terminate(solution:abandon / no cycles occur)

9.4 navigation

• define: (art of chaining multiple naming services && resolve name to resource)
• approaches : iterative / recursive / non-recursive  -image

9.5 DNS

• define ： a name service design
• functions: resolve the name(main)
• issus(2)
• hierarchical organizaiton
• example -- image
• diffierent with file-based : scale / privacy / speed)

9.6 directory services