CS606: Advanced Operating Systems				   Lecture 2

Overall organization:
	- Classic forms of distributed systems
	- Distributed system design
	- End to End Argument
	- Student Lectures
	- Projects

----------------------------------------------------------------------------

Topic: Classical forms of distributed systems

    * Client-server:

	Idea: To single-system abstraction, bundle common system services
	      from many machines to run in one "central" server.  Any
	      client machine wishing to get access to this service contacts
	      this server.

	<Draw sample system>

	Common forms of servers:
	    - file server
	    - print server
	    - time server
	    - directory (name) server
	    - network server (DNS)
	    - process server (as you get more integrated)
	    - security server
	    - authentication server

	    Q: What are some other kinds of servers?  <Ask class.>

    * Peer-to-peer (conventional):

	Idea: Rather than centralizing each service, let each node provide
	      a local version of the service, but make it available for
	      remote nodes.

	Example: LANtastic or Win95 shares

	<Draw sample system>

	In many ways, this is like client-server, but without as much
	transparency or performance.

    * Server clusters:

	Idea: Have a collection of nodes COOPERATE to provide a service.
	      Could think of this as a cross between C/S and peer-to-peer,
	      with high degrees of integration, replication, migration, 
	      fault tolerance, load balancing, etc.

	Example: VAXclusters, Inktomi's web search engine, many others.

	<Draw sample system>

	Advantages:
	    - scalability
	    - high availability (fault tolerant)
	    - load balancing

	Disadvantages:
	    - concurrency control
	    - complexity

    * Peer (client) clusters: [emerging]

	Idea: Do away with the strict notion of clients and servers.
	      Do away with the strict separation of functions vs peer-to-peer.
	      Each machine adds some of its resources to the `pool,' and
	      as an aggregate they provide some service.

	Example: Mango Medley	      

	Advantages:
	    - scalability
	    - high availability (fault tolerant)
	    - cost (no servers - fewer `overhead' resources)

	Disadvantages:
	    - complexity
	    - harder fault tolerance problems

	http://www.infoworld.com/cgi-bin/displayNew.pl?/metcalfe/metcalfe.htm

    * "True" distributed system (aka, single system image)

	Idea: Build a totally dynamic, self-healing, self-tuning,network-wide
	      operating system that manages the entire collection of resources
	      in the system as a single, large, powerful machine.

	Examples (partial): Apollo Domain, V, Locus, DCE, Orca.

	<Draw sample system>

	Unique features:
	    - integrated process management (adaptive cross-machine
	      scheduling and load balancing)
	    - integrated resource management
	    - process and service migration
	    - automatic service replication
	    - strong support for concurrency control

----------------------------------------------------------------------------

Topic: Distributed system design

    So many choices... which to choose for *your* service/application?

    Issues:
	- cost
	- criticality (inverse of fault tolerance)
	- performance requirements
	- scalability requirement
	- available intellectual resources
	- modularity
	- ... other ideas?...

    Characteristics of example services:
	- directory service (NDS, Grapevine, ...):
	    - 
	    - 
	    - 

	- file service (NFS, AFS, ...):
	    - 
	    - 
	    - 

	- web server:
	    - 
	    - 
	    - 

	- proxy cache:
	    - 
	    - 
	    - 

	- game server:
	    - 
	    - 
	    -

    [Get class to come up with key characteristics of each service.
     Use these lists to talk about how you might structure each one,
     in different circumstances.]

----------------------------------------------------------------------------

Topic: End to End Argument

    Q: Why do you think I asked you to read the End-to-End paper?

    Q: Should we ever perform operations at the low level that will need
       to be reperformed at a high level?  If so, when?

    Q: What does it tell us about:
	- blast versus stop-and-wait bulk data transfers?
	- hardware acknowledgements?
	- packet checksumming?
	- maintaining the coherence of distributed data (two-phase commit)?
	- how to split the functionality of a clustered web server?
	- when/how to encrypt secure data?

    Discuss tradeoffs and alternatives!

----------------------------------------------------------------------------

Topic: Aside

    CSL Seminar:
    - Lots of good material for people in here (some overlap)
    - This Wednesday's is particularly good (concurrency control)

----------------------------------------------------------------------------

Topic: Student Lectures

    Updated schedule of lectures:
    -----------------------------
	- In : distributed objects, security
	- Out: WWW issues, wide area file systems

    Need volunteers for different topics:
    -------------------------------------
	- papers to be filled in later
	- may request help doing literature survey for best papers
	  for the ones in the new topics
	- sign up ASAP (first-come, first-served)

----------------------------------------------------------------------------

Topic: Projects

    Let's brainstorm on possible project ideas...

    Here are some I have:
    ---------------------
    * Distributed object layer for GMS
    * Chat/whiteboard service: in GMS and on "bare system"
    * Clustered client proxy cache
    * Worm-like service (pick one - service itself not important)
	- self-replicating/healing
	- client `failover'

    What are you doing in other classes/projects that is related?
	- I have demo envy, so graphical projects get fuzzy points!

----------------------------------------------------------------------------

Topic: Optional overview of networking

    Types of networks:
	- broadcast (mostly gone in wired networks now)
	- switched (pretty much taken over the LAN market)
	- point-to-point (dominant in the WAN infrastructure)

    Components:
	- switches
	- bridges
	- routers

    Common LANS:
	- Ethernet
	- Ethernet
	- Ethernet

    Commons single-site backbones:
	- ATM
	- FDDI

    Common failure modes:
	- adapter failures (normally failstop, sometimes byzantine)
	- switch failures
	- router failures
	- SOFTWARE FAILURES (router tables, DNS entries, etc.)