Differences between pre-emptive
& co-operative multitasking?
Monday, January 15, 2001 at 19:48:00
With pre-emptive multitasking the operating system
defines the amount of time that a process can stay on the processor,
whereas co-operative multitasking relies on a process yielding
itself from the processor before another process can run on it.
Cooperative multitasking leads to processor hogging, and in the
worst case can lead to deadlock while a process is waiting for
a resource which cannot be released while the process is waiting
for it (typically one of the waiting processes has locked it).
Pre-emptive multitasking leads to smoother running systems, as
each process gets some time allocation on the processor.
Describe H/W interrupt & S/W
Monday, January 15, 2001 at 19:48:00
A hardware interrupt is generated when a device
actives a hardware interrupt line (typically an IRQ line), which
is then processed by an IRQ (Interrupt Service Routine). A typical
example might be when a character has arrived on the serial port.
The serial device causes a hardware interrupt (typically on a
PC, this is IRQ3 or IRQ4) which then causes the currently running
program to interrupt and the system then calls the required IRQ
which reads the character into a memory buffer, and then returns
back to the originally running program. A software interrupt is
generated by a program, such as an interrupt to determine the
current time, or reading a character from an external device.
Out line how RPC fits in to the
Wednesday, January 17, 2001 at 19:26:25
RPC (Remote Procedure Call) operates an the session
layer of the OSI model, and uses TCP/IP or UDP/IP for its transport
and network layer connection. It supports the running of remote
processes by passing run parameters to a remote systems and then
waiting for the results. RPC is a bit like logging into a remote
system (as you would with the session layer with telnet or rlogin)
and running a process and then getting the results back from it.
The process, itself, runs on the remote computer, and not on the
host computer. RPC is thus one of the best protocols to create
distributed systems, and processes are run of remote systems,
thus relieving the loading on the host system.
Have we got to know the Windows
NT and UNIX parts for the exam ? In your latest mail, are
the subjects you noticed the only ones for the exam or have
we got the whole part of NOS module ?
Friday, January 19, 2001 at 23:08:48
The areas I have outlined are the main areas that
you should read, but a basic understanding of NT and UNIX would
be helpful, especially related to some of the file permission
properties It might be a good idea to read the material on NDS
For the questions of the chapter
16 called "routine protocols", please tell me the answer
of the question "How is routing packet defined?" and "What
causes count-to-infinity?". Thanks Bill
Friday, January 19, 2001 at 20:00:48
The IP protocol allows for different types of packet
to be carried in the IP data packet. For example if the protocol
field value is a 1 it is ICMP (for ping, traceroute, and so on),
or 17 for UDP, and so on. Other values define routing packets
(such as RIP, IGRP, and so on). These IP packets contain routing
information, typically in the form of routing tables.
Count-to-infinity occurs with occurs with distance-vector
routing. With this there is a loop. Let's say that A, connects
to B, connect to C and back to A. If a Network connected to A
goes down then it tells B that it is unreachable, unfortunately
C tells B that it can reach still reach it in one hop, so B tells
that to A (which now makes it two hops), next A tells C than it
can reach it in two hops (so C thinks it can reach it in three
hops), and so it goes on until infinity occurs, and the data packets
will obviously just keep going round in a loop. The ways to get
round this problem are:
Setting infinity values. This stops a count to
infinity and limits it to a given value. In RIP this is
16, thus the maximum hop count is 15.
Split horizon. This is where routers do not updates
their tables when they know they are currently connected
to a given network which is currently down.
Hold down timers. Force routers to not take any
updates on a route that they know is currently down, for
a given amount of time, in order for the route to recover.
Why when using tracert to trace
a node from one subnet to another, it does only displays
(exit subnet gateway) not when it enters the other subnet
Friday, January 19, 2001 at 19:56:36
A router will have more than one port, each of which
will have an IP address which is related to the subnet that it
connects to. For example if a router has three ports which connect
to the subnets: 220.127.116.11, 18.104.22.168 and 22.214.171.124,
then the three ports of the router can be assigned the addresses:
126.96.36.199, 188.8.131.52 and 184.108.40.206. When tracing from
the 220.127.116.11 subnet it will go through the 18.104.22.168
port (which is the gateway for the 22.214.171.124 subnet), and
go out the 126.96.36.199 port onto the 188.8.131.52 subnet, but
this port will not be seen in the traceroute.
How would you describe Unix as
an operating system, one that competes with Windows or one
that works alongside it. What in your mind is Unix's biggest
advantage over Windows?
Sunday, January 21, 2001 at 14:16:18
UNIX is an excellent operating system, which is
freely avialable from many sources, and has been defined as an
open-system (as apposed to Windows NT which is a closed system).
The advantages of UNIX over NT are reducing as Windows becomes
more like UNIX, but here are a few of its current advantages:
Direct support for standard protocols. It directly
supports many of the standard protocols used on the Internet,
such as TCP/IP for communications, RPC for remote process
control, NFS for distributed file systems, and so on. With
Windows NT, UNIX is now easily integrated with any network,
as Windows NT directly supports NFS for mounting remote
Kernel-based. UNIX has the advantage over Windows
in that it has a clearly defined kernal, which can be tailored
to the required application (and is thus useful in embedded
applications). It is often difficult in Windows to differentiate
the kernal from the graphical user interface.
Easy-to-use attributes and command based. UNIX also
provides a set of easily modifiable set of file attributes
for the user (rwx),
their group (rwx)
and the rest of the world (rwx).
For users who understand the concepts of file attributes
these are easy-to-use, and to set up. Many users also prefer
to use text commands rather than windows-based programs
to control file attributes, thus chmod is a useful command
for changing the attributes of a file.
Less expensive in server applications. UNIX, as
a standard, supports most of the commonly avialable server
applications, such as e-mail, domain name services, telnet,
ftp, and so on, whereas Windows NT Server is required in
order to support server applications, which is often expensive.
Thus individual users can easily run server applications
(especially using LINUX). Windows NT does not directly support
Telnet, which can make remote connection difficult (although
there may be some support for it in Windows 2000 Server).
To reduce the reliance on any one vendor (typically
Microsoft), a well constructed network operating system for a
medium- to large-scale network is likely to be based around a
mixed UNIX/NT(/2000) provision.
What are the basic file attributes
that are used in NetWare?
Monday, January 22, 2001 at 12:16:18
Files have the SRWECMFA
attributes, which are:
Grants all rights to the folder or file. The Supervisor
right can't be blocked by an Inherited Rights Filter. Users
with this right can grant or deny other users rights to
the folder or file.
For a folder, grants the right to open files in the folder
and read the contents or run the programs. For a file, grants
the right to open and read the file.
For a folder, grants the right to open and change the contents
of files in the folder. For a file, grants the right to
open and write to the file.
For a folder, grants the right to create new files and
folders in the folder. For a file, grants the right to create
a file and to salvage a file after it has been deleted.
Grants the right to delete the folder or file.
Grants the right to change the attributes or name of the
folder or file, but does not grant the right to change its
contents. (Changing the contents requires the Write right.)
Grants the right to see the folder or file with the DIR
or NDIR command.
Grants the right to change the trustee assignments and
the Inherited Rights Filter of the folder or file.
Note that file rights apply only to the file they
are assigned to. The rights can be inherited from the folder that
contains the file. Folder rights apply not only to the folder
but also to the files and folders it contains.
NDS (written by Novell)?
on NDS (Network Computing Journal)
v. Windows 2000
with Windows 2000
How does an exterior routing
protocol differ from an interior routing protocol ?
Monday, January 22, 2001 at 21:28:07
An interior routing protocol is used by routers
within an Autonomous System (ASs - which are assigned by a unique
number) to route data packets around the AS. If data requires
to go out of the AS, it is sent to the gateway. An exterior routing
protocol is then used to route between the ASs. This has several
Routing information is not transmitted outside the AS.
Many organisations, especially military and defence-related,
do not with information on their networks to be transmitted
to routers outside their organisation, thus an interior
routing protocol will keep the routing data within the AS.
Simplifies the Internet. It would be almost impossible
for routers to built up routing tables which contained all
the routers in the whole of the Internet, thus ASs allow
for a simplified design, where routers only have to know
how to route within their own domain. The external routing
protocol will route between these domain. SuperJANET is
a good example of an AS, where all the incoming and outgoing
data goes through a few main gateways.
Optimised for condition. Typically an interior routing
protocol is based on a metric such as bandwidth, number
of hops, and so on, whereas an exterior routing protocol
would just use connectivity as its main objective (it really
just wants to route from one domain to another).
A typical exterior routing protocols are EGP, and typical interior
routing protocols are RIP and IGRP.
Processes and Scheduling (Chapter 5)
What are the basic states of process when it is
in a queue?
Outline the three main types of scheduling queues,
and where would they most likely be used.
Describe the basic objectives of a scheduling algorithm,
and how the schedular will try to deal with them (that is, maximise
Discuss the main scheduling methods that are used,
and what advantages and disadvantages do they have?
How are signals used in UNIX?
What is the main difference between signals and
What is the major problem with signals?
What is the main difference between hard real-time
and soft real-time?
Outline the operation of pipes, and how they are
implemented in UNIX.
Discuss the operation of message handling.
What commands does UNIX use to control processes?
Distributed Processing (Chapter 6)
What is the main advantage of distributed processing?
What methods are used with interprocess communications
Discuss a simple method of how a semaphore might
be used on mutually exclusive code.
Outline for the code given in the notes, describe
the operation of the producer-consumer problem.
What conditions occur for deadlock to occur?
How does the Banker's Algorithm overcome deadlock?
Give a simple example of the Banker's Algorithm.
How might a system recover from deadlock?
Discuss the basic structure of RPC.
How does RPC fit into the OSI model?
Outline the operation of RPC?
Distibuted File Systems (Chapter 7)
Outline some advantages of a distributed file system.
What are the two methods used to mount drives on
a system (that is, forest of drives or a single tree), and what
advantages and disadvantages of both?
What method is used to allow data to be represented
in a standard form?
Discuss how NFS creates a networked file system.
How do the NFS protocols fit into the OSI model?
Explain the operation of NIS, and what the important
files that it keeps a record of?
Outline the requirements for a primary and secondary
NIS servers, and how do they operate.
Routing Protocols (Chapter 16)
What the two main functions of the network layer?
How does a routed protocol (such as IP) differ from
a routing protocol (such as RIP)?
How does static routing differ from dynamic routing?
What routing metrics might a routing protocol use?
What methods do routing protocols use to transmit
information about their routing information?
Explain the difference between link-state and distance-vector
Outline the problems with distance-vector protocols,
and how they could be overcome.
What problem occur with link-state protocols?
How does an exterior routing protocol differ from
an interior routing protocol, and how are ASs used between them?
NetWare, NT and UNIX (Chapter 22)
Outline the main networking protocols used in NOS's.
State a set of best practices in the administration
of an NOS.
State a set of best practices for user accounts.
State a set of best practices for high security
Discuss the methods uses to create a robust networked
data storage environment.
Which file systems are used in Microsoft Windows?
Discuss the main problems with Novell NetWare 3.1,
and how they are ovecome with NetWare 4.1/NDS.
What enhanced features does NetWare 4.1/NDS have?
How is NDS structured, and give an example of an
How does a user change their current context?
How is a volume mapped in NDS?
How does NDS use time servers?
How are events timestamped in NDS?
What are the security models uses by FAT, UNIX and
For the NOS project, is there
a TCP port number which corresponds to the NIS servcie ?
Friday, January 26, 2001 at 15:33:55
NIS, like NFS, is an application layer protocol (as shown below),
and uses RPC for its session layer protocol. Thus an NIS server
must respond to RPC (which is port 111, I think), but the real
port will be defined in the /etc/services
directory (and the /etc/rpc
file will contain the rpc processes, along with the /etc/protocols
file which defines network layer protocols). The presentation
layer is XDR so that the data can be transferred transparently
Ref: Figure 7.3 Distributed Systems and
Networks, W.Buchanan, McGraw-Hill.