SYS364 - Seneca - School of Information & Communications
Download
Report
Transcript SYS364 - Seneca - School of Information & Communications
SYS364
Architecture
Today’s Agenda
System Architecture
Processing Methods
On-line, Batch
Centralized, De-centralized
Networking
LAN/WAN/MAN
Client/Server, File/Server
System Architecture
Refers to the Logical design and
physical structure of a system
Includes hardware, software, design and
processing methods
Environment & Platform
Server Farm with UNIX(s) + Apache
AS/400 with Websphere + NT partitions
Processing Methods
Environment and/or Platform refer to a
specific combination of hardware and
software
Influences system architecture
Reinforces bias (when all you’ve got is a
hammer, everything looks like a nail)
Becomes religious (but there is no “one
true” platform)
…just a few years ago
Two predominant environments
Mainframes with terminal access
Stand alone PC’s running one program at a
time
Today
The Internet connects everything
Businesses are more dynamic than ever
Competitive pressure: e-commerce, 24/7
Corporate “right sizing”
Mergers
Employees at all levels satisfy their own
information needs – IT provides users with
the technology, tools and resources
Requires that systems be cost effective, flexible
and user-friendly
Impact on Analyst’s strategy
Business operational requirements
Match software to business process
Level of integration with outside systems
EDI, XML, information/transaction exchange
Level of access for customers and
suppliers into your platform
Internet, telephony, call centers
On-line Data Processing
Transactions are processed completely when
and where they occur
Users interact directly with the information
system programs (not forms)
Programs update DB directly & immediately
(input verification is critical to system integrity)
system availability is crucial
Consider cost of downtime, manual backup
procedures
Batch Data Processing
Data is collected and processed in
batches, logically and physically.
Data captured, then entered,
then input.
May be more efficient than on-line
Typically used for large amounts of data
that must be processed on a routine
schedule
What in case study is batch?
Batch Data Processing
Transactions are collected and
processed periodically
Computer operators can run batch
programs on a pre-determined schedule
without user involvement
Payroll, Accounting close off
Data Warehouse, EIS, DSS updates
EDI
Level of Automation
Volume: How much info is there?
Volatility: How often does info change?
Timeliness: When does matter?
Forms and batch processing
PDA & handheld data collection devices
With wireless modems
Online screens everywhere
Centralized Systems
Server or Host
Mainframe and terminals
e.g. IBM “big iron”, AS/400 and large UNIX
In the past, slow + $$$ communications
networks limited remote data entry/input
Clustered systems
Many identical systems with load
balancing, fail-over. e.g. Server farms.
A horror when it’s time to upgrade/update.
Heterogeneous environments
Variety of platforms, use middleware
Distributed Systems
For business, computing is distributed
geographically, connected by networks
Local and regional tasks are performed
by decentralized computers
Summary uploads to central office
May use Client/Server model
Distributed Systems Design
processing involves local and central data
Data may be stored closer to users and
reduce network traffic
Easily scalable – new storage sites can be
added without affecting network operation
Less likely to experience a catastrophic failure
…what about security?
…network management?
Database replication and synchronization
issues
Distributed Processing Today
Distributed CPUs
SETI@Home
DistributedScience
Entropia
Distributed Storage (Peer-to-Peer)
Napster
Gnutella
Networks and the Net
The value of a network is the square of
the number of things connected to it. —
Metcalfe's law for Bob Metcalfe, the
inventor of Ethernet and 3COM.
The value of the Net is the number of
groups people on the network can form.
Reid’s Law
Moore's Law=
processing
power doubles
every 12-18
months
n = people
2
3
4
5
6
10
100
1,000
10,000
100,000
1,000,000
10,000,000
100,000,000
1,000,000,000
Metcalfe's Law = the value
of n items in a network is
proportional to n^2. I.e. the
number of one-to-one
connections.
n(2)-n = connections
2
6
12
20
30
90
9,900
999,000
99,990,000
9,999,900,000
999,999,000,000
99,999,990,000,000
9,999,999,900,000,000
999,999,999,000,000,000
Reed's Law: Net's value comes
from its enabling of groups, I.e.
the number of distinct subgroups
that can form in a set of N
members is 2^N
2(n)-n-1 = groups
1
4
11
26
57
1,013
1,267,650,600,228,230,000,000,000,000,000
1.072E+301
#NUM!
#NUM!
#NUM!
#NUM!
#NUM!
#NUM!
The Net
Intranet, extranet, internet
Accessibility – browsers make presentation
layer independent of the platform’s
software/hardware/operating systems
Low cost communications
Widely implement standards (HTML, XML)
Security slowly getting better
Reliability – no guarantees yet
Throughput – last mile problem
Volatile standards and tools
Local Area Networks (LANs)/
Wide Area Networks (WANs)
LANs
Hardwire a group of PCs together in a limited area
to a server
WANs
Uses a combination of phone lines, fiber optic
cables, microwave transmission and satellites to
connect PCs and LANs into a network that can
span the globe
X25, Frame Relay, ATM (Asynch Xfer Mode)
Internet built on TCP/IP
MANs (Metropolitan Area Networks)
New Ethernet wireless standard
Network Issues
Software licensing
Traffic and Bandwidth
Security and Firewalls
Topology, Configuration, Protocol
These issues need an expert
(I.e. your friends in CNS program)
Client/Server Systems
Divides processing between one or more
clients and a central server
A client handles the user interface
Data entry
Editing
Data query
A server handles
Data access
Processing
Database management functions
Benefits of Client/Server
? Easily scalable (up or down) to meet
changing business needs ?
? Transfer of expensive mainframe
systems to less expensive clients ?
In practice, benefits are dubious
and many C/S projects failed
Is a browser on the Web closer to
Client/Server or Mainframe/Terminal?
Systems Design Completion
System Design Specification
Created before the next SDLC Phase,
Implementation
Detailed costs
Staffing
Scheduling
Oriented to Programmers so they know
what to do
Oriented to Users so they know what they
are getting
Approvals of the Systems
Design Specification
Users approve all report and output
screen designs, menu and data entry
screens, source documents
IS Management will approve staffing,
costs, hardware and software and
effects on the operating environment
Other management – does the new
system meet business objectives?
Technical and Management
Presentations
Programmers/operators/IS staff
IS Management
Company management
Possible outcomes
Proceed with development
Perform additional work on the design phase
Terminate the project
Summary
Architecture
Processing
Networking
Completing the Design Phase of the
SDLC