1. What are the three basic ways to describe a distributed system?
Distributed system models help in ...
... classify and understand different implementations
... identify their weaknesses and strengths
... Creating a new system outside the previously validated building blocks
2. What are the difficulties and threats for distributed systems?
· Various usage models
o High workload variations, broken parts, or bad connections.
· Various system environments
o Heterogeneous hardware, operating systems, networks, and performance.
· Internal problems
o Hours that are not synchronized, conflicting updates, various hardware and software failures.
· External threats
o Attacks on data integrity, confidentiality, and denial of service.
3. Give an overview of the physical model
· Representation of the h / w element underlying DS that summarizes the specific details of computer / network technology.
· Basic physical model
· 3rd Generation DS:
o Early distributed systems [late 70-80s]: LAN-based
o Distributed internet scale systems: Cluster, grid, P2P, Clouds
o Contemporary distributed systems: dynamic nodes such as cellular-based services (very dynamic nodes are not static like other models).
4. Overview of architectural models - architectural elements
· The architectural model of a distributed system is related to the placement of its parts and the relationships between them. Example:
o Client-Server (CS) and peer process models.
o CS can be modified by:
§ Data partition / replication on cooperative servers
§ Caching data by a proxy server or client
§ Use of cellular codes and cellular agents
§ Requirements for adding or removing mobile devices.
5. Overview of architectural models - architectural patterns
Architectural Model - simplifies and abstracts the functions of individual components:
· Initial simplification is achieved by classifying the process as:
o Server process
o Client process
o The peer process
§ Work together and communicate symmetrically to perform tasks.
6. Determine the platform for distributed systems
· The lowest layer of hardware and software is often referred to as a platform for distributed systems and applications.
· This low-level layer provides services to the layer above, which is implemented independently on each computer.
· The main example
o Intel x86 / Windows
o Intel x86 / Linux
o Intel x86 / Solaris
o SPARC / SunOS
o PowerPC / MacOS
7. Determine distributed system middleware
· Software layers whose purpose is to cover the heterogeneity that exists in distributed systems and to provide a convenient programming model for application developers.
· Main example:
o Sun RPC (Remote Procedure Call)
o OMG CORBA (Architecture of Common Object Request Brokers)
o Microsoft D-COM (Distributed Component Object Model)
o Sun Java RMI (Remote Prayer Method)
o Examples of Modern Middleware:
Manjrasoft Various - for Cloud computing
§ IBM WebSphere
§ Microsoft .NET
§ Sun J2EE
§ Google AppEngine
§ Microsoft Azure
8. Describe the two-tier and three-tier architecture
· Two-tier architecture
This is a typical interaction for single threaded, monolithic clients. A client requests some processing or information from the server that he needs. He waits blockade for replies containing results, then can continue with the execution.
· Three-tier architecture
Services can be implemented as several server processes on separate host computers.
Example: Server and web Cluster based applications such as Google, Oracle parallel database
This topology is very common. Websites like Google serve around 100 million searches a day.
Obviously not feasible to serve them from one server.
Google uses clusters containing 10 of the thousands of machines that offer equivalent services, and you are transferred (through DNS and other ways) to one of them. Can also be transferred at the protocol or application level.
A similar technique can be used for Oracle databases, which are replicated through many servers to offer redundancy and performance
No comments:
Post a Comment