Architectural styles

Layered architectures System architecture Multitiered Architectures Hybrid Architectures

components are organized in a layered Deciding on software components, their

fashion where a component at layer Lj can interaction, and their placement leads to an
instance of a software architecture, also Two-tiered architecture Three-tiered architecture Edge-server systems Collaborative distributed systems
make a downcall to a component at a
lower-level layer Li (with i < j) and generally known as a system architecture.
expects a response. Only in exceptional
cases will an upcall be made to a organized according to a hybrid architecture
higher-level component. is formed by edge-server systems. These
systems are deployed on the Internet where
servers are placed ?at the edge? of the
network. This edge is formed by the
Many distributed applications are divided into three layers:
Centralized organizations Decentralized organizations boundary between enterprise networks and
the actual Internet, as provided by an
Internet Service Provider (ISP).

Simple client-server architecture Peer-to-peer systems

User interface layer Processing layer Data layer
a part that handles a middle part that a part that operates
interaction with a user generally contains the on a database or file
or some external core functionality of system.
General interaction between a
application. the application.
client and a server.
Structured peer-to-peer systems Unstructured peer-to-peer systems

in a structured peer-to-peer system the in an unstructured peer-to-peer system

nodes (i.e., processes) are organized in an each node maintains an ad hoc list of
overlay that adheres to a specific, neighbors. The resulting overlay resembles
deterministic topology: a ring, a binary tree, what is known as a random graph: a graph
a grid, etc. This topology is used to in which an edge ?u, v? between two nodes
efficiently look up data. Characteristic for u and v exists only with a certain probability
structured peer-to-peer systems, is that P[?u, v?].
they are generally based on using a
A server A client so-called semantic-free index.

A server is a process A client is a process key(data item) = hash(data item?s value) Search Methods in the Unstructured peer-to-peer systems:
implementing a that requests a
specific service. service from a server
by sending it a
request and Flooding Random walks policy-based search
This client-server interaction, subsequently waiting
also known as request-reply for the server?s reply. Flooding is a search or Random walks are a search method in distributed
the policy-based search
behavior. communication method in systems where an issuing node selects a
distributed systems where an involves selecting which
neighbor at random and sends a request for a
peers to query for a
issuing node sends a request to data item to that neighbor. If the neighbor has the
data item based on
all of its neighbors, and each data, it can respond directly to the issuing node. If
neighbor that receives the previous positive
not, the process is repeated with another
responses and giving
request forwards it on to all of its randomly chosen neighbor, continuing until the
preference to those
own neighbors, until the request data is found or a predefined number of steps is
reaches the node that has the neighbors that have
reached. The process of selecting random
many connections
requested data or all nodes have neighbors and forwarding requests is known as a
themselves.
been searched. random walk.