Transcript SoftLabFinalPres - Networked Software Systems Laboratory

Network Monitor
Final presentation
Project in systems programming,
Winter 2008
Students : Vitaly Zakharenko & Alex Tikh
Supervisor : Isask'har (Zigi) Walter
Software Systems Lab
Department of Electrical Engineering
Technion - Israel Institute of Technology
Introduction
Modern corporate networks often
contain large number of local networks,
which need to be monitored from a single
location. The latter is often difficult due to
security and accessibility restrictions
between distinct LANs.
Project Goal
Monitor large corporate networks
without passwords and special
permissions
Project Description

The system consists of multiple lightweight
agents and a single server application.
Monitor App.
Agent App.
Basic Server functionality
collects and stores measurement results sent by
agents and provides statistics.
 maintains database records of agents, monitored
hosts and host state measurements performed by
agents.
 remotely configures the functionality of agents.
 monitors state of the agents.

Basic Agent functionality
performs network measurements and
communicates results to the server.
 allows the server to get & set settings
defining the measurement functionality
and the scheduling scheme of the
measurements.

Server Details
Uses local SQL Server DB file to store
all information. Does not use any local
records to avoid DB synchronization
problems.
 Addition of new measurements is easy.
 Classes of a specific measurement on the
server side communicate only with
classes of the same measurement on the
agent side. Thus measurements are
independent.

Server Details (cont.)
The application framework is independent
of classes providing a specific
measurement. Nothing needs to be
changed in the framework when new
measurements are introduced.
 Implementation is fully multithreaded. No
blocking ever occurs.
 Server–agents communication is based on
.NET Remoting.

Server Details (cont.)
Intensively uses OOD methodology and
Design Patterns (Factory, Mediator,
Publisher-Subscriber).
 Server is implemented as a console
application.

Agent Details
Build with special care to facilitate
addition of new measurements.
 Each measurement be configured
remotely.
 Each measurement’s scheduling scheme
can be configured by the remote server.
 Agent remembers its configuration
(settings are serialized).

Agent Details (cont.)
All measurements are independent of
each other and each one runs on a
separate thread.
 Design is object-oriented and decoupled.

Implemented example
measurements

SNMP device detection
- agent broadcasts SNMP packets to IP
addresses within its subnet and reports
detected SNMP devices to the server

ICMP (Ping)
- agent broadcasts ICMP packets to IP
addresses within its subnet and reports
dead and live hosts, loss ratio etc.
Implemented example
measurements (cont.)

HTTP URL response time
- agent tries to download a web page
from a number of URLs specified by the
server, measures access time (or failure)
and reports it to the server.
Technology
C# and .NET Framework
 .NET Remoting and Sockets
 MS SQL Server
 ADO.NET and strongly-typed DataSet

Other topics used in project
Protocols SNMP, ICMP, HTTP.
 Sockets – previous versions of the
applications were based on sockets.
 C# TCP programming using Sockets and
Helper classes – in prev. versions. We
learned to deal with TCP buffering
problems.
 Serialization, formatters, streams etc.
 Using registry, WMI, DNS etc.

Literature used in project
C# for Java Programmers(642 pages)
– studied througly.
 C# Database Programming(~400 pages)
– studied througly.
 C# Network Programming(647 pages)
– studied throughly.

Detailed Server functionality
Managing Agents




The whole space of IP Addresses is divided
in sets. The sets are optionally overlapping.
To each set distinct functionality of Agents
within the set is ascribed.
The IP Address set together with settings of
Agents within the set are incorporated into
a “policy”.
User can change both the set of IP
Addresses covered by a policy and the agent
settings ascribed to the policy.
Detailed Server functionality
Managing Agents (cont.)
The policies are stored in database.
 Each change of a policy (IP Address
mapping or agent settings) is
communicated to the relevant known
agents.
 Only relevant agents are updated.
 Whenever a new (yet unknown) agent
communicates the server its settings are
updated according to the highest priority
policy it falls within.

Detailed Server functionality
Managing Agents (cont.)
Settings of each measurement are
managed by a distinct class on the server
side.
 The class both manages the database
table, where the settings are stored and
communicates the setting to the two
classes responsible for performing the
measurement on the agent side.

Detailed Server functionality
Managing Results



Results are stored in a local database.
Each measurement result is communicated by
agents to a distinct class responsible for storing
results of the specific measurement, providing
measurement statistics and interfacing the
measurement result record.
The measurement manager class notifies the
mediator class of each new measurement result
provided by an agent. As a result whenever a new
(yet unknown) agent communicates results to the
server it is registered in the database by the agent
manager class, subsequently communicated by the
agent.
Server: Class Diagram
Agents management
Server: Class Diagram
Hosts management
Server: Class Diagram
Main class
Server: Class Diagram
Factories
Server: Class Diagram
Measurement result classes
Server: Class Diagram
Measurement management classes
Server: Class Diagram
Setting management classes
Server: Class Diagram
Setting classes
Server: Class Diagram
Mediator class
Server: Class Diagram
Policy management classes
Server: Class Diagram
Setting set classes
Agent: Class Diagram
Main class and aux. class
Agent: Class Diagram
Factory classes
Agent: Class Diagram
Worker classes
Agent: Class Diagram
Scheduler classes
Server: Database Diagrams
Policy diagram
Server: Database Diagrams
Results diagram
Server: UML page 1
Server: UML page 2
Agent : UML
Summary and conclusions

By planting an Agent in a LAN one can
efficiently raise restrictions imposed by
LAN’s firewall. Thus, one can effectively
monitor computers as if they all belonged
to the same LAN.
Summary and conclusions (cont.)
C# is perfectly suited for programming
tasks addressed in the application.
 The application profited greatly from
OOD support of C#, from powerful
dedicated class libraries, easy DB access
(via ADO.NET) and modern Remoting
mechanisms.
 C# is also widely used and perfectly
compatible with many other technologies
provided by Microsoft.

Summary and conclusions (cont.)
Storing information to MS SQL server
database file instead of a regular data file
makes the data structure more orderly
and simplifies synchronization issues of a
multithreaded application.
 In short, we
Microsoft .NET.

Future work
The program is implemented as a console
application. GUI should be developed.
 The DB file may grow too large when
many network measurement results are
accumulated over time. The results
should therefore be periodically
summarized in form of statistics and saved
in Hosts DB table.

End
Great thanks to our supervisor, Isask'har (Zigi)
Walter for his great help and support
 Thanks to the software lab staff for the support
(Ilana David and Victor Kulik)
