Technical Overview Of DCOM

1
Technical Overview Of DCOM

• By
• Varada venkata

2
Contents

• Introduction
• Why distributed applications?
• DCOM architecture
• Design issues addressed by DCOM
• Summary

3
Introduction

• DCOM extends COM
• Supports communications among objects on
  different computers
• Applications can be distributed at locations
  that make sense to the customer and to the
  application
• DCOM handles low-level details of network
  protocols allowing us to focus on the business
  aspects
• DCOM solves the problems associated with
  distributed applications

4
Why to write distributed applications ?

• Some applications are inherently distributed.
• Applications designed with distribution in mind
  can accommodate different clients with different
  capabilities.
• Designing applications for distribution provides
  flexibility in deployment

5
DCOM Architecture

Com component in the same process
6
DCOM Architecture

• Com runtime Provides object oriented services
  to the clients and components
• RPC and security provider Generate standard
  network packets that conform to the DCOM
  wire-protocol standard

7
Design issues in DCOM

• COMPONENTS AND REUSE

• DCOM directly and transparently takes advantage
  of any existing in COM components and tools

LOCATION INDEPENDENCE

• Design constraints while implementing Distributed
  Applications
• Components that interact should be close to each
  other
• Some components can run on specific machines or
  at specific loc
• Smaller comps increase flexibility of
  deployment,inc ntwk traffic
• Larger comp reduce ntwk traffic, reduce
  flexibility of deployment

8
Location independence

• DCOM completely hides the location of the
  component
• If applications have numerous small components ,
  reduce ntwk loading by deploying them on same
  LAN segment, or on same machine
• If application have small number of large
  components , put them on the faster machines
  available , less scope of network loading

9
Language neutrality

• As an extension of com ,DCOM is completely
  language- independent
• Developers can choose the tools and languages
  that are familiar with
• CONNECTION MANAGEMENT
• DCOM manages connection to components by
  maintaining a reference count
• DCOM uses efficient pinging protocol
• DCOM provides a robust distributed garbage
  collection mechanism that is transparent to the
  application
• DCOM provides unidirectional client-server
  interaction and also rich , interactive
  communication between clients and servers

10
SCALABILITY

• DCOM provides a number of features that enhance
  appls scalability
• SMP(symmetric multiprocessing)
• DCOM takes advantage of the Win NT support for
  multiprocessing.
• DCOM shields the details of thread management and
  delivers the optimal performance.
• FLEXIBLE DEPLOYMENT DCOM location
  independence enables it to distribute components
  over other computers
• Ex?
• Efficient performance for a server
  is to handle 50 components.
• No of users 200,
• when no of users 400 , no of components 70,
• Deploy 30 components on second server, 20 are on
  both, 20 on first.

11
SCALABILITY
12
Versioning

• Com component can appear differently to different
  clients .
• A client that uses a certain feature needs to
  access only the methods and properties that it
  uses.

13
The broker component keeps track of the pool of
the server machine running identical components
The broker measure the current load on the
server and detects when servers are added or
removed.
14
Performance

• Vtable
• RPC infrastructure when client and components are
  not actually close.
• The overall performance and scalability
  advantages of DCOM can be reached by implementing
  thread-pool managers and pinging protocols
• BANDWIDTH AND LATENCY
• Bandwidth The size of the parameters passed
  into a method call directly affects the time it
  takes to complete the call
• Latency The physical distance and the number
  of network elements involved delay even the
  smallest data packet .
• DCOM minimizes the network round trips to avoid
  impact of latency.
• DCOM preferred transfer protocol is the
  connection less UDP subset of the TCP/IP protocol
  suite .
• This connectionless nature of this protocol
  allows DCOM to perform several optimizations by
  merging many packages with data and pinging msgs

15
Security
DCOM stores Access control list for components
16
Programmatic control over security
17
Load Balancing

• Static load balancing
• Permanently assign certain users to certain
  servers running the same application.
• Dynamic load balancing
• Reference component decides which server has
  less load. It chooses the server having less load
  and deploys the component.

18
Protocol Neutrality

• DCOM makes sure that the application is
  independent of the underlying network infra
  structure.
• DCOM uses any transport protocol, including
  TCP/IP ,UDP.
• DCOM provides security framework on all of these
  protocols , including connectionless and
  connection-oriented protocol.

19
Summary

• DCOM makes it easy to write a distributed
  application that
• scales from single computer environment in to
  biggest pool of server machines
• Can be robustly expanded to meet new functional
  requirements
• Uses network bandwidth carefully, while providing
  great response time for the end users
• Is inherently secure
• Provides a smooth migration path to sophisticated
  load balancing
• Can be used with any network protocol and
  integrated into any hardware platform

20
References

• "DCOM Architecture"
• "DCOM Solutions in Action" (MSDN Library,
  Conference Papers, PDC 97 Conference Papers)
• The Component Object Model Specification (MSDN
  Library, Specifications)
• Distributed Component Object Model Protocol
  DCOM/1.0 (MSDN Library, Specifications)
• COM Technologies Home Page at http//www.microsoft
  .com/com/comintro.htm
• Introduction to DCOM (http//devcentral.iftech.co
  m/articles/DCOM/intro_DCOM/part1/default.php_