IPCC Server Services

© 2002, Cisco Systems, Inc. All rights reserved.
Functional Description
© 2002, Cisco Systems, Inc. All rights reserved. 2

3
ACD1 PG1 AW2 PG2a PG2b ACD2 PG3a PG3b vru2 vru4
IP IP CMS IP vru1 vru3
2
IP
CallRouterA LoggerA
IXC Network NIC
© 2002, Cisco Systems, Inc. All rights reserved. ICMSA 5.0—Lesson 2-3
4
5 IP
CallRouterA LoggerA
IXC Network NIC
7
8
9
6
IP
11
10
CallRouterA LoggerA
1
3 4
2
IXC Network NIC
5
Router Processes
Prefix Process Name Description

Ccag CCAGENT Central Controller DMP Agent. Device Management
Protocol Agent that manages session layer
communications with ICR nodes.
Dbag DBAGENT Central Controller Database Agent. Communications
process that validates access to the central database.
Dbw DBWORKER Host Database Lookup. Process that queries external
databases and uses resulting data in call routing.
Mds MDSPROC Message Delivery Service. Process that provides
reliable message delivery between ICR processes.
Nm NODEMAN Node Manager. Process that manages, restarts, and
initializes processes on each ICR Node
Nmm NMM Node Manager Manager. Process that manages,
restarts, and initializes the Node Manager process on
each ICR Node.
Rtr ROUTER Call Router. Process that receives call routing requests,
determines call destinations, and collects information
about the entire system.
Mcia, mcib MCINIC MCI Network Interface Controller process. Interface
between the CallRouter and the MCI signaling
network. Others would be Sprint NIC; INAP NIC, etc.
CIC Netwrkcic Network ICR process that connects to Customer ICR
ICRPNIC process. Passes the route request to another
ICR system
AGI Appgw Application Gateway. Allows a routing script to pass
data to an external application and receive data in
return which can be used in routing decisions.
Rts RTSERVER Real Time Server. Process that takes real-time data
retrieved from PGs and forwards it to Admin
Workstations
Logger Processes

Csfs CSFS Customer Service Forwarding Service. Handles
communications with various services like SNMP,
RAS, and Serial Event Feed.
Dtp DTP Customer Support Data Transfer Process (Dials The
Phone). Transfers events from the Logger to the
GeoTel Customer Support Center (CSC)
Lgr LOGGER Database Logger. Process that stores historical data and
information about the entire system in the central
database.
Nm NODEMAN Node Manager. Process that manages, restarts, and
initializes processes on each ICR node.
each ICR Node.
Rcv RECOVERY Central Database Recovery. Recovers central database
historical data.
SNMP SNMP The SNMP Extension Agent receives an event feed
from the CSFS process and communications with the
Windows NT SNMP Agent to generate SNMP traps
when certain alarmable events occur.
PG Processes

Abs ABS Application Bridge Server. A server process that
operates between the PG and the Aspect CallCenter
ACD. The Application Bridge Server monitors
disconnect and transfer messages between the PG and
application running on the Aspect CallCenter
Application Bridge.
CGI CTIServer CTI Gateway Service. Optional component that allows
an external CTI application to communicate with a
Peripheral Gateway. Can be installed on the same
machine as the PG software or on a separate machine.
Mds MDSPROC Message Delivery Service. Process that provides
reliable message delivery between ICR processes.
Nm NODEMAN Node manager. Process that manages, restarts, and
initializes processes on each ICR node
each ICR Node.
Opc OPC Open Peripheral Controller. Interface between the PIM
and the CallRouter. Supplies the CallRouter with
uniform message sets from different PG types.
Pgag PGAGENT Peripheral Gateway DMP Agent. The Device
Management Protocol Agent that manages session
layer communications between the PG and the
CallRouter.
Pim1 Varies. ex: Aspect Peripheral Interface Manager. The GeoTel proprietary
EventLink PIM, is interface between a peripheral and the PG.
seen as ASPEVPIM
Pim2, Pim3,
Pim4
AW Processes

WebLink WebLink Enabled when WebView is enabled. In ICM 5.0
WebView will be installed for Reporting.
Nm NODEMAN Node manager. Process that manages, restarts, and
initializes processes on each ICR node.
each ICR Node.
Rtc RTCLIENT Real Time Client. Admin Workstation process that
receives real-time data from the Real-Time Distributor
Rtd RTDIST Real Time Distributor. Distributes real-time data to
Admin Workstations.
UAW UpdateAW UpdateAW. Monitors the CallRouter, looking for
configuration updates. If an AW sends up
configuration changes, all other AW’s note the
difference, and pull back the changes.
Lgr LOGGER Database Logger. Process that manages Real-Time and
Five-Minute data and information about the entire ICR
system. This process is similar to the Logger process on
the main loggers.
Replication Replication When AW is a Distributor and an HDS is installed this
process will be enabled to replicate the historicalData
from the Logger to the local HDS.
Node Manager
• Installed as a standard NT service
Node Manager

• One Node Manager per ICM Node (e.g. router, logger, PG,
CG, AW)
Node Manager

CG, AW)
• Starts processes as specified by its registry config
Node Manager

CG, AW)
• Each process has Node Manager registry key under
– software\geotel\icr\<cust>\<node>\NodeManager\CurrentVersion\Proce
sses
Node Manager

CG, AW)
sses
• Node Manager must be restarted to read changes to registry
Node Manager

CG, AW)
sses
• Monitors processes and restarts them if they fail or hang
Node Manager

CG, AW)
sses
• Monitors processes and restarts them if they fail or hang
• Can request system reboot if critical process fails (or fails
repeatedly)
Router Startup
• All processes start and wait for MDS to go “In Service”
Router Startup

• MDS initializes and goes “In Service”
Router Startup

• Router loads state from its peer if available. Otherwise
router requests config from logger
Router Startup

• ccagent accepts connections from DMP devices (PGS, AT&T
NICS and BTNUP NICS)
Router Startup

• Router responds to requests for configuration from PGS and
NICS
Router Startup

• Router responds to requests for configuration from PGS and
NICS
• rtsvr accepts connections from real time distributors
Logger Startup
• Logger and Recovery processes start and connect to

database.
Logger Startup

database.
• Logger connects to MDS on same side router and waits for
MDS to go “In Service.
Logger Startup

database.
• Recovery attempts to connect to its peer and sync up
historical data.
Logger Startup

database.
• Recovery attempts to connect to its peer and sync up
historical data.
• When logger sees MDS go in service it loads its state from
its peer if available. Otherwise it waits for instructions from
the router.
3
Visible / Public Network 1
NMM NMM
NM NM
In 4
DBA CCAG
Service NIC
RCV 2
RTS
7 5
SQL
MDS 6
Enabled Synchronizer
Router Logger
Private Network
Visible / Public Network
8
DBA CCAG
NIC
RCV
RTS
SQL
MDS 9
Router Logger
Private Network
10
DBA CCAG
NIC
RCV
RTS
SQL
MDS
Router Logger
Private Network
11
DBA CCAG
NIC
RCV
RTS
SQL
MDS
Router Logger
Private Network
DBA CCAG
NIC
RCV
RTS
4 3
SQL
MDS
Router Logger
2 5
Private Network
1
DBA CCAG
NIC
RCV
RTS
SQL
MDS 6
Router Logger
Private Network
IP Network Configuration for ICM
• Each synchronized ICM System (CC or PG) uses two

independent data networks:

– Private Network carries synchronization traffic
• Can be a WAN or Ethernet

– Private Network carries synchronization traffic
– Visible Network carries traffic between each side of the
synchronized system and foreign systems
PG1 ACD1 AW2 PG2a PG2b ACD2 PG3a PG3b vru2 vru4
AW1 IP IP
CallRouterA LoggerA CallRouterB LoggerB
IP IP
NIC NIC
IXC Network
Goals of ICM Fault Tolerance
• Minimize time periods during which system is

unresponsive to call routing requests.
• Eliminate single points of failures that could cause the
system to stop.
• Provide disaster protection by allowing the system
components to be geographically separated.
Overview of ICM Fault Tolerance
• All critical parts of the system can be physically

duplicated.
• Communication paths between critical
components can also be duplicated.
– Critical Components
• Central Controller (Router & Logger
• PGs
• NIC’s
• Provide disaster protection by allowing the system

components to be geographically separated.
ICM Approaches to Fault Tolerance
• ICM uses multiple approaches to Fault Tolerance

– ICM is “Self-Healing”
– Node Manager
– Hot Standby
– Synchronized Execution
• Node Manager (nm)

– “First Line of Defense”
– Each Component in the ICM includes a Node Manager process.
• As shown earlier it is in Charge of starting other ICM Processes
• If a process fails, NM will restart it.
• NM can reboot an ICM node in the event of a critical failure.
• NM Properties are established in ICM Setup.
• NM process is managed by the NMM process, Node Manager
Manager.
Hot Standby
• One set of processes is called the Primary or Active
– Active node or process does ALL the work
• One set of processes is called the Backup or

Standby
– Standby node or process watches and waits for the active
process to fail
• PGs and NICs primarily use Hot Standby Fault Tolerance

• PGs also use Synchronized Execution for certain processes.
ICM Fault Tolerance Basics
• Synchronized Processes
1. Synchronized process pairs base ALL action on message input
2. All inputs (messages) received in same order by both
processes
processes
3. Both processes compute same internal state and same outputs
processes
4. Main examples are Router, Logger, OPC
processes
5. When a duplexed synchronized process starts up it attempts to
receive state from its partner. If it can’t receive state, then it
attempts to initialize in simplexed mode.
2. All inputs (messages) received in same order by both processes
6. When a synchronized process receives a request to transfer its
state to its peer it takes a snapshot of its entire synchronized
state. While it is taking the snapshot, new messages are queued.
When it has finished taking the snapshot, it resumes processing
messages while the state data is transferred in the background.
2. All inputs (messages) received in same order by both processes
6. When a synchronized process receives a request to transfer its
state to its peer it takes a snapshot of its entire synchronized
state. While it is taking the snapshot, new messages are queued.
When it has finished taking the snapshot, it resumes processing
messages while the state data is transferred in the background.
7. While a synchronized process is in the processing of receiving
state from its peer, new messages are queued. After the process
has received and processed the entire state, it resumes
processing messages.
• MDS (Message Delivery Service)

– Provides message delivery between “local” processes

– Handles splitting and synchronization of messages

– Handles duplicate output messages (with agent processes)

– All messages sent from one “local” process to another go
through MDS

through MDS
– One MDS process is enabled the other is disabled

through MDS
– One MDS process is enabled the other is disabled
– MDS processes are said to operate in “paired” or “isolated”
mode
PG1 PG2 PG3
AW
IP IP
DBA CCAG DBA CCAG

NIC
RCV NIC
RCV
RTS RTS
SQL SQL
MDS Synchronized MDS
Enabled Synchronizer Zone Disabled Synchronizer
Router Logger Router Logger
IP IP
PG1 PG2 PG3
Running
AW Alone
IP IP
DBA CCAG DBA CCAG

NIC
RCV NIC
RCV
RTS RTS
SQL SQL
IP IP
PG1 PG2 PG3
AW
IP IP
DBA CCAG DBA CCAG

NIC
RCV NIC
RCV
RTS RTS
SQL SQL
IP IP
PG1 PG2 PG3
I’m duplexed.
AW Is my partner
IP IP still alive?
DBA CCAG DBA CCAG

NIC
RCV NIC
RCV
RTS RTS
SQL SQL
IP IP
PG1 PG2 PG3
AW
IP IP
DBA CCAG DBA CCAG

NIC
RCV NIC
RCV
RTS RTS
SQL SQL
IP IP
PG1 PG2 PG3
Yes.
AW I’m Enabled
IP IP
DBA CCAG DBA CCAG

NIC
RCV NIC
RCV
RTS RTS
SQL SQL
IP IP
PG1 PG2 PG3
Request State
AW Transfer
IP IP
DBA CCAG DBA CCAG

NIC
RCV NIC
RCV
RTS RTS
SQL SQL
Enabled Synchronizer Disabled Synchronizer
Zone
IP IP
PG1 PG2 PG3
(Build State)
AW Here’s the
IP IP State!
DBA CCAG DBA CCAG

NIC
RCV NIC
RCV
RTS RTS
SQL SQL
IP IP
PG1 PG2 PG3
Is my Config
AW up-to-date?
IP IP
DBA CCAG DBA CCAG

NIC
RCV NIC
RCV
RTS RTS
SQL SQL
IP IP
PG1 PG2 PG3
Yes or No.
If no, here is
AW the updated
IP IP info.
DBA CCAG DBA CCAG

NIC
RCV NIC
RCV
RTS RTS
SQL SQL
IP IP
PG1 PG2 PG3
Am I missing
AW any historical
IP IP data?
DBA CCAG DBA CCAG

NIC
RCV NIC
RCV
RTS RTS
SQL SQL
IP IP
PG1 PG2 PG3
Yes.
AW Here it is...
IP IP
DBA CCAG DBA CCAG

NIC
RCV NIC
RCV
RTS RTS
SQL SQL
IP IP
PG1 PG2 PG3
AW
IP IP
DBA CCAG DBA CCAG

NIC
RCV NIC
RCV
RTS RTS
SQL SQL
IP IP
PG1 PG2 PG3
AW
IP IP
DBA CCAG DBA CCAG

NIC
RCV NIC
RCV
RTS RTS
SQL SQL
IP IP
PG1 PG2 PG3
AW
IP IP
DBA CCAG DBA CCAG

NIC
RCV NIC
RCV
RTS RTS
SQL SQL
IP IP
PG1 PG2 PG3
AW
IP IP
DBA CCAG DBA CCAG

NIC
RCV NIC
RCV
RTS RTS
SQL SQL
IP IP
PG1 PG2 PG3
AW
IP IP
DBA CCAG DBA CCAG

NIC
RCV NIC
RCV
RTS RTS
SQL SQL
IP IP
PG1 PG2 PG3
AW
IP IP
DBA CCAG DBA CCAG

NIC
RCV NIC
RCV
RTS RTS
SQL SQL
IP IP
PG1 PG2 PG3
AW
IP IP
DBA CCAG DBA CCAG

NIC
RCV NIC
RCV
RTS RTS
SQL SQL
IP IP
PG1 PG2 PG3
AW
IP IP
DBA CCAG DBA CCAG

NIC
RCV NIC
RCV
RTS RTS
SQL SQL
IP IP
PG1 PG2 PG3
AW
IP IP
DBA CCAG DBA CCAG

NIC
RCV NIC
RCV
RTS RTS
SQL SQL
IP IP
PG1 PG2 PG3
AW
IP IP
DBA CCAG DBA CCAG

NIC
RCV NIC
RCV
RTS RTS
SQL SQL
IP IP
PG1 PG2 PG3
AW
IP IP
DBA CCAG DBA CCAG

NIC RCV NIC RCV
RTS RTS
SQL SQL
IP IP
PG1 PG2 PG3
AW
IP IP
DBA CCAG DBA CCAG

NIC
RCV NIC
RCV
RTS RTS
SQL SQL
IP IP
PG1 PG2 PG3
AW
IP IP
DBA CCAG DBA CCAG

NIC
RCV NIC
RCV
RTS RTS
SQL SQL
IP IP
PG1 PG2 PG3
AW
IP IP
DBA CCAG DBA CCAG

NIC
RCV NIC
RCV
RTS RTS
SQL SQL
IP IP
PG1 PG2 PG3
AW
IP IP
DBA CCAG DBA CCAG

NIC
RCV NIC
RCV
RTS RTS
SQL SQL
IP IP
PG1 PG2 PG3
AW
IP IP
DBA CCAG DBA CCAG

NIC
RCV NIC
RCV
RTS RTS
SQL SQL
IP IP
PG Startup
• All processes start and wait for MDS to go “In

Service”
PG Startup

Service”
PG Startup

Service”
• OPC loads state from its peer if available. Otherwise
OPC sends initialize request to one pgagent
PG Startup

Service”
• OPC loads state from its peer if available. Otherwise
OPC sends initialize request to one pgagent
• pgagent attempts to establish an active path to one
side of the central controller (and an idle path to the
other side).
PG Startup

Service”
• OPC loads state from its peer if available.
Otherwise OPC sends initialize request to one
pgagent
• pgagent attempts to establish an active path to one
side of the central controller (and an idle path to the
other side).
• Once active path to central controller is
established, OPC sends request to the router
asking to be told its configuration information.
PG Startup

• OPC loads state from its peer if available. Otherwise OPC sends
initialize request to one pgagent
• pgagent attempts to establish an active path to one side of the
central controller (and an idle path to the other side).
• Once active path to central controller is established, OPC sends
request to the router asking to be told its configuration
information.
• Once OPC successfully receives its config from the router it
attempts to activate the configured PIMS. This causes the PIMS to
connect to the ACDs. If this is successful the PG reports
“Peripheral Online” to the router.
1 NMM 2 PGnA In Service
NM 3
OPC
MDS
PIM1 PIM2 PGAG
ACD1 5 ACD2
IP Router 6
PGnA
OPC
7
MDS
PIM1 PIM2 PGAG

8
ACD1 ACD2
IP Router
PGnA
OPC
MDS
PIM1 PIM2 PGAG
ACD1 ACD2
IP Router
PGnA PGnB
OPC OPC
Synchronized
Zone
MDS MDS
Idle Active Idle Active Idle Active

PIM1 PIM2 PGAG PIM1 PIM2 PGAG
ACD1 ACD2
Idle Data Connection
to Side A CallRouter. PGAG
IP Router Active Data connection to
sends and receives Side B CallRouter. PGAG sends
heartbeats and receives Data & Heartbeats.
and acknowledgments.
Are you
PGnA PGnB
there?
OPC Synchronized
OPC
Zone
MDS MDS

ACD1 ACD2
to Side A CallRouter. PGAG IP Router Active Data connection to
State
PGnA PGnB
Transfer
OPC OPC
Synchronized
MDS Zone MDS

ACD1 ACD2
to Side A CallRouter. PGAG IP Router Active Data connection to
PGnA PGnB
OPC OPC
Synchronized
Zone
MDS MDS

ACD1 ACD2
PGnA PGnB
OPC OPC
Synchronized
Zone
MDS MDS

ACD1 ACD2
PGnA PGnB
OPC OPC
Synchronized
Zone
MDS MDS

ACD1 ACD2
AW Startup
• Real Time Client (rtc) connects to its distributor and sends a

registration request.
AW Startup

• Real Time Distributor (rtd) receives connect request from
client and attempts to connect to preferred Real Time Server
(rts).
AW Startup

(rts).
• Real Time Server receives connect request from distributor
and forwards the initial state of all real time tables (base
records).
AW Startup

(rts).
• Real Time Server receives connect request from distributor
and forwards the initial state of all real time tables (base
records).
• UpdateAW service waits for real time feed to become active
and then brings local config DB up to date according to
config sequence number.
Admin
1 NMM
Real-Time Data from 2 RTDistributor NM Workstation

CallRouter’s RTS process Processes
3
RTClient cust_awdb
Logger
UpdateAW
cust_hds
Historical Data from Replication
Central Controller Logger
Admin
Real-Time Data from RTDistributor Workstation
CallRouter’s RTS process Processes
4
RTClient cust_awdb
5
Logger
UpdateAW
cust_hds
Historical Data from Replication
Central Controller Logger
Just showing Processes involved in Configuration change
(Intra-process communication on the Router and Logger flow thru MDS)
Logger Call Router Dist AW
RTServer
9 RTDist
8 10
5
Logger Router RTClient
3 2 6 11
UpdateAW
DBAgent
4 12 13
Logger 14 AWDB
12
CC DB 1 7
15
upcc.dll AWRTClient
Configure ICM
(on an AW)
Central Controller Failure Scenarios
Quiz
Quiz
In the context of ICM,

what is a NIC? 1
a) Network Interface Card
b) Network Intermediary Connector
c) Network Interface Controller
d) Non-Interruptible Computer
Quiz
What is the primary function

of the Logger? 2
a) To store Configuration Information and
Historical Data
b) To cut down trees in the forest
c) To connect Peripheral devices to the Call
Router
d) To respond to route requests from a
carrier network
Quiz
What are examples of

Peripheral Devices? 3
a) Modem, Scanner, or Printer
b) Automatic Call Distributor or Voice
Response Unit
c) Admin Workstations
d) AT&T, MCI, or Sprint NICs
Quiz
What three things comprise

a route request? 4
a) Dialed Number, Calling Line ID, and Caller
Entered Digits
b) Dialed Number, Caller Entered Digits, and
Peripheral Gateway Number
c) Dialed Number, ACD, and Calling Line ID
d) Calling Line ID, ANI, and Dialed Number
Quiz
PGs connect to ACDs over

the following logical 5
connections?
a) MIS Link
b) CTI Link
c) TCP/IP Link
d) A and B
Quiz
The AT&T NIC connects to

the AT&T Network via:
6
a) Cross Links
b) A-Links
c) Missing Links
d) Cuff Links
Quiz
AW is an acronym for: 7
a) Motor Oil Weight
b) Wire Gauge
c) Administrative Workstation
d) Admin Widget
Quiz
The Private Network is used

for: 8
a) Synchronization and State Transfer of
Central Controller data
b) Passing Private messages
c) Connecting to Peripherals
d) Connecting to Admin Workstations
Quiz
The AWDB is used for: 9

a) Storage of Historical Data
b) Storage of Real-time data
c) Storage of Configuration data
d) A & B
e) B & C
Quiz
The ICM uses what approach 10

to fault tolerance:
a) Synchronized Execution
b) Duplicate Data Paths
c) Hot Standby
d) Node Manager
e) All of the above
Quiz
MDS is an acronym for: 11

a) Monitor Distribution Service
b) Message Delivery Service
c) Magical Data Synchronizer
d) Mail Delivery System
Quiz
What happens if an individual

process fails? 12
a) Node Manager will attempt to restart it.
b) The process is restarted by the INIT
kernel
c) It remains stopped
d) The PC reboots
Quiz
Which of the following best

describes "synchronized 13
execution?"
a) Duplicate active processes receive
identical input messages, and produce
identical output
b) There is one active process, and one or
more idle backups
c) Two systems each do half the work
Quiz
Which of the following best describes

the Hot Standby approach to fault
tolerance?
14
a) Duplicate active processes receive
identical input messages, and produce
identical output
b) Two systems each do half the work
c) Two electric chairs
d) There is one active process, and one or
more idle backups
Quiz
What is the responsibility of the Central

Controller Agent (CCAgent)? 15
a) To eliminate 007 and take over the world
b) To maintain session layer connectivity
with PGs and NICs.
c) To store and forward real-time data to
AWs
d) To enforce all the laws that haven't been
passed yet
Quiz
What are the functions of the

Admin Workstation? 16
a) Create, Modify, and Edit configuration
information
b) Create, Modify, Edit, Schedule, and
Monitor Routing Scripts
c) Create, Edit, and Monitor Reports
d) All of the above
Quiz
What is PIM an acronym for? 17

a) Personal Information Manager
b) Peripheral Intermediary Module
c) Peripheral Interface Manager
d) PC Interface Monitor
Quiz
It is possible to run ICM on 18

Windows 95/98? T/F.
T.
F.
© 2002, Cisco Systems, Inc. All rights reserved.

IPCC Server Services

Cargado por

Información del documento

Derechos de autor

Formatos disponibles

Compartir este documento

Compartir o incrustar documentos

Opciones para compartir

¿Le pareció útil este documento?

¿Este contenido es inapropiado?

Copyright:

Formatos disponibles

IPCC Server Services

Cargado por

Copyright:

Formatos disponibles

© 2002, Cisco Systems, Inc. All rights reserved.

© 2002, Cisco Systems, Inc. All rights reserved. 2

IP IP CMS IP vru1 vru3

IXC Network NIC

IP IP CMS IP vru1 vru3

IXC Network NIC

IP IP CMS IP vru1 vru3

Prefix Process Name Description

Prefix Process Name Description

Prefix Process Name Description

Prefix Process Name Description

• Installed as a standard NT service

• Installed as a standard NT service

• Installed as a standard NT service

• Installed as a standard NT service

• Installed as a standard NT service

• Installed as a standard NT service

• Installed as a standard NT service

• All processes start and wait for MDS to go “In Service”

• All processes start and wait for MDS to go “In Service”

• All processes start and wait for MDS to go “In Service”

• All processes start and wait for MDS to go “In Service”

• All processes start and wait for MDS to go “In Service”

• All processes start and wait for MDS to go “In Service”

• Logger and Recovery processes start and connect to

• Logger and Recovery processes start and connect to

• Logger and Recovery processes start and connect to

• Logger and Recovery processes start and connect to

• Each synchronized ICM System (CC or PG) uses two

• Each synchronized ICM System (CC or PG) uses two

• Each synchronized ICM System (CC or PG) uses two

IP IP CMS IP vru1 vru3

CallRouterA LoggerA CallRouterB LoggerB

• Minimize time periods during which system is

• All critical parts of the system can be physically

• Provide disaster protection by allowing the system

• ICM uses multiple approaches to Fault Tolerance

• Node Manager (nm)

• One set of processes is called the Backup or

• PGs and NICs primarily use Hot Standby Fault Tolerance

• MDS (Message Delivery Service)

• MDS (Message Delivery Service)

• MDS (Message Delivery Service)

• MDS (Message Delivery Service)

• MDS (Message Delivery Service)

• MDS (Message Delivery Service)

DBA CCAG DBA CCAG

Router Logger Router Logger

DBA CCAG DBA CCAG

Router Logger Router Logger

DBA CCAG DBA CCAG

Router Logger Router Logger

DBA CCAG DBA CCAG

Router Logger Router Logger

DBA CCAG DBA CCAG

Router Logger Router Logger

DBA CCAG DBA CCAG

Router Logger Router Logger

DBA CCAG DBA CCAG

DBA CCAG DBA CCAG

Router Logger Router Logger

DBA CCAG DBA CCAG

Router Logger Router Logger

DBA CCAG DBA CCAG