Telecoms And Data Network Management, Services And Billing

T-110.300/301

The FCAPS Model

Fault Management Configuration Management Accounting Performance Management Security Management

Part of Telecommunications Management Network (TMN) standard from ITU Same concepts are also important for data networks

Fault Management
Detect, isolate, repair abnormal network behavior Monitor network status
SNMP, CMIP, TMN standards Commercial and homebrew tools On traditional telco side the most important part of FCAPS

Long term aim of the fault management is to:

Recognize trends Find better solutions

Fault Management

The immediate fault handling process is

Receive an alarm Identify the symptoms of the problem Open ticket Send an immediate report if needed Detect the actual cause Fix the problem Document the solution Report and close the ticket

Configuration Management
Physical and logical network configuration How to
Add, remove and update equipment Install it physically Configure the equipment Manage versions of the configuration and Fall back to a previous configuration Keep track of the equipment Do preventive maintenance

Configuration and change management is a very important part of network management Autoconfiguration is often preferable, especially when the hardware is physically located in a separate place
E.g. cable modems, plug and play components

Accounting
Collect and analyze information about resource usage Monitoring the usage and charging for it is often an important requirement
Like, companies that dont make a profit dont live long

High detail level accounting is very expensive

E.g. each transaction in a web proxy This is the specialty of telcos

Accounting is related to financial functions, mostly billing Produces also internal data
Required for performance monitoring Also used for fraud monitoring

Performance Management
Collect and analyze throughput and statistical info How to ensure then enough capacity is available for services
Network bandwidth Over each link, taking latency into account Router memory and CPU Disk space etc.

How to optimize the usage of available capacity

90% optimal is generally good enough Perfect is the worst enemy of good Planning for faults Extra capacity, redundancy Change of operations mode

Performance Management
How to find other solutions for providing performance
besides adding capacity Smarter router interface cards instead of adding CPU HTTP caches

Performance management is constant tuning and monitoring

Like how much to pre-fetch to the HTTP cache

The requirements must be planned beforehand

Sometimes an order can take years to complete (undersea cables)

Differences in performance can make or break an operation

Security Management
Control access to resources through authentication and authorization policies The production network must be protected
The security must be incorporated in the design On IP based networks separate management and user data networks (like telco user and control planes) Implementation is important Constant monitoring Constant design review Practical issues, like key and password distribution

Security can also be provided to the customers as a service

Virtual Private Networks Firewalls Content filtering

Telecoms Network Management Logical Layered Architecture

TMN is a standard from ITU Business Management Layer
Vaguely defined, business strategic goal setting

Service Management Layer

QoS management, customer configuration, accounting etc.

Network Management Layer

Routing and dedicated network paths, link utilization monitoring, optimizing etc.

Element Management Layer

Collects information from individual network elements and sets the configuration in NEs

Network Element Layer

Contains the actual network elements (switches, SS7 components etc.)

TMN In Practice
Concepts are currently in use Only a few network equipment manufacturers are actually implementing CMISE/CMIP
Common Management Information Service Element Common Management Information Protocol E.g. latest Nokia GSM networks Much of current infrastructure relies on SNMP or proprietary protocols Simple Network Management Protocol, uses UDP/IP

HP Open View is often used to implement the higher level functionalities

Implements the idea of unified, advanced network management system Supports SNMP, CMIP and other protocols

Network Management With SNMP

Simple Network Management Protocol
Defined in RFC-1157

Network elements (routers, hosts, printers etc) have a SNMP agent Management station queries network elements for information
The agent is a server, the management station is the client

Management Information Base (MIB) describes the information served by SNMP agents
A specification

Network Management With SNMP

Router

Switch

Server Management station Managed network elements have agent software

Management Information Base

Specifies what variables the network elements maintain Variables are the information that can be queried and set by the manager Specifies standardized object identifiers Variables are named using a scheme that is hierarchical and is unlimited in expansion,
E.g. iso.org.dod.internet.mgmt.mib-2

There is a branch in the naming tree for private enterprises (usually manufacturers of network hardware) to locate their own MIBs

MIB Example
sysUpTime OBJECT-TYPE SYNTAX TimeTicks MAX-ACCESS read-only STATUS current DESCRIPTION "The time (in hundredths of a second) since the network management portion of the system was last re-initialized." ::= { system 3 }

ASN.1 is used for the definition Data types: Integer, DisplayString, TimeTicks, .. More complex data types can be constructed from the basic types

SNMP Network Management In Action

Network manager software is configured with the network layout and the MIBs of different network elements Network manager regularly queries the network elements and displays the information to human supervisor When the management software finds something wrong, for example a router does not reply to queries for a while, the software alerts the human supervisor Network manager may set variables in a network element, e.g.. the address of a DNS server A network element may send a "trap", for example a printer may signal that it is out of paper

CMIP
Common Management Information Protocol A more complex and capable protocol comparable to SNMP Addresses many of the shortcomings of SNMP, is also more complicated and requires more resources In many cases agents might be too heavy for practical use as compared to SNMP Currently should be considered only if network management is of serious importance

Other Network And Service Monitoring Tools

SNMP is oriented towards collecting data on network equipment
A MIB and an agent is needed for each service to be monitored

Often services can be monitored with simpler tools Example: Big Brother, http://www.bb4.org
Unix based collection of scripts, that connect the servers on the network and gather data A Web based display that shows if everything is working OK or not The source code is included (priceless!)

Independent and homebrew tools provide more depth for analysis

Duplicate measurement using dissimilar methods help to separate actual problems from problems in the measuring system

Solving Network Problems

Things will go wrong
Make a checklist beforehand Try to have diagnostic tools for all possible problems

Never confuse a measurement with the thing being measured

Tools show what they think is the fault

Stay calm, focused and methodical

Check the cables

If you spend all your time handling problems, something is wrong

Stay on the ball instead of chasing it

Service Management
Services are above the network layer, they are what the users see
Properly configured networks with redundancy might experience partial failure without affecting the services

Managing and monitoring services is not as well standardized and layered as the network level
This is becoming an important business area There is a strong trend towards allowing the users configure the services themselves "self provisioning"

After the service management is figured out, business management is the next level

Providing And Managing Services

Know your services, the network is only the delivery mechanism Generally the services drive the network design
How much bandwidth needed and where? How many servers, how to balance the load?

Sometimes network problems can be solved by adapting the services

One solution to web user overload (flash crowd) is to make the service require less bandwidht Eg. the CNN "war mode" with less graphics

Content Delivery Networks

How to provide a very popular (Web) service to all the users of the Internet? The content is distributed to several web servers in different physical (or network topological) locations Domain Name System (DNS) is used to direct the traffic to the server considered nearest Content can be duplicated to different servers Managing a CDN is quite complex Currently Akamai is the market leader

Network Management Thoughts

When designing a network:
Think about the big picture, what is the reason for this network Think about the future

When maintaining the network daily:

Think about making your work easier Take care of the details

When solving a network problem:

Think calmly and methodically Do not rush, if the first three attempts do not fix the problem, go get a cup of coffee

Billing

Billing Basics
Billing is a very important feature of the modern telecommunications networks Initially paper tickets written by operators
Low cost work power made this feasible

Next an electrical counter for each device (telephone line)

Active connections received pulses, which increased the counter Value added billing (long distance calls) was generated by increasing the pulse frequency

Currently billing is entirely computerized In modern networks:

The user is charged per minute/second or unmetered or prepaid Usage data is collected in Call Detail Records (CDR) Large files or databases containing CDRs are processed to produce a bill

Telecom Billing System Overview

etc WAP GW SMSC CDRs GPRS MSC gw MSC IN Roaming Mediation Subscriber management Billing process

Printing

Delivery

Operative systems

The Billing Process

Services, switches, SMSCs and other Network Elements produce
Session, usage & transaction data: CDRs, tickets, events, other logs

Mediation functionalities
Aggregate, correlate, filter and normalize the tickets, producing Records of who, when, what, (where, why)

Rating function
Prices the records, producing Priced usage records

Billing function
Creates the invoices taking into account: Accounting, payments, collections, tax, discounts

Presentation
Formatting, printing and delivery of the bill

Call Detail Records

The CDR can contain
Called / Received / Forwarded Type of basic (GSM) service: speech, data, fax, sms Type of supplementary service IN triggering Event time and date Event duration Charging principle (IMSI, MSISDN) B subscriber, C subscriber Pre-billing information

Mediation
Collects or receives the information from the various network elements
CDRs or raw data

Aggregates the records

For batch processing

Normalizes the records

Information can be lacking, must be collated etc.

Filters unmeaningful records

Operator's own rules

Produces statistics

Rest Of The Billing Process

Rating
Price tags are put on unified usage records Price is based on tariff tables A function of time, duration, local/long distance etc.

Billing
Discounts (family, volume) Usage matched with customer accounts Connection to Customer Care Software (CRM) Tax

Invoicing
Formatting the actual bill Customers can have several subscriptions Printing and delivery

Hot Billing
Also called real time billing Needed for spending limits and pre-paid accounts In theory customer's account is monitored in real time and the service is shut off when the limit is reached In practice this is impossible
Real time is expensive (processing requirements) Simultaneous services available (voice, SMS, GPRS)

Real time billing requires specialized software and tight systems integration across the services

Billing Schemes
Monthly flat rate fee (ADSL, GPRS, local calls) Per minute (speech, circuit data, video conferencing) Per megabyte (GPRS, WLAN) Per message (SMS, e-mail) Per view (video on demand) Per transaction (music downloads) Per click (click through advertising (billing can be negative, too)) Per page (fax) Per $ (commission) Per bullet (interactive games) Or per combinations of the above Or modified by daytime, volume etc.

An Estimate On CDR Volumes

by Jean-Claude Sotto GSM
2 CDRs / call 3 calls / subscriber / day ~6 CDRs / subscriber / day

GPRS
15 CDRs / PDP context (WAP + GGSN + SGSN) 3 contexts / subscriber / day ~45 CDRs / subscriber / day

UMTS
GPRS + Services + QoS -> 15 + 4 + 2 CDRs 6 contexts / subscriber / day ~130 CDRs / subscriber / day

Changing Billing Models

More actors
Network operator Service (virtual network) operator Content service provider Content provider Sponsoring Cost sharing Service provisioning Commission

New billing models

Payment Brokering
Who should manage the billing
More services and actors The telco billing system is going to be used for purchasing hard goods instead of just services? But the operator does not want to become a financial institution (regulation)

A broker (small or large financial institution) could manage the billing service
post-paid billing prepayment (wallet) customer and service identification micro (average <1 ) or macro (average >10) payments

Summary Of Billing
Billing is an important telco process It requires heavy duty processing Many operators have multiple billing systems for historical reasons
Fixed line telephony, mobile telephony, Internet services Also there is a psychological limit to what the customer is willing to accept as a "communications bill"

So Long
Source and No Money will get you through the hard times better than Money and No Source.
Bob Kodner

Facts without theory are useless Theory without facts is bullshit

Unknown

When in fear, and when in doubt; Run in circles, scream and shout!
Unknown