SlideShare une entreprise Scribd logo
1  sur  11
Télécharger pour lire hors ligne
Tail-f Systems Whitepaper

Closing the Services Configuration Gap

Introduction
Delivering valued-added services, like MPLS VPNS, Metro Ethernet, and IP TV is critical to the profitability
and growth of service providers. In most operator networks, there are bottlenecks that get in the way of
activating services in a timely and cost effective manner. We call this the “Services Configuration Gap.”

Current systems use either legacy software or third party applications that are expensive and inflexible, often
requiring manual intervention, frequent software updates, and maintenance of layers of mediation software.
These systems are being strained by the increased rate of change as services are enhanced, new services are
introduced, and network devices are replaced and added. Both OSS and EMS developers are burdened with
these problems. For network equipment vendors, developing EMS applications, the focus is more on device
adaptations than service applications.

This whitepaper describes the Services Configuration Gap and proposes a new generation of configuration
management tools to deal with the fast changing environment faced by service providers and OSS and EMS
developers. The proposed approach leverages dynamic data modeling, automated configuration, and
mechanisms for transaction integrity. By automating the connection between service activation and device
configuration, the excess cost and delay associated with hardwired, service-specific software is eliminated.

Service Configuration Today

The steps involved in creating and activating a service include many processes, support systems, and network
devices. As shown below in Figure 1, the TM Forum’s eTOM Process Map divides the processes into two
major areas: strategy (left) and operations (right).




                                        Figure 1: The eTOM Process Map

The strategy product lifecycle addresses service introduction and retirement. The design of new services
includes how the service is represented in systems like customer care, assurance, and billing and requires
services to be modeled so that all systems can have a common view of their characteristics and decomposition.




                                              Tail-f Systems © 2010
                                                   Page 1 of 11
Tail-f Systems Whitepaper

The design and modeling includes the mapping to the
resource layer, as well as defining what devices and
configurations are needed to support a service. When a
service has been designed it is handed over to operations
processes which are responsible for activating the service
in the network. The activation process which includes
assurance and billing is shown in Figure 2.

For example, if a customer orders an MPLS VPN, the
order handling process makes sure that the order is
properly managed including credit authorization and
overall order tracking. The order process then triggers the
service configuration and activation process which in turn
configures the customer service. The process might
include manual steps such as the installation of                      Figure 2: eTOM Service Activation
customer equipment.

Service models are commonly expressed in UML models and implemented in vendor-specific repositories at
the OSS level. The device data models are typically available as SNMP MIBs and vendor-specific CLIs.

Challenges with Existing Service Configuration Approaches
Service Models

The Services Configuration Gap is a function of the way existing service models and their mapping processes
connect to the corresponding device data models. Due to the rapid change of customer requirements, adoption
of new services, and a constantly changing network topology, these models and mappings must be frequently
revised. Today this process is semi-automated and not dynamic, resulting in changes often being reflected
throughout the system in weeks rather than hours.

To reach new time-to-market expectations, the next generation of service management tools must have the
following capabilities:
     • Be model-driven and allow user and programmatic interfaces to be automatically updated from the
        service model without other inputs.
     • Support model-aware APIs that let programmers work directly with classes from the service model
        rather than low-level protocol based APIs.
     • Ability to load new versions and models at runtime and propagate changes immediately.
     • Modeling tools and support that allow for quick changes while still validating the model correctness
        and completeness.
     • The model must interface seamlessly with the subsequent mapping step to ensure that the complete
        chain from service model to device model is managed.
     • Modern software engineering approaches including modularity, aspects, and annotations (rather than
        inheritance) should be used to avoid dependencies and long build times.




                                              Tail-f Systems © 2010
                                                   Page 2 of 11
Tail-f Systems Whitepaper

Device Management

The process of configuring network devices is generally a manual and time-consuming task. Unlike alarm and
performance management, configuration activities involve writing data and changing states on devices whereas
alarm and performance management is mostly a read-only activity. Furthermore, the configuration process also
has more complex operation requirements such as validation and transactions. An additional challenge is the
fact that configuration data models are vendor-specific and often hard to understand.

While network configuration in large networks will always be demanding and need skilled network
administrators, this should not prevent the use of automated configuration management at the "plumbing
level.” Currently, most configuration management operations are performed using scripts on top of command
line interfaces or even partly with SNMP. The use of CLIs and SNMP comes with major limitations as
discussed below.

SNMP MIBS

    •   It is hard to find the correct MIB Modules using SNMP. There is no defined discovery procedure
        other than an ad hoc SNMP walk and, even after a MIB walk, there is no formal and automated way to
        find the correct MIB module.
    •   All MIBs do not define fundamental concepts, including which data is configuration data versus
        operational data, sequencing and transactional dependencies, or constraints between variables.
    •   The module structure of SNMP MIBs has worked well in guaranteeing unique identifiers. However,
        the administrative based tree-structure does not suit automated rendering of interfaces or favor
        readability and extensibility in a logical way.

CLI Models

    •  The data-model for CLIs is informally defined and is managed by scripts that parse the actual strings.
      This is a fragile way of working since misspellings and version changes break the scripts.
    • There is no room for automated rendering of tools since no formal model is defined.

While improvements can be made to traditional solutions at considerable cost, the result falls short of a fully
automated approach where the latter includes the following capabilities:
    • Discovery of data models.
    • Resolution of data model versions.
    • Rendering of interfaces from data models.
    • Synchronization of configuration data from and to devices.
    • Validations and transactions to ensure correct configuration.
    • Transactions that are tolerant of unreachable devices and capability variants.

The Service Configuration Gap is a byproduct of inflexible, partially automated systems that struggle to keep
up with the rate of change in a service operator’s environment. Symptoms of the Service Configuration Gap
include:
    • Delays in the introduction of new services that result in the postponement of incremental revenue.
    • The need for large teams of custom programmers to update and maintain adapter software.
    • Configuration errors that result in loss of service and negatively impact customer satisfaction.




                                               Tail-f Systems © 2010
                                                    Page 3 of 11
Tail-f Systems Whitepaper

Dynamic Service Configuration

The combination of dynamic service modeling and automated configuration is the basis of the next generation
of configuration management tools and we call this new approach “Dynamic Service Configuration.” With
Dynamic Service Configuration operators are able to dynamically adapt the service configuration solution
according to changes in their service portfolio, rather than being based on raw device configurations.

The configuration management tools that enable
Dynamic Service Configuration sit between the network
devices and the order management system. As shown in
Figure 3, with this new approach there will be a Service
Layer and a Device Layer. The Service Layer supports
dynamic modeling of services with short development
time and model-driven dynamic updates of dependant
interfaces. The Device Layer automatically manages the
device interfaces, such as synchronization of
configuration data, multinode configuration
deployment, and interface versioning.

In the context of the eTOM processes (see Figure 2),
Dynamic Service Configuration implements Service
Configuration and Activation and Resource
Provisioning.

Many operators have systems in place for managing
work-flows across the complete OSS chain, our
proposed approach sits below such systems and
provides an efficient network facing transaction engine.
Furthermore, it replaces in-house CLI scripts or
expensive adaptors and provides a dynamic and                         Figure 3: Dynamic Service Configuration
managed solution for such environments.

The input to the configuration management system is a request to change or add a service, such as a NGOSS
Contract or an MTOSI request. The output from the system is a validated network change, an alert that
validation has failed, or the proposed change was rejected due to some other reason.

Dynamic Service Configuration uses the NETCONF protocol and YANG data modeling as foundations.
YANG is used to bridge the higher level models to device interfaces and NETCONF is used as the means of
deploying the configuration to the devices. This approach can benefit from many automated features such as
model and transaction management where devices support these standards directly. Non-NETCONF devices
such as SNMP and CLI managed network elements can also be mapped by using a mediator and still benefit
from the NETCONF design principle with support of transactions and model-based management.

NETCONF and YANG are described in more detail below.




                                              Tail-f Systems © 2010
                                                   Page 4 of 11
Tail-f Systems Whitepaper

NETCONF

The Network Configuration Protocol, NETCONF, is an IETF network management protocol and is published
in RFC 4741. NETCONF is being adopted by major network equipment providers and has gained strong
industry support.

NETCONF provides mechanisms to install, manipulate, and delete the configuration of network devices.
Its operations are realized on top of a simple Remote Procedure Call (RPC) layer. The NETCONF protocol
uses XML based data encoding for the configuration data as well as the protocol messages. NETCONF is
designed to be a replacement for CLI based programmatic interfaces, such as Perl + Expect over Secure Shell
(SSH). NETCONF is usually transported over the SSH protocol, using the ’NETCONF’ sub-system and in
many ways it mimics native proprietary CLI over SSH interfaces available in a device. However, it uses
structured schema-driven data and provides detailed structured error return information, which the CLI cannot
provide.

Operators need a way to distribute changes to the devices and validate them locally before activating them.
This is indicated by the two bottom options in Figure 4 where configuration data can be sent to candidate
databases in the devices before they are committed to running in production applications. All NETCONF
devices must allow the configuration data to be locked, edited, saved, and unlocked. In addition, all
modifications to the configuration data must be saved in non-volatile storage.




                                      Figure 4: NETCONF Configuration Databases


YANG

YANG is a data modeling language that provides breakthrough advantages to modeling configuration and state
data. The YANG modeling language is currently being standardized by the IETF in the NETMOD working
group. The authors of YANG were involved in the development of the next generation SNMP SMI and are
very experienced in the IETF network management arena. Modern network standards groups such as the Metro
Ethernet Forum are adopting YANG as a device configuration standard.



                                             Tail-f Systems © 2010
                                                  Page 5 of 11
Tail-f Systems Whitepaper

YANG is tree-structured rather than object-oriented. Configuration data is structured into a tree and the data
can include complex types such as lists and unions. The definitions are contained in modules and one module
can augment the tree in another module. Strong revision rules are defined for modules.

YANG also differs from previous network management data model languages through its strong support of
constraints and data validation rules. The advantages of using YANG for data models include:
    • YANG is simple, easy to learn, and most important of all, it is easy to read and understand. The
         complexity of other languages has made it hard to review information models and find problems,
         thereby undermining the ability of developers to build management applications.
    • YANG is written for network management. It is a domain specific language designed to do what it
         needs to do well. It is not constrained by or overloaded by any framework it is derived from.
    • YANG is flexible and modular. Existing modules can be augmented in a controlled manner.

An Example of Dynamic Service Configuration in Action

The following is an illustration of Dynamic Service Configuration in action using YANG as a general
modeling language and showing the power of dynamically modeling services and automated configuration of
network devices.

We are using the scenario of activating MPLS VPN for customers. The hypothetical network topology for this
example is shown in Figure 5 below.




                                           Figure 5: Network Topology

A VPN contains customer devices attached to the Customer Edge (CE) routers. The customer devices use a
MPLS VPN to exchange information between devices within the VPN. The CE routers are not aware of the
VPN as this is managed by the Provider Edge (PE) routers. At service fulfillment time, network resources are
allocated in the required geographic sites to suit the customer. Having selected the devices to be used, the
configuration system can build the required resource facing service template and then populate it with the
customer specific instance details. The service and device model for this example is shown in Figure 6 below.
                                              Tail-f Systems © 2010
                                                   Page 6 of 11
Tail-f Systems Whitepaper




                                      Figure 6: Service and Device Model

The classes at the top of Figure 6 are standard classes in the Service and Device Layers. The Device Class is
automatically augmented with the YANG modules that are discovered from the device (in this case, with PE
and CE classes). The CE router, being VPN agnostic, contains a simple model for IP interfaces while the PE
routers model MPLS information in addition to standard IP interfaces.

The SID standard Resource and Customer Facing Service Classes are augmented as part of a dynamic service
modeling activity. The Customer facing Service Class is augmented with the Customer VPN service and the
Resource facing service with the MPLS VPN. The Customer VPN service represents what the customer
bought, a VPN service connecting the points of presence. The network facing service (MPLS VPN) is the
implementation of the customer service using MPLS technology and it refers to the access links connecting the
CE with the PE. The separation of customer facing services versus the resource facing services is an important
modeling pattern inherited from the TM Forum SID model.

Next, a fully meshed MPLS VPN “RED” is created. The core network is assumed to be configured (PE
routers). This scenario involves the following steps:
    • Setup connectivity PE1-CE1 and PE2-CE3
    • Creation of the VPN RED resource facing service instance “vpn-red” and its access links “northwest”
         (PE1-CE1) “southeast” (PE2-CE3).
    • Creation of the customer facing service instance “acme-vpn.
    • Addition of “acme-vpn” to customer “acme”.

This model is depicted in Figure 7 below.




                                              Tail-f Systems © 2010
                                                   Page 7 of 11
Tail-f Systems Whitepaper




                                          Figure 7: Meshed MPLS VPN

First, the CE interfaces on the PE routers facing VPN RED service are created as shown in Listing 1 below.

ncscli >set ncs managed-device pe1 config interface eth1 enabled ip 192.168.7.2 mask 255.255.255.0
ncscli >set ncs managed-device pe2 config interface eth3 enabled ip 192.168.6.2 mask 255.255.255.0
ncscli>commit
                                         Listing 1: Create Service in CLI

Note the commit command at the end. This will result in a multi-node transaction towards two PE routers.
Either the complete transaction succeeds or fails and in the latter case can be roll-backed.

The next step is to create the mpls vpn “vpn-red” with its access-links “north-west” and “south-east”:

ncscli> set ncs service vpn-red service-type mplsVpn access-link north-west ce ce1 ce-if eth0 pe pe1 pe-
if eth1 subnet 192.168.7.0/24
Value for 'route-distinguisher' (<string, max: 16 chars, min: 3 chars>): 300:1
Value for 'ncs service VPN_GREEN properties mplsVpn vpn-id' (<RFC 2685>): 300:1

ncscli>set ncs service vpn-red service-type mplsVpn access-link south-east ce ce3 ce-if eth0 pe pe2 pe-if
eth3 subnet 192.168.6.0/24
ncscli>commit
                             Listing 2: Create Resource Facing MPLS Service in CLI

The constraints expressed in the model will ensure that only references are made to valid objects for the
access-link. The “Value for” prompts are automatically generated from the model and the CLI asks for these
attributes since they are mandatory. While not possible to illustrate here, the automatically rendered CLI will
only accept keyboard entries according to the data type, so for example only digits and dots in the right places
are allowed when the IP address is entered.

If for some reason the network changes are undesired, a rollback “undo” can easily be performed.

Next the customer facing VPN service “acme-vpn” is created as shown in Listing 3 below. The command
refers to the existing customer “acme”.

                                              Tail-f Systems © 2010
                                                   Page 8 of 11
Tail-f Systems Whitepaper

ncscli> set ncs service acme-vpn service-type custVpn mplsVpn vpn-red vpnCust acme
ncscli> commit

                                  Listing 3: Create Customer Facing VPN in CLI

This refers to the VPN service “vpn-red” allocated previously and to the customer “acme” that needs to be
associated with the service resource. The customer is defined in an external system and queried for validity at
creation time.

The above examples of modeling and configuration have been illustrated with CLI commands, but this could
have equally been achieved with a Web Interface dynamically generated by the model.

Mapping Service Models
When a customer wants to add access to their VPN, the Service Logic needs to define the new CE, allocate the
required PE port, and define what VPN types are needed to be configured. Our approach is to define the
Customer Service in a YANG model so that the service fulfillment is simply the mapping of the Service Layer
model to the underlying Device Layer models.

If a device does not support NETCONF, a mediator embeds transactional mapping logic which resolves many
repetitive and error-prone transformation tasks. The two main interface categories for OSS systems, such as
inventory systems or other network management systems, are SQL databases or APIs. In these cases, a YANG
view of the required data is created and mapped to the external database or API.

From the perspective of the configuration management application, it looks as though the data resides within
the application while every access is forwarded to the external system. This approach gives the following
benefits to model mapping:
     • Allows for federation as an approach to data integration.
     • The service logic is always a YANG to YANG mapping procedure. No service mapping logic is
         contained in adaptors to other interfaces.
     • It is possible to provide solutions to hard problems like transactions, calculating delta operations,
         validation of data, and pre-provisioning.
     • Device configuration integration is reduced to a minimum for NETCONF devices.

Rather than requiring the implementer to write code for every model attribute and operation, the definition of a
single create template is all that is required. All other operations like sets and deletes are automatically
calculated. In coding terms, this means 1,000 lines of code can be collapsed to 100 lines, reducing service
activation lead times, programming, and maintenance costs.

Implementing Dynamic Service Configuration
There are many implementation details that will be required to make systems using Dynamic Service
Configuration effectively integrate with existing systems and operations.

Examples of these implementation details include:
   • Capability to automatically discover the supported device modules and resolve transactions across
       different versions.
   • Managing the propagation of changes from service to device models. When a service is changed, the
       corresponding device changes are managed in a separate transaction. These changes can be inspected
       and used for reverse operation calculations.

                                              Tail-f Systems © 2010
                                                   Page 9 of 11
Tail-f Systems Whitepaper

    •   Pre-provisioning – used when the configuration management systems caches the operations for
        network elements not installed. In normal circumstances, transactions would fail because a device
        does not respond, but with this capability the missing commands are executed whenever devices are
        available.
    •   Backlogging – used when changes to the network must be made as soon as possible and the
        transaction should succeed even if some devices are not responding. Here operations to non-
        responsive devices are cached.
    •   Capability to enforce all service models to implement self-test functions. For example, in the
        MPLS scenario the MPLS VPN will perform a MPLS ping to verify the connectivity. Actual test
        instrumentation may be available on the devices themselves or in external probes.
    •   Ability to restore a broken service. A typical scenario is where technicians modify devices manually
        without being aware of negative effects on services.
    •   Replay function to show the differences between elements with inconsistent configuration parameters.
        This is used to see which devices are out of synch versus the desired configuration and the
        corresponding audit log.

Conclusions

To effectively deliver new services in a rapidly changing environment, service providers cannot continue to
rely on existing configuration management systems.

Current systems use either legacy software or third party applications that are expensive, inflexible, require
manual intervention, and frequent software updates plus maintenance of layers of mediation software. These
systems are being strained by the increased rate of change as existing services are enhanced, new services are
introduced, and network devices are replaced and added. As a result, there is a Services Configuration Gap
that can be measured in terms of delays in introducing new services, lost revenue, and excess software
maintenance costs.

A new approach to configuration management is proposed called Dynamic Service Configuration. Dynamic
Service Configuration leverages the NETCONF and YANG standards and is based on dynamic data modeling,
automated configuration, and transaction integrity. From a software implementation point of view, 1,000 lines
of code can be collapsed to 100 lines, reducing service activation lead times, programming, and maintenance
costs.

About Tail-f Systems

Tail-f Systems is a leading developer of configuration management software for networking equipment
providers, EMS providers, and OSS development organizations. Six of the ten largest global networking
equipment providers are Tail-f Systems’ customers.

Tail-f Systems provides a full family of products to build and extend network management systems.
ConfD, the company’s on-device configuration management toolkit, renders management interfaces from a
single data model and includes a lightweight embedded database. Customers using ConfD radically reduce
their time-to-market and benefit from carrier-grade implementations of NETCONF, CLI, Web, and SNMP
interfaces.

Active in standards development organizations, Tail-f Systems is the leading provider of configuration
management solutions based on the NETCONF and YANG standards.


                                              Tail-f Systems © 2010
                                                   Page 10 of 11
Tail-f Systems Whitepaper

Tail-f Systems has implemented the concepts behind Dynamic Service Configuration in a family of
configuration management products and tools. For more information on Tail-f Systems’ order fulfillment and
configuration solutions, please contact us at mailto:info@tail-f.com.

Tail-f Systems is headquartered in Stockholm, Sweden. For more information on the company, please visit
http://www.tail-f.com.

Further Reading

    1. Martin Bjorklund. Why YANG. http://www.yang-central.org/twiki/bin/view/Main/WhyYang
    2. W. Enck, P. McDaniel, S. Sen, P. Sebos, S. Spoerel, A. Greenberg, S. Rao, and W. Aiello.
        Configuration management at massive scale: System design and experience, Pages 73–86, 2007.
    3. R. Enns. NETCONF Configuration Protocol. RFC 4741 (Proposed Standard), December 2006.
    4. TM Forum. eTom, http://www.tmforum.org/browse.aspx, TM Forum, 2005.
    5. Balasz Lengyel. www.yang-central.org/twiki/pub/Main/YangDocuments/draft-lengyel-why-yang-
        00.txt.
    6. ITIL. http://www.itil-officialsite.com.
    7. Stefan Wallin and Viktor Leijon. Telecom network and service management: An operator survey. In
        12th IFIP/IEEE International Conference on Management of Multimedia and Mobile Networks and
        Services (MMNS), October 2009.
    8. H. Xu and D. Xiao. Considerations on NETCONF-Based Data Modeling. In Proceedings of the 11th
        Asia-Pacific Symposium on Network Operations and Management: Challenges for Next Generation
        Network Operations and Service Management, 2008.
    9. Schonwalder, J., Pras, A., and Martin-Flatin, J.P. On the Future of Internet Management
        Technologies. IEEE Communications Magazine, vol 41, pages 90- 97
    10. Phil Shafer, Juniper. An NETCONF- and NETMOD-based Architecture for Network Management.
        http://www.ietf.org/id/draft-ietf-netmod-arch-05.txt.




                                            Tail-f Systems © 2010
                                                 Page 11 of 11

Contenu connexe

En vedette

NCS: NEtwork Control System Hands-on Labs
NCS:  NEtwork Control System Hands-on Labs NCS:  NEtwork Control System Hands-on Labs
NCS: NEtwork Control System Hands-on Labs Cisco Canada
 
NSO Introduction
NSO IntroductionNSO Introduction
NSO IntroductionJunho Lee
 
Tail-f Webinar OpenFlow Switch Management Using NETCONF and YANG
Tail-f Webinar OpenFlow Switch Management Using NETCONF and YANGTail-f Webinar OpenFlow Switch Management Using NETCONF and YANG
Tail-f Webinar OpenFlow Switch Management Using NETCONF and YANGTail-f Systems
 
Module 6: YANG Tutorial - part 2
Module 6: YANG Tutorial - part 2Module 6: YANG Tutorial - part 2
Module 6: YANG Tutorial - part 2Tail-f Systems
 
Dynamic Service Chaining
Dynamic Service Chaining Dynamic Service Chaining
Dynamic Service Chaining Tail-f Systems
 
Module 3: NETCONF and YANG Concepts
Module 3: NETCONF and YANG ConceptsModule 3: NETCONF and YANG Concepts
Module 3: NETCONF and YANG ConceptsTail-f Systems
 
Module 5: YANG Tutorial - part 1
Module 5: YANG Tutorial - part 1Module 5: YANG Tutorial - part 1
Module 5: YANG Tutorial - part 1Tail-f Systems
 
Module 1: ConfD Technical Introduction
Module 1: ConfD Technical IntroductionModule 1: ConfD Technical Introduction
Module 1: ConfD Technical IntroductionTail-f Systems
 
Module 2: Why NETCONF and YANG
Module 2: Why NETCONF and YANGModule 2: Why NETCONF and YANG
Module 2: Why NETCONF and YANGTail-f Systems
 
Module 4: NETCONF Tutorial
Module 4: NETCONF Tutorial Module 4: NETCONF Tutorial
Module 4: NETCONF Tutorial Tail-f Systems
 

En vedette (12)

NCS: NEtwork Control System Hands-on Labs
NCS:  NEtwork Control System Hands-on Labs NCS:  NEtwork Control System Hands-on Labs
NCS: NEtwork Control System Hands-on Labs
 
Tail-f - Why NETCONF
Tail-f - Why NETCONFTail-f - Why NETCONF
Tail-f - Why NETCONF
 
NSO Introduction
NSO IntroductionNSO Introduction
NSO Introduction
 
Tail-f Webinar OpenFlow Switch Management Using NETCONF and YANG
Tail-f Webinar OpenFlow Switch Management Using NETCONF and YANGTail-f Webinar OpenFlow Switch Management Using NETCONF and YANG
Tail-f Webinar OpenFlow Switch Management Using NETCONF and YANG
 
Module 6: YANG Tutorial - part 2
Module 6: YANG Tutorial - part 2Module 6: YANG Tutorial - part 2
Module 6: YANG Tutorial - part 2
 
Dynamic Service Chaining
Dynamic Service Chaining Dynamic Service Chaining
Dynamic Service Chaining
 
Module 3: NETCONF and YANG Concepts
Module 3: NETCONF and YANG ConceptsModule 3: NETCONF and YANG Concepts
Module 3: NETCONF and YANG Concepts
 
Module 5: YANG Tutorial - part 1
Module 5: YANG Tutorial - part 1Module 5: YANG Tutorial - part 1
Module 5: YANG Tutorial - part 1
 
Module 1: ConfD Technical Introduction
Module 1: ConfD Technical IntroductionModule 1: ConfD Technical Introduction
Module 1: ConfD Technical Introduction
 
Module 2: Why NETCONF and YANG
Module 2: Why NETCONF and YANGModule 2: Why NETCONF and YANG
Module 2: Why NETCONF and YANG
 
Module 4: NETCONF Tutorial
Module 4: NETCONF Tutorial Module 4: NETCONF Tutorial
Module 4: NETCONF Tutorial
 
NETCONF YANG tutorial
NETCONF YANG tutorialNETCONF YANG tutorial
NETCONF YANG tutorial
 

Dernier

Generative AI - Gitex v1Generative AI - Gitex v1.pptx
Generative AI - Gitex v1Generative AI - Gitex v1.pptxGenerative AI - Gitex v1Generative AI - Gitex v1.pptx
Generative AI - Gitex v1Generative AI - Gitex v1.pptxfnnc6jmgwh
 
2024 April Patch Tuesday
2024 April Patch Tuesday2024 April Patch Tuesday
2024 April Patch TuesdayIvanti
 
How AI, OpenAI, and ChatGPT impact business and software.
How AI, OpenAI, and ChatGPT impact business and software.How AI, OpenAI, and ChatGPT impact business and software.
How AI, OpenAI, and ChatGPT impact business and software.Curtis Poe
 
Emixa Mendix Meetup 11 April 2024 about Mendix Native development
Emixa Mendix Meetup 11 April 2024 about Mendix Native developmentEmixa Mendix Meetup 11 April 2024 about Mendix Native development
Emixa Mendix Meetup 11 April 2024 about Mendix Native developmentPim van der Noll
 
Digital Identity is Under Attack: FIDO Paris Seminar.pptx
Digital Identity is Under Attack: FIDO Paris Seminar.pptxDigital Identity is Under Attack: FIDO Paris Seminar.pptx
Digital Identity is Under Attack: FIDO Paris Seminar.pptxLoriGlavin3
 
Bridging Between CAD & GIS: 6 Ways to Automate Your Data Integration
Bridging Between CAD & GIS:  6 Ways to Automate Your Data IntegrationBridging Between CAD & GIS:  6 Ways to Automate Your Data Integration
Bridging Between CAD & GIS: 6 Ways to Automate Your Data Integrationmarketing932765
 
The Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptx
The Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptxThe Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptx
The Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptxLoriGlavin3
 
Modern Roaming for Notes and Nomad – Cheaper Faster Better Stronger
Modern Roaming for Notes and Nomad – Cheaper Faster Better StrongerModern Roaming for Notes and Nomad – Cheaper Faster Better Stronger
Modern Roaming for Notes and Nomad – Cheaper Faster Better Strongerpanagenda
 
TeamStation AI System Report LATAM IT Salaries 2024
TeamStation AI System Report LATAM IT Salaries 2024TeamStation AI System Report LATAM IT Salaries 2024
TeamStation AI System Report LATAM IT Salaries 2024Lonnie McRorey
 
Time Series Foundation Models - current state and future directions
Time Series Foundation Models - current state and future directionsTime Series Foundation Models - current state and future directions
Time Series Foundation Models - current state and future directionsNathaniel Shimoni
 
Microsoft 365 Copilot: How to boost your productivity with AI – Part one: Ado...
Microsoft 365 Copilot: How to boost your productivity with AI – Part one: Ado...Microsoft 365 Copilot: How to boost your productivity with AI – Part one: Ado...
Microsoft 365 Copilot: How to boost your productivity with AI – Part one: Ado...Nikki Chapple
 
Passkey Providers and Enabling Portability: FIDO Paris Seminar.pptx
Passkey Providers and Enabling Portability: FIDO Paris Seminar.pptxPasskey Providers and Enabling Portability: FIDO Paris Seminar.pptx
Passkey Providers and Enabling Portability: FIDO Paris Seminar.pptxLoriGlavin3
 
The Future Roadmap for the Composable Data Stack - Wes McKinney - Data Counci...
The Future Roadmap for the Composable Data Stack - Wes McKinney - Data Counci...The Future Roadmap for the Composable Data Stack - Wes McKinney - Data Counci...
The Future Roadmap for the Composable Data Stack - Wes McKinney - Data Counci...Wes McKinney
 
Decarbonising Buildings: Making a net-zero built environment a reality
Decarbonising Buildings: Making a net-zero built environment a realityDecarbonising Buildings: Making a net-zero built environment a reality
Decarbonising Buildings: Making a net-zero built environment a realityIES VE
 
Moving Beyond Passwords: FIDO Paris Seminar.pdf
Moving Beyond Passwords: FIDO Paris Seminar.pdfMoving Beyond Passwords: FIDO Paris Seminar.pdf
Moving Beyond Passwords: FIDO Paris Seminar.pdfLoriGlavin3
 
QCon London: Mastering long-running processes in modern architectures
QCon London: Mastering long-running processes in modern architecturesQCon London: Mastering long-running processes in modern architectures
QCon London: Mastering long-running processes in modern architecturesBernd Ruecker
 
Glenn Lazarus- Why Your Observability Strategy Needs Security Observability
Glenn Lazarus- Why Your Observability Strategy Needs Security ObservabilityGlenn Lazarus- Why Your Observability Strategy Needs Security Observability
Glenn Lazarus- Why Your Observability Strategy Needs Security Observabilityitnewsafrica
 
Genislab builds better products and faster go-to-market with Lean project man...
Genislab builds better products and faster go-to-market with Lean project man...Genislab builds better products and faster go-to-market with Lean project man...
Genislab builds better products and faster go-to-market with Lean project man...Farhan Tariq
 
Top 10 Hubspot Development Companies in 2024
Top 10 Hubspot Development Companies in 2024Top 10 Hubspot Development Companies in 2024
Top 10 Hubspot Development Companies in 2024TopCSSGallery
 
A Journey Into the Emotions of Software Developers
A Journey Into the Emotions of Software DevelopersA Journey Into the Emotions of Software Developers
A Journey Into the Emotions of Software DevelopersNicole Novielli
 

Dernier (20)

Generative AI - Gitex v1Generative AI - Gitex v1.pptx
Generative AI - Gitex v1Generative AI - Gitex v1.pptxGenerative AI - Gitex v1Generative AI - Gitex v1.pptx
Generative AI - Gitex v1Generative AI - Gitex v1.pptx
 
2024 April Patch Tuesday
2024 April Patch Tuesday2024 April Patch Tuesday
2024 April Patch Tuesday
 
How AI, OpenAI, and ChatGPT impact business and software.
How AI, OpenAI, and ChatGPT impact business and software.How AI, OpenAI, and ChatGPT impact business and software.
How AI, OpenAI, and ChatGPT impact business and software.
 
Emixa Mendix Meetup 11 April 2024 about Mendix Native development
Emixa Mendix Meetup 11 April 2024 about Mendix Native developmentEmixa Mendix Meetup 11 April 2024 about Mendix Native development
Emixa Mendix Meetup 11 April 2024 about Mendix Native development
 
Digital Identity is Under Attack: FIDO Paris Seminar.pptx
Digital Identity is Under Attack: FIDO Paris Seminar.pptxDigital Identity is Under Attack: FIDO Paris Seminar.pptx
Digital Identity is Under Attack: FIDO Paris Seminar.pptx
 
Bridging Between CAD & GIS: 6 Ways to Automate Your Data Integration
Bridging Between CAD & GIS:  6 Ways to Automate Your Data IntegrationBridging Between CAD & GIS:  6 Ways to Automate Your Data Integration
Bridging Between CAD & GIS: 6 Ways to Automate Your Data Integration
 
The Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptx
The Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptxThe Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptx
The Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptx
 
Modern Roaming for Notes and Nomad – Cheaper Faster Better Stronger
Modern Roaming for Notes and Nomad – Cheaper Faster Better StrongerModern Roaming for Notes and Nomad – Cheaper Faster Better Stronger
Modern Roaming for Notes and Nomad – Cheaper Faster Better Stronger
 
TeamStation AI System Report LATAM IT Salaries 2024
TeamStation AI System Report LATAM IT Salaries 2024TeamStation AI System Report LATAM IT Salaries 2024
TeamStation AI System Report LATAM IT Salaries 2024
 
Time Series Foundation Models - current state and future directions
Time Series Foundation Models - current state and future directionsTime Series Foundation Models - current state and future directions
Time Series Foundation Models - current state and future directions
 
Microsoft 365 Copilot: How to boost your productivity with AI – Part one: Ado...
Microsoft 365 Copilot: How to boost your productivity with AI – Part one: Ado...Microsoft 365 Copilot: How to boost your productivity with AI – Part one: Ado...
Microsoft 365 Copilot: How to boost your productivity with AI – Part one: Ado...
 
Passkey Providers and Enabling Portability: FIDO Paris Seminar.pptx
Passkey Providers and Enabling Portability: FIDO Paris Seminar.pptxPasskey Providers and Enabling Portability: FIDO Paris Seminar.pptx
Passkey Providers and Enabling Portability: FIDO Paris Seminar.pptx
 
The Future Roadmap for the Composable Data Stack - Wes McKinney - Data Counci...
The Future Roadmap for the Composable Data Stack - Wes McKinney - Data Counci...The Future Roadmap for the Composable Data Stack - Wes McKinney - Data Counci...
The Future Roadmap for the Composable Data Stack - Wes McKinney - Data Counci...
 
Decarbonising Buildings: Making a net-zero built environment a reality
Decarbonising Buildings: Making a net-zero built environment a realityDecarbonising Buildings: Making a net-zero built environment a reality
Decarbonising Buildings: Making a net-zero built environment a reality
 
Moving Beyond Passwords: FIDO Paris Seminar.pdf
Moving Beyond Passwords: FIDO Paris Seminar.pdfMoving Beyond Passwords: FIDO Paris Seminar.pdf
Moving Beyond Passwords: FIDO Paris Seminar.pdf
 
QCon London: Mastering long-running processes in modern architectures
QCon London: Mastering long-running processes in modern architecturesQCon London: Mastering long-running processes in modern architectures
QCon London: Mastering long-running processes in modern architectures
 
Glenn Lazarus- Why Your Observability Strategy Needs Security Observability
Glenn Lazarus- Why Your Observability Strategy Needs Security ObservabilityGlenn Lazarus- Why Your Observability Strategy Needs Security Observability
Glenn Lazarus- Why Your Observability Strategy Needs Security Observability
 
Genislab builds better products and faster go-to-market with Lean project man...
Genislab builds better products and faster go-to-market with Lean project man...Genislab builds better products and faster go-to-market with Lean project man...
Genislab builds better products and faster go-to-market with Lean project man...
 
Top 10 Hubspot Development Companies in 2024
Top 10 Hubspot Development Companies in 2024Top 10 Hubspot Development Companies in 2024
Top 10 Hubspot Development Companies in 2024
 
A Journey Into the Emotions of Software Developers
A Journey Into the Emotions of Software DevelopersA Journey Into the Emotions of Software Developers
A Journey Into the Emotions of Software Developers
 

Tail-f Systems Whitepaper - Configuration Management and Service Provisioning

  • 1. Tail-f Systems Whitepaper Closing the Services Configuration Gap Introduction Delivering valued-added services, like MPLS VPNS, Metro Ethernet, and IP TV is critical to the profitability and growth of service providers. In most operator networks, there are bottlenecks that get in the way of activating services in a timely and cost effective manner. We call this the “Services Configuration Gap.” Current systems use either legacy software or third party applications that are expensive and inflexible, often requiring manual intervention, frequent software updates, and maintenance of layers of mediation software. These systems are being strained by the increased rate of change as services are enhanced, new services are introduced, and network devices are replaced and added. Both OSS and EMS developers are burdened with these problems. For network equipment vendors, developing EMS applications, the focus is more on device adaptations than service applications. This whitepaper describes the Services Configuration Gap and proposes a new generation of configuration management tools to deal with the fast changing environment faced by service providers and OSS and EMS developers. The proposed approach leverages dynamic data modeling, automated configuration, and mechanisms for transaction integrity. By automating the connection between service activation and device configuration, the excess cost and delay associated with hardwired, service-specific software is eliminated. Service Configuration Today The steps involved in creating and activating a service include many processes, support systems, and network devices. As shown below in Figure 1, the TM Forum’s eTOM Process Map divides the processes into two major areas: strategy (left) and operations (right). Figure 1: The eTOM Process Map The strategy product lifecycle addresses service introduction and retirement. The design of new services includes how the service is represented in systems like customer care, assurance, and billing and requires services to be modeled so that all systems can have a common view of their characteristics and decomposition. Tail-f Systems © 2010 Page 1 of 11
  • 2. Tail-f Systems Whitepaper The design and modeling includes the mapping to the resource layer, as well as defining what devices and configurations are needed to support a service. When a service has been designed it is handed over to operations processes which are responsible for activating the service in the network. The activation process which includes assurance and billing is shown in Figure 2. For example, if a customer orders an MPLS VPN, the order handling process makes sure that the order is properly managed including credit authorization and overall order tracking. The order process then triggers the service configuration and activation process which in turn configures the customer service. The process might include manual steps such as the installation of Figure 2: eTOM Service Activation customer equipment. Service models are commonly expressed in UML models and implemented in vendor-specific repositories at the OSS level. The device data models are typically available as SNMP MIBs and vendor-specific CLIs. Challenges with Existing Service Configuration Approaches Service Models The Services Configuration Gap is a function of the way existing service models and their mapping processes connect to the corresponding device data models. Due to the rapid change of customer requirements, adoption of new services, and a constantly changing network topology, these models and mappings must be frequently revised. Today this process is semi-automated and not dynamic, resulting in changes often being reflected throughout the system in weeks rather than hours. To reach new time-to-market expectations, the next generation of service management tools must have the following capabilities: • Be model-driven and allow user and programmatic interfaces to be automatically updated from the service model without other inputs. • Support model-aware APIs that let programmers work directly with classes from the service model rather than low-level protocol based APIs. • Ability to load new versions and models at runtime and propagate changes immediately. • Modeling tools and support that allow for quick changes while still validating the model correctness and completeness. • The model must interface seamlessly with the subsequent mapping step to ensure that the complete chain from service model to device model is managed. • Modern software engineering approaches including modularity, aspects, and annotations (rather than inheritance) should be used to avoid dependencies and long build times. Tail-f Systems © 2010 Page 2 of 11
  • 3. Tail-f Systems Whitepaper Device Management The process of configuring network devices is generally a manual and time-consuming task. Unlike alarm and performance management, configuration activities involve writing data and changing states on devices whereas alarm and performance management is mostly a read-only activity. Furthermore, the configuration process also has more complex operation requirements such as validation and transactions. An additional challenge is the fact that configuration data models are vendor-specific and often hard to understand. While network configuration in large networks will always be demanding and need skilled network administrators, this should not prevent the use of automated configuration management at the "plumbing level.” Currently, most configuration management operations are performed using scripts on top of command line interfaces or even partly with SNMP. The use of CLIs and SNMP comes with major limitations as discussed below. SNMP MIBS • It is hard to find the correct MIB Modules using SNMP. There is no defined discovery procedure other than an ad hoc SNMP walk and, even after a MIB walk, there is no formal and automated way to find the correct MIB module. • All MIBs do not define fundamental concepts, including which data is configuration data versus operational data, sequencing and transactional dependencies, or constraints between variables. • The module structure of SNMP MIBs has worked well in guaranteeing unique identifiers. However, the administrative based tree-structure does not suit automated rendering of interfaces or favor readability and extensibility in a logical way. CLI Models • The data-model for CLIs is informally defined and is managed by scripts that parse the actual strings. This is a fragile way of working since misspellings and version changes break the scripts. • There is no room for automated rendering of tools since no formal model is defined. While improvements can be made to traditional solutions at considerable cost, the result falls short of a fully automated approach where the latter includes the following capabilities: • Discovery of data models. • Resolution of data model versions. • Rendering of interfaces from data models. • Synchronization of configuration data from and to devices. • Validations and transactions to ensure correct configuration. • Transactions that are tolerant of unreachable devices and capability variants. The Service Configuration Gap is a byproduct of inflexible, partially automated systems that struggle to keep up with the rate of change in a service operator’s environment. Symptoms of the Service Configuration Gap include: • Delays in the introduction of new services that result in the postponement of incremental revenue. • The need for large teams of custom programmers to update and maintain adapter software. • Configuration errors that result in loss of service and negatively impact customer satisfaction. Tail-f Systems © 2010 Page 3 of 11
  • 4. Tail-f Systems Whitepaper Dynamic Service Configuration The combination of dynamic service modeling and automated configuration is the basis of the next generation of configuration management tools and we call this new approach “Dynamic Service Configuration.” With Dynamic Service Configuration operators are able to dynamically adapt the service configuration solution according to changes in their service portfolio, rather than being based on raw device configurations. The configuration management tools that enable Dynamic Service Configuration sit between the network devices and the order management system. As shown in Figure 3, with this new approach there will be a Service Layer and a Device Layer. The Service Layer supports dynamic modeling of services with short development time and model-driven dynamic updates of dependant interfaces. The Device Layer automatically manages the device interfaces, such as synchronization of configuration data, multinode configuration deployment, and interface versioning. In the context of the eTOM processes (see Figure 2), Dynamic Service Configuration implements Service Configuration and Activation and Resource Provisioning. Many operators have systems in place for managing work-flows across the complete OSS chain, our proposed approach sits below such systems and provides an efficient network facing transaction engine. Furthermore, it replaces in-house CLI scripts or expensive adaptors and provides a dynamic and Figure 3: Dynamic Service Configuration managed solution for such environments. The input to the configuration management system is a request to change or add a service, such as a NGOSS Contract or an MTOSI request. The output from the system is a validated network change, an alert that validation has failed, or the proposed change was rejected due to some other reason. Dynamic Service Configuration uses the NETCONF protocol and YANG data modeling as foundations. YANG is used to bridge the higher level models to device interfaces and NETCONF is used as the means of deploying the configuration to the devices. This approach can benefit from many automated features such as model and transaction management where devices support these standards directly. Non-NETCONF devices such as SNMP and CLI managed network elements can also be mapped by using a mediator and still benefit from the NETCONF design principle with support of transactions and model-based management. NETCONF and YANG are described in more detail below. Tail-f Systems © 2010 Page 4 of 11
  • 5. Tail-f Systems Whitepaper NETCONF The Network Configuration Protocol, NETCONF, is an IETF network management protocol and is published in RFC 4741. NETCONF is being adopted by major network equipment providers and has gained strong industry support. NETCONF provides mechanisms to install, manipulate, and delete the configuration of network devices. Its operations are realized on top of a simple Remote Procedure Call (RPC) layer. The NETCONF protocol uses XML based data encoding for the configuration data as well as the protocol messages. NETCONF is designed to be a replacement for CLI based programmatic interfaces, such as Perl + Expect over Secure Shell (SSH). NETCONF is usually transported over the SSH protocol, using the ’NETCONF’ sub-system and in many ways it mimics native proprietary CLI over SSH interfaces available in a device. However, it uses structured schema-driven data and provides detailed structured error return information, which the CLI cannot provide. Operators need a way to distribute changes to the devices and validate them locally before activating them. This is indicated by the two bottom options in Figure 4 where configuration data can be sent to candidate databases in the devices before they are committed to running in production applications. All NETCONF devices must allow the configuration data to be locked, edited, saved, and unlocked. In addition, all modifications to the configuration data must be saved in non-volatile storage. Figure 4: NETCONF Configuration Databases YANG YANG is a data modeling language that provides breakthrough advantages to modeling configuration and state data. The YANG modeling language is currently being standardized by the IETF in the NETMOD working group. The authors of YANG were involved in the development of the next generation SNMP SMI and are very experienced in the IETF network management arena. Modern network standards groups such as the Metro Ethernet Forum are adopting YANG as a device configuration standard. Tail-f Systems © 2010 Page 5 of 11
  • 6. Tail-f Systems Whitepaper YANG is tree-structured rather than object-oriented. Configuration data is structured into a tree and the data can include complex types such as lists and unions. The definitions are contained in modules and one module can augment the tree in another module. Strong revision rules are defined for modules. YANG also differs from previous network management data model languages through its strong support of constraints and data validation rules. The advantages of using YANG for data models include: • YANG is simple, easy to learn, and most important of all, it is easy to read and understand. The complexity of other languages has made it hard to review information models and find problems, thereby undermining the ability of developers to build management applications. • YANG is written for network management. It is a domain specific language designed to do what it needs to do well. It is not constrained by or overloaded by any framework it is derived from. • YANG is flexible and modular. Existing modules can be augmented in a controlled manner. An Example of Dynamic Service Configuration in Action The following is an illustration of Dynamic Service Configuration in action using YANG as a general modeling language and showing the power of dynamically modeling services and automated configuration of network devices. We are using the scenario of activating MPLS VPN for customers. The hypothetical network topology for this example is shown in Figure 5 below. Figure 5: Network Topology A VPN contains customer devices attached to the Customer Edge (CE) routers. The customer devices use a MPLS VPN to exchange information between devices within the VPN. The CE routers are not aware of the VPN as this is managed by the Provider Edge (PE) routers. At service fulfillment time, network resources are allocated in the required geographic sites to suit the customer. Having selected the devices to be used, the configuration system can build the required resource facing service template and then populate it with the customer specific instance details. The service and device model for this example is shown in Figure 6 below. Tail-f Systems © 2010 Page 6 of 11
  • 7. Tail-f Systems Whitepaper Figure 6: Service and Device Model The classes at the top of Figure 6 are standard classes in the Service and Device Layers. The Device Class is automatically augmented with the YANG modules that are discovered from the device (in this case, with PE and CE classes). The CE router, being VPN agnostic, contains a simple model for IP interfaces while the PE routers model MPLS information in addition to standard IP interfaces. The SID standard Resource and Customer Facing Service Classes are augmented as part of a dynamic service modeling activity. The Customer facing Service Class is augmented with the Customer VPN service and the Resource facing service with the MPLS VPN. The Customer VPN service represents what the customer bought, a VPN service connecting the points of presence. The network facing service (MPLS VPN) is the implementation of the customer service using MPLS technology and it refers to the access links connecting the CE with the PE. The separation of customer facing services versus the resource facing services is an important modeling pattern inherited from the TM Forum SID model. Next, a fully meshed MPLS VPN “RED” is created. The core network is assumed to be configured (PE routers). This scenario involves the following steps: • Setup connectivity PE1-CE1 and PE2-CE3 • Creation of the VPN RED resource facing service instance “vpn-red” and its access links “northwest” (PE1-CE1) “southeast” (PE2-CE3). • Creation of the customer facing service instance “acme-vpn. • Addition of “acme-vpn” to customer “acme”. This model is depicted in Figure 7 below. Tail-f Systems © 2010 Page 7 of 11
  • 8. Tail-f Systems Whitepaper Figure 7: Meshed MPLS VPN First, the CE interfaces on the PE routers facing VPN RED service are created as shown in Listing 1 below. ncscli >set ncs managed-device pe1 config interface eth1 enabled ip 192.168.7.2 mask 255.255.255.0 ncscli >set ncs managed-device pe2 config interface eth3 enabled ip 192.168.6.2 mask 255.255.255.0 ncscli>commit Listing 1: Create Service in CLI Note the commit command at the end. This will result in a multi-node transaction towards two PE routers. Either the complete transaction succeeds or fails and in the latter case can be roll-backed. The next step is to create the mpls vpn “vpn-red” with its access-links “north-west” and “south-east”: ncscli> set ncs service vpn-red service-type mplsVpn access-link north-west ce ce1 ce-if eth0 pe pe1 pe- if eth1 subnet 192.168.7.0/24 Value for 'route-distinguisher' (<string, max: 16 chars, min: 3 chars>): 300:1 Value for 'ncs service VPN_GREEN properties mplsVpn vpn-id' (<RFC 2685>): 300:1 ncscli>set ncs service vpn-red service-type mplsVpn access-link south-east ce ce3 ce-if eth0 pe pe2 pe-if eth3 subnet 192.168.6.0/24 ncscli>commit Listing 2: Create Resource Facing MPLS Service in CLI The constraints expressed in the model will ensure that only references are made to valid objects for the access-link. The “Value for” prompts are automatically generated from the model and the CLI asks for these attributes since they are mandatory. While not possible to illustrate here, the automatically rendered CLI will only accept keyboard entries according to the data type, so for example only digits and dots in the right places are allowed when the IP address is entered. If for some reason the network changes are undesired, a rollback “undo” can easily be performed. Next the customer facing VPN service “acme-vpn” is created as shown in Listing 3 below. The command refers to the existing customer “acme”. Tail-f Systems © 2010 Page 8 of 11
  • 9. Tail-f Systems Whitepaper ncscli> set ncs service acme-vpn service-type custVpn mplsVpn vpn-red vpnCust acme ncscli> commit Listing 3: Create Customer Facing VPN in CLI This refers to the VPN service “vpn-red” allocated previously and to the customer “acme” that needs to be associated with the service resource. The customer is defined in an external system and queried for validity at creation time. The above examples of modeling and configuration have been illustrated with CLI commands, but this could have equally been achieved with a Web Interface dynamically generated by the model. Mapping Service Models When a customer wants to add access to their VPN, the Service Logic needs to define the new CE, allocate the required PE port, and define what VPN types are needed to be configured. Our approach is to define the Customer Service in a YANG model so that the service fulfillment is simply the mapping of the Service Layer model to the underlying Device Layer models. If a device does not support NETCONF, a mediator embeds transactional mapping logic which resolves many repetitive and error-prone transformation tasks. The two main interface categories for OSS systems, such as inventory systems or other network management systems, are SQL databases or APIs. In these cases, a YANG view of the required data is created and mapped to the external database or API. From the perspective of the configuration management application, it looks as though the data resides within the application while every access is forwarded to the external system. This approach gives the following benefits to model mapping: • Allows for federation as an approach to data integration. • The service logic is always a YANG to YANG mapping procedure. No service mapping logic is contained in adaptors to other interfaces. • It is possible to provide solutions to hard problems like transactions, calculating delta operations, validation of data, and pre-provisioning. • Device configuration integration is reduced to a minimum for NETCONF devices. Rather than requiring the implementer to write code for every model attribute and operation, the definition of a single create template is all that is required. All other operations like sets and deletes are automatically calculated. In coding terms, this means 1,000 lines of code can be collapsed to 100 lines, reducing service activation lead times, programming, and maintenance costs. Implementing Dynamic Service Configuration There are many implementation details that will be required to make systems using Dynamic Service Configuration effectively integrate with existing systems and operations. Examples of these implementation details include: • Capability to automatically discover the supported device modules and resolve transactions across different versions. • Managing the propagation of changes from service to device models. When a service is changed, the corresponding device changes are managed in a separate transaction. These changes can be inspected and used for reverse operation calculations. Tail-f Systems © 2010 Page 9 of 11
  • 10. Tail-f Systems Whitepaper • Pre-provisioning – used when the configuration management systems caches the operations for network elements not installed. In normal circumstances, transactions would fail because a device does not respond, but with this capability the missing commands are executed whenever devices are available. • Backlogging – used when changes to the network must be made as soon as possible and the transaction should succeed even if some devices are not responding. Here operations to non- responsive devices are cached. • Capability to enforce all service models to implement self-test functions. For example, in the MPLS scenario the MPLS VPN will perform a MPLS ping to verify the connectivity. Actual test instrumentation may be available on the devices themselves or in external probes. • Ability to restore a broken service. A typical scenario is where technicians modify devices manually without being aware of negative effects on services. • Replay function to show the differences between elements with inconsistent configuration parameters. This is used to see which devices are out of synch versus the desired configuration and the corresponding audit log. Conclusions To effectively deliver new services in a rapidly changing environment, service providers cannot continue to rely on existing configuration management systems. Current systems use either legacy software or third party applications that are expensive, inflexible, require manual intervention, and frequent software updates plus maintenance of layers of mediation software. These systems are being strained by the increased rate of change as existing services are enhanced, new services are introduced, and network devices are replaced and added. As a result, there is a Services Configuration Gap that can be measured in terms of delays in introducing new services, lost revenue, and excess software maintenance costs. A new approach to configuration management is proposed called Dynamic Service Configuration. Dynamic Service Configuration leverages the NETCONF and YANG standards and is based on dynamic data modeling, automated configuration, and transaction integrity. From a software implementation point of view, 1,000 lines of code can be collapsed to 100 lines, reducing service activation lead times, programming, and maintenance costs. About Tail-f Systems Tail-f Systems is a leading developer of configuration management software for networking equipment providers, EMS providers, and OSS development organizations. Six of the ten largest global networking equipment providers are Tail-f Systems’ customers. Tail-f Systems provides a full family of products to build and extend network management systems. ConfD, the company’s on-device configuration management toolkit, renders management interfaces from a single data model and includes a lightweight embedded database. Customers using ConfD radically reduce their time-to-market and benefit from carrier-grade implementations of NETCONF, CLI, Web, and SNMP interfaces. Active in standards development organizations, Tail-f Systems is the leading provider of configuration management solutions based on the NETCONF and YANG standards. Tail-f Systems © 2010 Page 10 of 11
  • 11. Tail-f Systems Whitepaper Tail-f Systems has implemented the concepts behind Dynamic Service Configuration in a family of configuration management products and tools. For more information on Tail-f Systems’ order fulfillment and configuration solutions, please contact us at mailto:info@tail-f.com. Tail-f Systems is headquartered in Stockholm, Sweden. For more information on the company, please visit http://www.tail-f.com. Further Reading 1. Martin Bjorklund. Why YANG. http://www.yang-central.org/twiki/bin/view/Main/WhyYang 2. W. Enck, P. McDaniel, S. Sen, P. Sebos, S. Spoerel, A. Greenberg, S. Rao, and W. Aiello. Configuration management at massive scale: System design and experience, Pages 73–86, 2007. 3. R. Enns. NETCONF Configuration Protocol. RFC 4741 (Proposed Standard), December 2006. 4. TM Forum. eTom, http://www.tmforum.org/browse.aspx, TM Forum, 2005. 5. Balasz Lengyel. www.yang-central.org/twiki/pub/Main/YangDocuments/draft-lengyel-why-yang- 00.txt. 6. ITIL. http://www.itil-officialsite.com. 7. Stefan Wallin and Viktor Leijon. Telecom network and service management: An operator survey. In 12th IFIP/IEEE International Conference on Management of Multimedia and Mobile Networks and Services (MMNS), October 2009. 8. H. Xu and D. Xiao. Considerations on NETCONF-Based Data Modeling. In Proceedings of the 11th Asia-Pacific Symposium on Network Operations and Management: Challenges for Next Generation Network Operations and Service Management, 2008. 9. Schonwalder, J., Pras, A., and Martin-Flatin, J.P. On the Future of Internet Management Technologies. IEEE Communications Magazine, vol 41, pages 90- 97 10. Phil Shafer, Juniper. An NETCONF- and NETMOD-based Architecture for Network Management. http://www.ietf.org/id/draft-ietf-netmod-arch-05.txt. Tail-f Systems © 2010 Page 11 of 11