Internet DRAFT - draft-klee-teas-actn-connectivity-multi-domain
draft-klee-teas-actn-connectivity-multi-domain
Network Working Group Kwang-koog Lee
Internet Draft Hosong Lee
Intended status: Informational KT
Expires January 2018 Ricard Vilalta
CTTC
Victor Lopez
Telefonica
July 31, 2017
ACTN use-case for E2E network services in multiple vendor domain
transport networks
draft-klee-teas-actn-connectivity-multi-domain-03.txt
Status of this Memo
This Internet-Draft is submitted to IETF in full conformance with
the provisions of BCP 78 and BCP 79.
Internet-Drafts are working documents of the Internet Engineering
Task Force (IETF), its areas, and its working groups. Note that
other groups may also distribute working documents as Internet-
Drafts.
Internet-Drafts are draft documents valid for a maximum of six
months and may be updated, replaced, or obsoleted by other documents
at any time. It is inappropriate to use Internet-Drafts as
reference material or to cite them other than as "work in progress."
The list of current Internet-Drafts can be accessed at
http://www.ietf.org/ietf/1id-abstracts.txt
The list of Internet-Draft Shadow Directories can be accessed at
http://www.ietf.org/shadow.html.
This Internet-Draft will expire on January 31, 2018.
Copyright Notice
Copyright (c) 2014 IETF Trust and the persons identified as the
document authors. All rights reserved.
This document is subject to BCP 78 and the IETF Trust's Legal
Provisions Relating to IETF Documents
(http://trustee.ietf.org/license-info) in effect on the date of
publication of this document. Please review these documents
K. Lee & H. Lee Expires January 31, 2018 [Page 1]
Internet-Draft ACTN Multiple Vendor Domains July 2017
carefully, as they describe your rights and restrictions with
respect to this document.
Abstract
This document provides a use-case that addresses the need for
facilitating the application of virtual network abstractions and the
control and management of end-to-end network services that traverse
multiple vendor domain transport networks.
These abstractions shall help create a virtualized environment
supporting operators in viewing and controlling different vendor
domains, especially for end-to-end network connectivity service for
a single operator.
Table of Contents
1. Introduction...................................................2
2. End-to-End Network Services in Multi-vendor Domain Transport
Networks..........................................................3
2.1. Example A - Leased Line Services..........................5
2.2. Example B - Data Center Interconnect (DCI)................6
3. Requirements...................................................7
4. References....................................................10
5. Acknowledgement...............................................10
6. Contributors..................................................11
1. Introduction
Network operators build and operate their network using multiple
domains (i.e., core and access networks) in different dimensions.
Domains may be defined by a collection of links and nodes (each of a
different technology), administrative zones under the concern of a
particular business entity, or vendor-specific "islands" where
specific control mechanisms have to be applied. Due to the
technology of each vendor, the optical components cannot be
interconnected. Even, the interconnection of components supporting
the same technology is not easy since their control and management
system is tightly coupled with their own devices. Therefore each
optical domain becomes an isolated island in terms of the control
and management of end-to-end services traversing multiple domains.
The network operators use vendor-specific NMS implementations along
with an operator-tailored umbrella provisioning system, which may
include a technology specific Operations Support System (OSS).
Thanks to the evolution of vendor specific SDN controllers, the
K. Lee & H. Lee Expires January 31, 2018 [Page 2]
Internet-Draft ACTN Multiple Vendor Domains July 2017
network operators require a network entity, which abstracts the
details of the optical layer while enabling control and management
of the end-to-end services traversing multiple domains. The
establishment of end-to-end connections spanning several of these
domains is a perpetual problem for operators, which need to address
both interoperability and operational concerns at the control and
data planes.
The introduction of new services, often requiring connections that
traverse multiple domains, needs significant planning, and several
manual operations to interface multiple vendor-specific domains in
which specific control/management mechanisms of the vendor equipment
have to be applied (e.g., EMS/NMS, OSS/BSS, control plane, SDN
controller, etc.). Undoubtedly, establishing an on-demand end-to-end
connection which requires provisioning based on dynamic resource
information is more difficult in the current network context.
This document provides a use-case that addresses the need for
creating a virtualized environment supporting operators in viewing
and controlling different vendor domains, especially for end-to-end
network services for a single operator. Surely, the use-case could
be also available for the interconnection of end-to-end services for
multiple operators. This will accelerate rapid deployment of new
services, including more dynamic and elastic services, and improve
overall network operations and scaling of existing services.
This use-case is a part of the overarching work, called Abstraction
and Control of Transport Networks (ACTN). Related documents are the
ACTN Requirements [ACTN-Req], ACTN-framework [ACTN-Frame] and the
problem statement [ACTN-PS].
2. End-to-End Network Services in Multi-vendor Domain Transport
Networks
This section provides an architecture example to illustrate the
context of the current challenges and issues operators face in
delivering end-to-end network services in operators' multi-vendor
domain transport networks.
K. Lee & H. Lee Expires January 31, 2018 [Page 3]
Internet-Draft ACTN Multiple Vendor Domains July 2017
|
| / End-to-End Connection \ |
|/-----------------------------------------------------------\|
|\-----------------------------------------------------------/|
| \ / |
| |
| +----------------+ |
| | | |
| | Converged | |
| | Packet-Optical| |
| +-------------+ | Core Domain | +-------------+ |
| | |--| (Vendor A) |--| | |
+----+ | Access | +----------------+ | Access | +----+
| CE1|--| Domain 1 | | | | Domain 3 |--| CE2|
+----+ | (Vendor B) |----- -----| (Vendor C) | +----+
+-------------+ +-------------+
Figure 1. Multi-vendor Domains
As an illustrative example, consider a multi-domain transport
network consisting of three domains: one core converged packet-
optical domain (Vendor A) and two access domains (Vendors B and C).
Each access domain is managed by its domain control/management
mechanism which is often a proprietary vendor-specific scheme. The
core domain is also managed by Vendor A's proprietary
control/management mechanism (e.g., EMS/NMS, OSS/BSS, Control Plane,
SDN Controller, or any combination of these entities, etc.) that may
not interoperate with access domain control/management mechanisms or
at best partially interoperate if Vendor A is same as Vendor B or
Vendor C.
Due to these domain boundaries, facilitating end-to-end connections
(e.g., Ethernet Virtual Connections, etc.) that traverse multi-
domains is not readily achieved. These domain controls are optimized
for its local operation and in most cases not suited for controlling
the end-to-end connectivity services. For instance, the discovery of
the edge nodes that belong to other domains is hard to achieve
partly because of the lack of the common API and its information
model and control mechanisms thereof to disseminate the relevant
information.
Moreover, the path computation for any on-demand end-to-end
connection would need abstraction of dynamic network resources and
ways to find an optimal path that meets the connection's service
requirements. This would require knowledge of both the domain level
dynamic network resource information and the inter-domain
K. Lee & H. Lee Expires January 31, 2018 [Page 4]
Internet-Draft ACTN Multiple Vendor Domains July 2017
connectivity information including domain gateway/peering points and
the local domain policy.
From an end-to-end connection provisioning perspective, in order to
facilitate a fast and reliable end-to-end signaling, each domain
operation and management elements should ideally speak the same
control protocols to its neighboring domains. However, this is not
possible for the current network context unless a folk-lift green
field technology deployment with a single vendor solution would be
done. Although each domain applies the same protocol for the data
plane, an end-to-end connectivity traversing multiple vendor domains
might not be provided due to a management and control mechanism
focusing only on its own domain.
In addition, the end-to-end connection provisioning via multiple
domains might not be treated by a single layer but the control and
management of multi-layer should be necessary. In this case, the
control plane of an upper layer first interacts with its lower layer
in order to check whether the lower layer can provide network
resources to meet service-level requirements such as SLA (Service
Level Agreement) and user-defined policies. Then, the upper layer
could proceed to provision its own layer referring to provisioning
results provided from its lower layer. However, vendor-specific
management systems have no awareness of adjacent layers of network
resources. Even, a single vendor management system covering multi-
layer avoids the interaction with control planes of the multi-layer
due to the complexity of implementation.
From a network connectivity management perspective, it would require
a mechanism to disseminate any connectivity issues from the local
domain to the other domains whenever the local domain cannot resolve
a connectivity issues. This is hard to achieve due to the lack of
the common API and its agreed-upon information model and control
mechanisms thereof to disseminate the relevant information.
From an operation's perspective, the current network environments
are not conducive to offering end-to-end connectivity services in
multi-vendor domain transport networks. For instance, when the
performance monitoring inquiry is requested, operators manually
monitor each domain and aggregate the performance results. However,
it may not be precise because of the different measurement timing
employed by each domain.
2.1. Example A - Leased Line Services
Service providers offer to customers a leased line service
connecting geographically distant two or more locations.
Traditionally, leased lines have been offered using SONET/SDH
networks in speed ranging from E1 to STM64 to meet specific business
K. Lee & H. Lee Expires January 31, 2018 [Page 5]
Internet-Draft ACTN Multiple Vendor Domains July 2017
needs. But, as demand grew on data network, service providers
started to offer Ethernet-based leased line services in speed
ranging from 10Mbps and 10Gbps. Especially, MPLS-TP defined by IETF
is mainly used to achieve the carrier-grade characteristics (high-
level availability, guaranteed bandwidth, and OAM capabilities) of
SONET/SDH.
However, unlike the SONET/SDH networks, there are some limitations
to provide an end-to-end connectivity on top of telco infrastructure
with multi-vendor domains, because the EMS/NMS system of each vendor
does not communicate with each other in order to operate the multi-
vendor MPLS-TP networks due to the lack of common API. Each vendor
devices can support the same OAM and protection mechanisms, but end-
to-end connections are only created through data planes manually
configured from operators. Operators should manually give specific
MPLS label values to network elements acting as LSRs/LERs
considering the collision of used label values. In addition, the
exchange of specific OAM and protection switching information should
be achieved for appropriate end-to-end OAM and protection switching
operations. Even, it is hard to operate such kinds of connections
because of network elements non-visible from each vendor's EMS/NMS
system. As a result, operators might have difficulties in terms of
execution of on-demand OAM functions and manual protection switching
commands.
To achieve an end-to-end connectivity, each vendor domain creates
their own MPLS-TP tunnels by their EMS/NMS systems and then, MEF-
based ENNIs are used at the interconnection links. But, this
solution still requires manual configurations from operators for S-
VLAN handling and various ENNI parameters (connectivity information,
and bandwidth profile).
2.2. Example B - Data Center Interconnect (DCI)
Enterprise customer data centers are consolidating, growing and
becoming more redundant and dynamic. Thereby, service providers
provide a new data center interconnect (DCI) service connecting
geographically separate data centers to their customers to extend
their computing resources or enhance the service availability
between data centers or main sites.
DCI supports two possible transport scenarios over metro and
regional area networks.
- DWDM or CWDM: This is a Layer 1 type of DCI service using optical
WDM equipment between data centers to meet the highest bandwidth
and lowest latency requirements. Depending on distance and
capacity requirements between data centers, provider choose a
short-reach CWDM or long-reach DWDM technology. Then, customers
K. Lee & H. Lee Expires January 31, 2018 [Page 6]
Internet-Draft ACTN Multiple Vendor Domains July 2017
are provided a private optical network using an optical
wavelength.
- L2VPN: This is a Layer 2 type of DCI service using Ethernet-based
L2VPN technologies for customers requiring seamless LAN extension.
Two kinds of L2VPN technologies are applicable where one is
carrier Ethernet transport technology and another is L2VPN
solutions based on IP/MPLS or overlay network technologies such as
VxLAN, NVGRE or GRE. The Carrier Ethernet technology provides
predictable performance in terms of various Ethernet services
(i.e., E-Line, E-Tree, E-LAN) defined by MEF (Metro Ethernet
Forum). In the meantime, L2VPN based on IP technologies cover less
stringent latency and bandwidth requirements.
The scenarios are mapped into networking technologies that can meet
the underlying requirements. From the provider perspectives, the
providers should support a number of different network technologies
metro and regional area network between data centers in order for
offering appropriate virtual private networks depending on the
customer requirements. Unlike the case of Example A in this section,
some of IP-based overlay solutions only provide SDN-based interface
technologies such as NetConf, RestConf, OpenFlow and SNMP without
any EMS/NMS systems. Therefore, it is necessary to implement an
integration system for the control and management of the L1 and
L2/L3 devices. Resource discovery and control throughout the
multiple layers should be achieved in provider networks to offer
more rapid service agility and efficiency of the DCI service.
3. Requirements
In the previous section, we discussed the current challenges and
issues that prevent operators from offering end-to-end connectivity
services in multi-vendor domain transport networks.
This section provides a high-level requirement for enabling end-to-
end connectivity services in multi-vendor domain transport networks.
Figure 2 shows information flow requirements of the aforementioned
context.
K. Lee & H. Lee Expires January 31, 2018 [Page 7]
Internet-Draft ACTN Multiple Vendor Domains July 2017
+-------------------------------------------------+
| |
| Customer On-demand Network Service |
| |
+-------------------------------------------------+
/|\
|
\|/
+-------------------------------------------------+
| |
| Abstracted Global View |
| |
+-------------------------------------------------+
/|\
|
\|/
+-------------------------------------------------+
| |
| Single Integrated E2E Network View |
| |
+-------------------------------------------------+
/|\ /|\ /|\
| | |
\|/ \|/ \|/
+-------------+ +-------------+ +-------------+
| | | | | |
| Domain A | | Domain B | | Domain C |
| Control(NMS)| | Control(NMS)| | Control(Dev)|
+-------------+ +-------------+ +-------------+
Figure 2. Information Flow Requirements for Enabling End-to-End
Network Connectivity Service in Multi-vendor Domain Networks
There are a number of key requirements from Figure 2.
- A single integrated end-to-end network view is necessary to be
able to provision the end-to-end paths that traverse multiple
vendor domains. In this approach the scalability and
confidentiality problems are solved, but new considerations must
be taken into account:
o Limited awareness, by the VNC, of the intra-domain resources
availability.
o Sub-optimal path selection.
K. Lee & H. Lee Expires January 31, 2018 [Page 8]
Internet-Draft ACTN Multiple Vendor Domains July 2017
- The path computations shall be performed in two stages: first on
the abstracted end-to-end network view (happening at VNC), and on
the second stage it shall be expanded by each PNC.
- In order to create a single integrated end-to-end network view,
discovery of inter-connection data between domains including the
domain border nodes/links is necessary. (The entity to collect
domain-level data is responsible for collecting inter-connection
links/nodes)
- The entity to collect domain-level data should recognize
interoperability method between each domain. (There might be
several interoperability mechanisms according to technology being
applied.)
- The entity responsible to collect domain-level data and create an
integrated end-to-end view should support push/pull model with
respect to all its interfaces.
- The same entity should coordinate a signaling flow for end-to-end
connections to each domain involved. (This entity to domain
control is analogous to an NMS to EMS relationship)
- The entity responsible to create abstract global view should
support push/pull model with respect to all its interfaces. (Note
that the two entities (an entity to create an integrated end-to-
end view and an entity to create an abstracted global view) can be
assumed by the same entity, which is an implementation issue.
- Hierarchical composition of integrated network views should be
enabled by a common API between NorthBound Interface of the Single
Integrated End-to-End view (handled by VNC) and Domain Control
(handled by PNC).
- There is a need for a common API between each domain control to
the entity that is responsible for creating a single integrated
end-to-end network view. At the minimum, the following items are
required on the API:
o Programmability of the API.
o The multiple levels/granularities of the abstraction of
network resource (which is subject to policy and service
need).
o The abstraction of network resource should include customer
end points and inter-domain gateway nodes/links.
K. Lee & H. Lee Expires January 31, 2018 [Page 9]
Internet-Draft ACTN Multiple Vendor Domains July 2017
o Any physical network constraints (such as SRLG, link
distance, etc.) should be reflected in abstraction.
o Domain preference and local policy (such as preferred peering
point(s), preferred route, etc.)
o Domain network capability (e.g., support of push/pull model).
- The entity responsible for abstraction of a global view into a
customer view should provide a programmable API to allow the
flexibility. Abstraction might be provided by representing each
domain as a virtual node (node abstraction) or a set of virtual
nodes and links (link abstraction). Node abstraction creates a
network topology composed by nodes representing each network
domain and the inter-domain links between the border nodes of each
domain.
o Abstraction of a global view into a customer view should be
provided to allow customer to dynamically request network on-
demand services including connectivity services.
o What level of details customer should be allowed to view
network is subject to negotiation between the customer and
the operator.
4. References
[ACTN-Frame] D. Ceccarelli, L. Fang, Y. Lee and D. Lopez, "Framework
for Abstraction and Control of Transport Networks," draft-
ceccarelli-actn-framework, work in progress.
[ACTN-PS] Y. Lee, D. King, M. Boucadair, R. Jing, and L. Murillo,
"Problem Statement for the Abstraction and Control of
Transport Networks," draft-leeking-actn-problem-statement,
work in progress.
[ACTN-Req] Y. Lee, D. Dhody, S. Belotti, K. Pithewan, D. Ceccarelli,
"Requirements for Abstraction and Control of Transport
Networks", draft-lee-teas-actn-requirements, work in
progress.
5. Acknowledgement
The authors wish to thank Young Lee for the discussions in the
document.
K. Lee & H. Lee Expires January 31, 2018 [Page 10]
Internet-Draft ACTN Multiple Vendor Domains July 2017
6. Contributors
Authors' Addresses
Kwang-koog Lee
KT
Email: kwangkoog.lee@kt.com
Hosong Lee
KT
Email: hosong.lee@kt.com
Ricard Vilalta
CTTC
Email: ricard.vilalta@cttc.es
Victor Lopez
Telefonica
Email: victor.lopezalvarez@telefonica.com
K. Lee & H. Lee Expires January 31, 2018 [Page 11]