| Rfc | 6139 |
|---|
| Title | Routing and Addressing in Networks with Global Enterprise Recursion
(RANGER) Scenarios |
|---|
| Author | S. Russert, Ed., E. Fleischman, Ed., F. Templin,
Ed. |
|---|
| Date | February 2011 |
|---|
| Format: | TXT, HTML |
|---|
| Status: | INFORMATIONAL |
|---|
|
Independent Submission S. Russert, Ed.
Request for Comments: 6139 Unaffiliated
Category: Informational E. Fleischman, Ed.
ISSN: 2070-1721 F. Templin, Ed.
Boeing Research & Technology
February 2011
Routing and Addressing in Networks with
Global Enterprise Recursion (RANGER) Scenarios
Abstract
"Routing and Addressing in Networks with Global Enterprise Recursion
(RANGER)" (RFC 5720) provides an architectural framework for scalable
routing and addressing. It provides an incrementally deployable
approach for scalability, provider independence, mobility,
multihoming, traffic engineering, and security. This document
describes a series of use cases in order to showcase the
architectural capabilities. It further shows how the RANGER
architecture restores the network-within-network principles
originally intended for the sustained growth of the Internet.
Status of This Memo
This document is not an Internet Standards Track specification; it is
published for informational purposes.
This is a contribution to the RFC Series, independently of any
other RFC stream. The RFC Editor has chosen to publish this
document at its discretion and makes no statement about its value
for implementation or deployment. Documents approved for
publication by the RFC Editor are not a candidate for any level of
Internet Standard; see Section 2 of RFC 5741.
Information about the current status of this document, any
errata, and how to provide feedback on it may be obtained at
http://www.rfc-editor.org/info/rfc6139.
Copyright Notice
Copyright (c) 2011 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
carefully, as they describe your rights and restrictions with respect
to this document.
Table of Contents
1. Introduction ....................................................3
2. Terminology .....................................................4
3. Approach ........................................................7
4. Scenarios ......................................................11
4.1. Global Concerns ...........................................11
4.1.1. Scaling the Global Inter-Domain Routing Core .......11
4.1.2. Supporting Large Corporate Enterprise Networks .....13
4.2. Autonomous System Concerns ................................16
4.3. Small Enterprise Concerns .................................16
4.4. IPv4/IPv6 Transition and Coexistence ......................18
4.5. Mobility and MANET ........................................21
4.5.1. Global Mobility Management .........................21
4.5.2. First-Responder Mobile Ad Hoc Networks (MANETs) ....23
4.5.3. Tactical Military MANETs ...........................24
4.6. Provider Concerns .........................................27
4.6.1. ISP Networks .......................................27
4.6.2. Cellular Operator Networks .........................28
4.6.3. Aeronautical Telecommunications Network (ATN) ......28
4.6.4. Unmanaged Networks .................................31
5. Mapping and Encapsulation Concerns .............................32
6. Problem Statement and Call for Solutions .......................32
7. Summary ........................................................33
8. Security Considerations ........................................33
9. Acknowledgements ...............................................34
10. References ....................................................34
10.1. Normative References .....................................34
10.2. Informative References ...................................34
1. Introduction
The Internet is continually required to support more users, more
internetwork connections, and increasing complexity due to diverse
policy requirements. This growth and change strains the
infrastructure and demands new solutions. Some of the complementary
approaches to transform Internet technology are being pursued
concurrently within the IETF: translation (including Network Address
Translation (NAT)), tunneling (map and encapsulate), and native IPv6
[RFC2460] deployment. Routing and Addressing in Networks with Global
Enterprise Recursion (RANGER) [RFC5720] describes the architectural
elements of a "map and encapsulate" approach that also facilitates
the other two approaches. This document discusses RANGER operational
scenarios.
RANGER provides an architectural framework for scalable routing and
addressing. It provides for scalability, provider independence,
mobility, multihoming, and security for the next-generation Internet.
The RANGER architectural principles are not new. They can be traced
to the deliberations of the ROAD group [RFC1380], and also to still
earlier works including NIMROD [RFC1753] and the Catenet model for
internetworking [CATENET] [IEN48] [RFC2775]. [RFC1955] captures the
high-level architectural aspects of the ROAD group deliberations in a
"New Scheme for Internet Routing and Addressing (ENCAPS) for IPNG".
The Internet has grown tremendously since these architectural
principles were first developed, and that evolution increases the
need for these capabilities. The Internet has become a critical
resource for business, for government, and for individual users
throughout the developed world. RANGER carries forward these
historic architectural principles, creating a ubiquitous enterprise
network structure that can represent collections of network elements
ranging from the granularity of a singleton router all the way up to
an entire Internet. This enterprise network structure uses border
routers that configure tunnel endpoints to connect potentially
recursively nested networks. Each enterprise network may use
completely independent internal Routing Locator (RLOC) address
spaces, supporting a virtual overlay network connecting edge networks
and devices that are addressed with globally unique Endpoint
Interface iDentifiers (EIDs). The RANGER virtual overlay can
transcend traditional administrative and organizational boundaries.
In its purest form, this overlay network could therefore span the
entire Internet and restore the end-to-end transparency envisioned in
[RFC2775].
The RANGER architecture drew early observations from the Intra-Site
Automatic Tunnel Addressing Protocol (ISATAP) [RFC5214] [RFC5579] but
now uses Virtual Enterprise Traversal (VET) [RFC5558], the Subnetwork
Encapsulation and Adaptation Layer (SEAL) [RFC5320], and other
mechanisms including IPsec [RFC4301] as its functional building
blocks. This document describes use cases and shows how the RANGER
mechanisms apply. Complementary mechanisms (e.g., DNS, DHCP, NAT,
etc.) are included to show how the various pieces can work together.
It expands on the concepts introduced in "IPv6 Enterprise Network
Scenarios" [RFC4057] and "IPv6 Enterprise Network Analysis - IP Layer
3 Focus" [RFC4852], and shows how the enterprise network model
generalizes to a broad range of scenarios. These use cases are
included to provide examples, invite criticism and comment, and
explore the potential for creating the next-generation Internet using
the RANGER architecture. Familiarity with RANGER, VET, SEAL, and
ISATAP are assumed.
2. Terminology
Internet Topology Hierarchy
The Internet Protocol (IP) natively supports a topology hierarchy
comprised of increasing aggregations of networked elements.
Network interfaces of devices are grouped into subnetworks, and
subnetworks are grouped into larger aggregations. Subnetworks can
be optionally grouped into areas and the areas grouped into an
autonomous system (AS). Alternatively, subnetworks can be
directly grouped into an AS. The foundation of the IP Topology
Hierarchy is the AS, which determines the administrative
boundaries of a network deployment including its routing,
addressing, quality of service, security, and management.
Intra-domain routing occurs within an autonomous system, and
inter-domain routing links autonomous systems into a "network of
networks" (Internet).
Routing Locator (RLOC)
an address assigned to an interface in an enterprise-interior
routing region. Note that RLOC space is local to each enterprise
network.
The IPv4 public address space currently in use today can be
considered as the RLOC space for the global Internet as a giant
"enterprise network".
Endpoint Interface iDentifier (EID)
an address assigned to an edge network interface of an end system.
Note that EID space is global in scope, and must be separate and
distinct from any RLOC space.
commons
an enterprise-interior routing region that provides a subnetwork
for cooperative peering between the border routers of diverse
organizations that may have competing interests. An example of a
commons is the Default-Free Zone (DFZ) of the global Internet.
The enterprise-interior routing region within the commons uses an
addressing plan taken from RLOC space.
enterprise network
the same as defined in [RFC4852], where the enterprise network
deploys a unified RLOC space addressing plan within the commons,
but may also contain partitions with disjoint RLOC spaces and/or
organizational groupings that can be considered as enterprises
unto themselves. An enterprise network therefore need not be "one
big happy family", but instead provides a commons for the
cooperative interconnection of diverse organizations that may have
competing interests (e.g., such as the case within the global
Internet Default-Free Zone).
Historically, enterprise networks are associated with large
corporations or academic campuses. However, in RANGER an
enterprise network may exist at any IP Topology Hierarchy level.
The RANGER architectural principles apply to any networked entity
that has some degree of cooperative active management. This
definition therefore extends to home networks, small office
networks, a wide variety of Mobile Ad hoc Networks (MANETs), and
even to the global Internet itself.
site
a logical and/or physical grouping of interfaces within an
enterprise network commons, where the topology of the site is a
proper subset of the topology of the enterprise network. A site
may contain many interior sites, which may themselves contain many
interior sites in a recursive fashion.
Throughout the remainder of this document, the term "enterprise"
refers to either enterprise or site; i.e., the RANGER principles
apply equally to enterprises and sites of any size or shape. At
the lowest level of recursive decomposition, a singleton
Enterprise Border Router can be considered as an enterprise unto
itself.
Enterprise Border Router (EBR)
a node at the edge of an enterprise network that is also
configured as a tunnel endpoint in an overlay network. EBRs
connect their directly attached networks to the overlay network,
and connect to other networks via IP-in-IP tunneling across the
commons to other EBRs. This definition is intended as an
architectural equivalent of the functional term "EBR" defined in
[RFC5558], and is synonymous with the term "xTR" used in other
contexts (e.g., [LISP]).
Enterprise Border Gateway (EBG)
an EBR that also connects the enterprise network to provider
networks and/or to the global Internet. EBGs are typically
configured as default routers in the overlay, and provide
forwarding services for accessing IP networks not reachable via an
EBR within the commons. This definition is intended as an
architectural equivalent of the functional term "EBG" defined in
[RFC5558], and is synonymous with the term "default mapper" used
in other contexts (e.g., [APT]).
overlay network
a virtual network manifested by routing and addressing over
virtual links formed through automatic tunneling. An overlay
network may span many underlying enterprise networks.
6over4
"Transmission of IPv6 over IPv4 Domains without Explicit Tunnels"
[RFC2529]; functional specifications and operational practices for
automatic tunneling of unicast/multicast IPv6 packets over
multicast-capable IPv4 enterprise networks.
ISATAP
Intra-Site Automatic Tunnel Addressing Protocol (ISATAP) [RFC5214]
[RFC5579]; functional specifications and operational practices for
automatic tunneling over unicast-only enterprise networks.
VET
Virtual Enterprise Traversal (VET) [RFC5558]; functional
specifications and operational practices that provide a functional
superset of 6over4 and ISATAP. In addition to both unicast and
multicast tunneling, VET also supports address/prefix
autoconfiguration as well as additional encapsulations such as
IPsec, SEAL, UDP, etc.
SEAL
Subnetwork Encapsulation and Adaptation Layer (SEAL) [RFC5320]; a
functional specification for robust packet identification and link
MTU adaptation over tunnels. SEAL supports effective ingress
filtering and adapts to subnetworks configured over links with
diverse characteristics.
Within the RANGER architectural context, the SEAL "subnetwork" and
RANGER "enterprise" should be considered as identical
abstractions.
Provider-Independent (PI) prefix
an EID prefix (e.g., 2001:DB8::/48, 192.0.2/24, etc.) that is
routable within a limited scope and may also appear in enterprise
network mapping tables. PI prefixes that can appear in mapping
tables are typically delegated to a BR by a registry, but are not
aggregated by a provider network.
Provider-Aggregated (PA) prefix
an EID prefix that is either derived from a PI prefix or delegated
directly to a provider network by a registry. Although not widely
discussed, it bears specific mention that a prefix taken from a
delegating router's PI space becomes a PA prefix from the
perspective of the requesting router.
Customer Premises Equipment (CPE) Router
a residential or small office router that provides IPv4 and/or
IPv6 support. The user or the service provider may manage the
router.
Carrier-Grade NAT (CGN)
a special (usually high capacity) IPv4-to-IPv4 NAT deployed within
the service provider network that serves multiple subnets.
3. Approach
The RANGER [RFC5720] architecture seeks to fulfill the objectives set
forth in [RFC1955]:
o No Changes to Hosts
o No Changes to Most Routers
o No New Routing Protocols
o No New Internet Protocols
o No Translation of Addresses in Packets
o Reduce the Routing Table Size in All Routers
o Use the Current Internet Address Structure
The RANGER enterprise network is a cooperative networked collective
sharing a common (business, social, political, etc.) goal. An
enterprise network can be simple or complex in composition and can
operate at any IP Topology Hierarchy level. Although RANGER focuses
on encapsulation, it is also compatible with both native and
translated routing and addressing.
RANGER enables a protocol and/or addressing system to be connected in
a virtual overlay across an untrusted transit network, or "commons".
While it does not show all possible uses, Figure 1 illustrates that
RANGER supports the creation of a distributed network across an
intervening commons, which could implement a dissimilar IP version,
routing protocol, or addressing system.
.--------------. .--------------. .-------------.
/ \_ _/ \_ _/ \
\ Enterprise A / \ Commons / \ Enterprise B /
\_ _ _ _ _ _ _ / \_ _ _ _ _ _ _ / \_ _ _ _ _ _ _/
Domains
Network / IPvx IPvy IPvz
Protocol \ IPv6 IPv4 IPv6
IP Security secured unsecured secured
Mgmt Domain Entity A ISP Entity B
/
| Public Addresses Private Addresses Public Addresses
Addressing |Private Addresses Public Addresses Private Addresses
| PA Addresses PI Addresses PA Addresses
\ PI Addresses PA Addresses PI Addresses
Figure 1. RANGER Links Distributed Enterprise Networks
The RANGER concepts can be applied recursively. They can be
implemented at any level within the IP Topology Hierarchy to create
an enterprise-within-enterprise organizational structure extending
traditional AS, area, or subnetwork boundaries. This structure uses
border routers that configure tunnel endpoints to enable
communications between potentially recursively nested enterprise
networks in a virtual overlay network that transcends traditional
administrative and organizational boundaries. In its purest form,
this overlay network could therefore span the entire Internet and
restore end-to-end transparency [RFC2775].
The RANGER architecture applies the best current practice insights
from previous encapsulation systems as they are currently articulated
within the Virtual Enterprise Traversal [RFC5558], and Subnetwork
Encapsulation and Adaptation Layer [RFC5320] functional
specifications. The result is an architecture and protocol system
that can be used to create arbitrarily complex, scalable IP
deployments that support both unicast and multicast routing and
addressing systems.
RANGER supports scalable routing through a recursively nested
enterprise-within-enterprise network capability. The fundamental
building block is the Enterprise Border Router (EBR) (see Figure 2).
The EBR is the limiting factor for RANGER recursion, and in certain
contexts a singleton EBR can be viewed as an enterprise network unto
itself. Traditional network infrastructures can be extended to
support complex structures solely with the addition of EBRs with no
other modification to any networked entity.
An EBR can be a commercial off-the-shelf router, a tactical military
radio, an aircraft mobile router, etc., but it can also be an end
system (e.g., a laptop computer, a soldiers' handheld device, etc.)
with an embedded gateway function [RFC1122].
Provider-Edge Interfaces
x x x
| | |
+--------------------+---+--------+----------+ E
| | | | | n
| I | | .... | | t
| n +---+---+--------+---+ | e
| t | +--------+ /| | r
| e I x----+ | Host | I /*+------+--< p I
| r n | |Function| n|**| | r n
| n t | +--------+ t|**| | i t
| a e x----+ V e|**+------+--< s e
| l r . | E r|**| . | e r
| f . | T f|**| . | f
| V a . | +--------+ a|**| . | I a
| i c . | | Router | c|**| . | n c
| r e x----+ |Function| e \*+------+--< t e
| t s | +--------+ \| | e s
| u +---+---+--------+---+ | r
| a | | .... | | i
| l | | | | o
+--------------------+---+--------+----------+ r
| | |
x x x
Enterprise-Edge Interfaces
Figure 2. Enterprise Border Router (EBR)
EBRs connect networks and end systems to one or more enterprise
networks via a repertoire of interface types. Enterprise-interior
interfaces attach to a commons. Provider-edge interfaces support
traditional routing relationships up the IP Topology Hierarchy, and
enterprise-edge interfaces support traditional relationships down the
IP Topology Hierarchy. Internal virtual interfaces are typically
loopback interfaces or VMware-like host-in-host interfaces.
VET interfaces support RANGER recursion and IP-in-IP encapsulation.
VET interfaces are configured over provider-edge, enterprise-
interior, or enterprise-edge interfaces to allow recursion
horizontally or vertically within the IP Topology Hierarchy. A VET
interface may be configured over several underlying interfaces that
all connect to the same enterprise network. This creates a link-
layer multiplexing capability that can provide several advantages
(see [RFC1122], Section 3.3.4). One important advantage is
continuous operation across failovers between multiple links attached
to the same enterprise network, without any need for readdressing.
Figure 3 shows two enterprise networks (each with their own internal
addressing and routing systems) that communicate over a virtual
overlay network across a commons. The virtual overlay is manifested
by tunneling, which links enterprise networks separated by
geographical remoteness, protocol incompatibility, or both. An
ingress EBR (iEBR) within the left enterprise network seeks to
forward encapsulated packets across the commons to the egress EBR
(eEBR) within the right enterprise network.
The figure shows that the eEBR assigns a Routing Locator (RLOC)
address on its interface to the commons' interior IP routing and
address space, while the destination host assigns an Endpoint
Interface iDentifier (EID) on its enterprise-edge interface. The
iEBR uses a mapping system to discover the RLOC of an eEBR on the
path to the destination EID address. A distinct mapping system is
maintained within each recursively nested enterprise network instance
operating at a specific level of the IP Topology Hierarchy. RANGER
uses the mapping system to join peer enterprise networks via a
virtual overlay across a commons.
Mapping System RLOC EID
. (BGP, DNS, etc.) . .
.---.------. .----------. . .------.---.
/ . \ / \ . / . \
/ (O) iEBR------/--------------\------eEBR * \
\ / \ Commons / \ /
\_ _ _ _ _ _ / \_ _ _ _ _ _ / \_ _ _ _ _ _/
Enterprise Network A Enterprise Network B
Figure 3. The RANGER Model
EBRs must configure both RLOC and EID addresses and/or prefixes.
Autoconfiguration is coordinated with Enterprise Border Gateways
(EBGs) that connect to the next-higher layer in the recursive
hierarchy, as specified in VET. Standard mechanisms including DHCP
[RFC2131] [RFC3315] and Stateless Address Autoconfiguration (SLAAC)
[RFC4862] are used for this purpose.
Similarly, EBRs require a means to discover other EBRs and EBGs that
can be used as enterprise network exit points. VET specifies
mechanisms for border router discovery using the global DNS and/or
enterprise-local name services such as Link-Local Multicast Name
Resolution (LLMNR) [RFC4795].
The mapping system is a distributed database that is synchronized
among a limited set of mapping agents. Database synchronization can
be achieved by many different protocol alternatives. The most
commonly used alternatives are either the Border Gateway Protocol
(BGP) [RFC4271] or the Domain Name System (DNS) [RFC1035]. Mapping-
system databases can be populated by many different mechanisms
including administrative configuration and automated prefix
registrations.
EBRs forward initial packets for which they have no mapping to an
EBG. The EBG in turn forwards the packet toward the final
destination and returns a redirect to inform the EBR of a better next
hop if necessary. The EBR then receives a mapping reply that it can
use to populate its Forwarding Information Base (FIB). It then
encapsulates each forwarded packet in an outer IP header for
transmission across the commons to the remote RLOC address of an
eEBR. The eEBR in turn decapsulates the packets and forwards them to
the destination EID address. The Routing Information Base (RIB)
within the commons only needs to maintain state regarding RLOCs and
not EIDs. The synchronized EID-to-RLOC mapping state is not subject
to oscillations due to link state changes within the commons. RANGER
supports scalable addressing by selecting a suitably large EID
addressing range that is distinct from any enterprise-interior RLOC
addressing ranges.
4. Scenarios
4.1. Global Concerns
4.1.1. Scaling the Global Inter-Domain Routing Core
Growth in the Internet has created challenges in routing and
addressing that have been recognized for many years
[RADIR-PROB-STATE]. IPv4 [RFC0791] address space is limited, and
Regional Internet Registry (RIR) allocation is passing the "very
painful" Host Density (HD) ratio threshold of 86% (that is, 192M
allocated addresses) [RFC3194]. As a result, exhaustion of the IPv4
address pool is predicted within the next two years [IPv4POOL],
[HUSTON-END]. IPv6 promises to resolve the address shortage with a
much larger address space, but transition is costly and could
exacerbate BGP problems described below. Richer interconnection,
increased multihoming (especially with provider-independent (PI)
addresses), and a desire to support traffic engineering via finer
control of routing has led to super-linear growth of BGP routing
tables in the Default-Free Zone, or "DFZ", of the Internet. This
growth is placing increasing pressures on router capacities and
technology costs that are unsustainable for the longer term within
the current Internet routing framework.
RANGER allows the coordinated reuse of addresses from enterprise to
enterprise by making RLOC address spaces independent of one another.
Figure 4 shows how the RANGER architecture allows the use of separate
address spaces for RLOC and EID addressing in the Internet. This
yields more endpoint address space, especially with the use of IPv6,
and also reduces the load on BGP in the Internet routing core. Note
that Figure 4 could represent variants of RFC 4057 scenarios 1 and 2.
EID RLOC EID
PA Spaces PI
Allocation Registration
.-------------------------------. ^
/ Internet Commons \ |
| .---------------------------. | |
2001:DB8::/40 | / Enterprise A \ | 2001:DB8:10::/56
| |/ 10.1/16 \ | ^
| || .-------------------------. || |
V || / Enterprise A.1 \ || |
2001:DB8::/48 || | 10.1/16 | || 2001:DB8:11::/56
|| \_________________________/ / |
| \ / |
| --------------------------- |
| |
| .---------------------------. |
| / Enterprise B \ |
2001:DB8:100::/40 | | 10.1/16 | | 2001:DB8:12::/56
| \____________________________/ |
\ /
\_______________________________/
Figure 4. Enterprise Networks and the Internet
RLOC address spaces are entirely independent of one another, as they
are used only within an enterprise network (recall that an enterprise
network can exist at any level of the IP Topology Hierarchy). Such
an arrangement allows each RLOC space to maintain an independent
routing system and thereby avoid the inherent scaling issues if a
single monolithic routing system were used for all.
EID address space can be provider-aggregated (PA) or PI, and taken
from either IPv4 or IPv6. EID addresses (barring the use of Network
Address Translation (NAT)) are globally unique, even when routable
only within a more limited scope (e.g., in their own edge networks).
The IRTF routing research group is investigating a Preliminary
Recommendation for a routing architecture [RFC6115] that provides a
taxonomy for routing scaling solutions for the global Internet
inter-domain routing core. RANGER presents a core/edge separation
architecture within this taxonomy that uniquely shows applicability
from the core all the way out to edge networks via its recursive
enterprise-within-enterprise framework. RANGER is further compatible
with a number of schemes intending to address routing scaling issues,
including "APT: A Practical Transit Mapping Service" [APT], "FIB
Suppression with Virtual Aggregation" [GROW-VA], "Locator/ID
Separation Protocol (LISP)" [LISP], and others.
4.1.2. Supporting Large Corporate Enterprise Networks
Each enterprise network operator must be able to manage its internal
networks and use the Internet infrastructure to achieve its
performance and reliability goals. Enterprise networks that are
multihomed or have mobile components frequently require provider-
independent addressing and the ability to coordinate with multiple
providers without renumbering "flag days" [RFC4192] [RFC5887].
RANGER provides a way to coordinate addressing plans and
inter-enterprise routing, with full support for scalability, provider
independence, mobility, multihoming, and security.
_.--------------------._
_.---'' -.
,--'' ,---. `---.
,-' X5 X6 .---.. `-.
,\' ,.X1-.. / \ ,' `. `.
,\' ,' `. .' E2 '. X8 Em \ `.
/ / \ | ,--. | / _,.._ \ \
/ / E1 \ | Y3 `. | | / Y7 | \
; | ___ | | ` W Y4 |... | `Y6 ,' | :
| | ,-' `. X2 | `--' | | `'' | |
: | | V Y2 | \ _ / | | ;
\ | `-Y1,,' | \ .' Y5 / \ ,-Y8'`- / /
\ \ / \ \_' / X9 `. ,'/ /
`. \ X3 `.__,,' `._ Y9'',' ,'
` `._ _,' ___.......X7_ `---' ,'
` `---' ,-' `-. -'
`---. `. E3 Z _' _.--'
`-----. \---.......---' _.---''
`----------------''
<------------------- Global IPv4 Internet ------------------>
Figure 5. Enterprise Networks within the Internet Commons
Figure 5 depicts enterprise networks E1 through Em connected to the
global IPv4 Internet via Enterprise Border Routers (EBRs) X1 through
X9. These same border nodes also act as Enterprise Border Gateways
(EBGs) that provide default routing services for nodes within their
respective enterprise networks. The global Internet forms a commons
across which the various enterprise networks connect as cooperating
yet potentially competing entities. Within each enterprise network
there may be arbitrarily many hosts, routers, and networks (not shown
in the diagram) that use addresses taken from that enterprise
network's RLOC space and over which both encapsulated IP packets with
(global-scoped) EID addresses and unencapsulated IP packets with
(enterprise-local) RLOC addresses can be forwarded.
Each enterprise network may encompass lower-tier networks; for
instance, the singleton EBR "W" in network E2 resides in a lower-tier
network (say E2.1), and (along with any of its attached devices) may
be considered as an enterprise unto itself. W sees Y3 and Y4 as
EBGs, which in turn see X5 and X6 as EBGs that connect to a common
provider network (in this case, the Internet). Each enterprise
network has one or more Endpoint Interface iDentifier (EID) address
prefixes used for addressing nodes on edge networks. RANGER's map-
and-encaps approach separates the mapping of EIDs to Routing Locators
(RLOCs) from the Routing Information Base (RIB) in the Internet
commons that are assigned to EBR router interfaces. Not only does
BGP in the Internet commons only need to maintain state regarding
RLOCs in the Internet commons, it has fewer unique routes to maintain
because only routes to EBRs are needed; traffic engineering can
therefore be accommodated via the mapping database.
In Figure 5, enterprise network E2 represents a corporation that has
multiple locations and connections to multiple ISPs. The corporation
has recently merged with another corporation so that its internal
network has two disjoint RLOC address spaces, but neither of the
formerly separate entities can bear the burden of address
renumbering. Enterprise network E2 can use a suitably large IPv4
and/or IPv6 EID addressing range (that is distinct from any
enterprise-interior RLOC addressing range) to support end systems on
enterprise-edge networks with no disruption to preexisting address
numbering.
As EBRs are deployed to connect enterprise networks together,
ordinary routers within the enterprise network continue to function
as normal and deliver both ordinary and encapsulated packets across
the existing Internet infrastructure and the network's own RLOC
commons. Legacy IPv4 services that bind to RLOC addresses continue
to be supported even as EID-based services are rolled out. Where a
legacy IP client and server are within the same RLOC address space,
they simply communicate by using RLOC-based routing across the
enterprise network commons. If the client and server are not within
the same RLOC address space, they communicate through some form of
network address and/or protocol translation (see [RFC5720],
Section 3.3.4 for details). EBRs from the various enterprise
networks publish their EID prefixes to an enterprise-specific mapping
system, so that other EBRs from the various enterprise networks can
consult the mapping system to receive the RLOC address of one or more
EBRs that serve the EID prefix.
As an example, when an end system connected to W in E2.1 has a packet
to send to node Z in enterprise network E3, W sends the packet to EBR
Y4, which encapsulates the packet in an outer IP packet with its own
source address and the RLOC address of the next-hop EBR as the
destination -- in this case, X6. X6 decapsulates the packet and
looks up the destination EID prefix, obtaining the RLOC of X7 as
next-hop. X6 then encapsulates the IPv6 packet in a packet with RLOC
address X6 as the source and X7 as the destination. X7 decapsulates
the packet on receipt and forwards it via its enterprise-edge
interface to node Z.
This example uses one thread out of many that are possible using
RANGER; see [RFC5720] and [RFC5558] for other options and details.
Many enterprise networks that use proxies and firewalls at their
border routers today will wish to maintain that control over their
enterprise borders, and the use of RANGER does not preclude such
configurations (for example, see Section 4.3).
4.2. Autonomous System Concerns
An enterprise network such as E2 in Figure 5 above can represent an
AS within the IP Topology Hierarchy. A possible configuration for
enterprise network E2 is for each of its enterprise components to
also be recursive ASs linked together using the RANGER constructs.
Such a configuration is increasingly commonplace today for the
networks of very large corporations (e.g., Boeing's corporate
enterprise network). These networks support an internal instance of
the BGP linking many corporate-internal ASs and independent from the
BGP instance that maintains the RIB within the global Internet
Default-Free Zone (DFZ). Such configurations are often motivated by
scaling or administrative requirements.
Such a corporate entity is internally an Internet unto itself, albeit
with separate default routes leading to the true global Internet.
The enterprise network E2 therefore appears to the rest of the
Internet as if it were a traditional IP Topology Hierarchy AS. Since
RANGER supports recursion, each AS within such a network may itself
use BGP internally in place of an IGP, and can therefore also
internally be composed of a locally internal Internet in a recursive
fashion. This enterprise-within-enterprise framework can recursively
be extended as broadly and as deeply as required in order to achieve
the specific requirements of the deployment (e.g., scaling, unique
administration, and/or functional compartmentalization).
4.3. Small Enterprise Concerns
Global enterprise networks operating at the autonomous system level
of the IP Topology Hierarchy include multiple geographical regions,
multiple ISPs, and complex internal structures that naturally benefit
from the application of RANGER techniques. However, all other
enterprise network instances (both large and small) can also be
served by RANGER. For example, Small and Home Office (SOHO) networks
may comprise only a few computers on a single network segment or may
extend to larger configurations with security islands, internal
routers and switches, etc.
An important concern of the small enterprise network is the ability
to grow the network, change ISPs, or expand to more locations without
readdressing the existing network. Consider a small company that has
a single location in California. The ISP connection is via a router
that acts as a Network Address Translator and firewall for the
company. Addresses of the few computers ("Wksta") are taken from the
[RFC1918] private address space.
ISP
-------|----- Wksta Wksta
| Firewall |_____________|____________|
| NAT |
-------------
Figure 6. Simple SOHO Network
This configuration has been adequate for the few employees performing
software development work, since there is no need to expose services
within the site to the outside world. But now a web presence is
required as product introduction approaches. The network manager
deploys an EBR either as a co-resident function on the existing NAT/
firewall platform (as depicted in Figure 7) or on a separate
platform.
The EBR has a provider-edge interface connected to the ISP; the
preexisting workstations; the preexisting enterprise-edge interfaces
connecting the workstations; and enterprise-edge interfaces
connecting several network segments connected by routers that host
web servers, workstations, and other enterprise network services. A
VET interface is configured over the new service network to allow the
servers to be addressed from the public Internet.
ISP
|
+------|-----+
| <|--
| VET2 < |
| <|---
| |
| | Server Server
| VET1 <|--------|-----------|-------
| |
| +--------+ | Wksta Wksta
| |Firewall| |_____________|____________|
| | NAT | |
| +--------+ |
+------------+
Figure 7. RANGER Serving the Small Company
In this new configuration, the EBR maintains the services within a
"demilitarized zone (DMZ)" that is accessible from the public
Internet without exposing other corporate assets that are still
protected by the preexisting firewall/NAT functions.
Shortly afterward, an infusion of venture capital allows acceleration
of the product development and marketing work by adding programmers
in Tokyo and sales offices in New York and London. These new
branches connect via Virtual Private Network (VPN) links across the
Internet, and a new VET interface (VET2) is configured over these
links to form a new sub-enterprise:
ISP
|
+------|-----+
| <|------------London
| VET2 < |
| <|--------------------New York
| |
| | Server Server
| VET1 <|--------|-----------|-------
| |
| +--------+ | Wksta Wksta
| |Firewall| |_____________|____________|
| | NAT | |
| +--------+ |
+------------+
Figure 8. RANGER for Multiple Locations
4.4. IPv4/IPv6 Transition and Coexistence
End systems and networks need to accommodate long-term support for
both IPv4 and IPv6. Requirements for transition include support for
IPv4 applications running over IPv4 protocol stacks, IPv4
applications over IPv6 stacks, IPv4 applications over dual stacks,
and IPv6 or IPv4/IPv6-capable applications over both IPv6 and dual
stacks. Both encapsulation and translation will likely be needed to
allow applications, enterprises, and providers to incorporate IPv6,
including all intermediate states, without global coordination or a
"flag day".
The RANGER architecture facilitates the addition of IPv6 addressing
to existing IPv4 end systems and routers (i.e., via dual stack) as
well as the addition of IPv6 networks to the existing set of IPv4
networks. RANGER (with VET and SEAL) makes it possible to carry
packets originated in one protocol across a network infrastructure
supporting another protocol or routing system. Figure 1 shows how
RANGER supports various combinations of edge (EID) and core (RLOC
commons) technologies, going beyond IP version differences to include
mixed security, management, and addressing as well.
The RANGER architecture supports end-to-end communications across
arbitrarily long paths of concatenated enterprise networks connected
by EBRs. When IPv6 is used as Endpoint Interface iDentifier (EID)
space, each EBR can provision a globally unique set of IPv6 EID
prefixes without scaling limitations, due to the expanded IPv6
address space. For example, Figure 9 shows a pair of end systems,
"H" and "J", separated by an intervening set of enterprise networks
spanned by VET interfaces labeled "vet1" through "vet4", where the
path between "H" and "J" traverses the EBR path "V->Y1->X2->X7->Z":
+------+
| IPv6 |
|Server|
" " " " " " " "" " " " " " " " " " " " " " " " | S1 |
" " +--+---+
" . . . . . . . . . . . . . . . " |
" . . . . . . " |
" . +----+ v +----+ v +----+ +----+ +-----+-------+
" . | V += e =+ Y1 += e =+ X2 += =+ R2 +==+ Internet |
" . +-+--+ t +----+ t +----+ +----+ +-----+-------+
" . | 1 . . 2 . . . " |
" . H . . . . v . " |
" . . . . . . . . . . . e . " +--+---+
" . t . " | IPv4 |
" . . . . . . , . 3 . " |Server|
" . +----+ v +----+ . " | S2 |
" . | Z += e =+ X7 += . " +------+
" . +-+--+ t +----+ . "
" . | 4 . . . "
" . J . . . . . "
" . . . . . . . "
" "
" " " " " " " " " " " " " " "" " " " " " " "
Figure 9. EBR Waypoint Navigation Using IPv6
When each EBR in the path is assigned a unique set of IPv6 EID
prefixes (and registers these prefixes in the appropriate routing/
mapping tables), IPv6 can be used for navigation purposes with each
EBR in the path seen as a waypoint for navigation. This is true even
if IPv4 is used as the enterprise-local Routing Locator (RLOC)
address space and there were many IPv4 hops on the path between each
pair of neighboring EBRs.
RANGER further provides a compatible framework for incorporating
supporting mechanisms including protocol translation, application-
layer aspects of IPv4/IPv6 transition discussed in [RFC4038], and DNS
issues for IPv6 from [RFC4472]. For instances where IPv4
applications remain in use, RANGER expects that IPv4<->IPv6
translation will be supported via network-based [BEHAVE-v6v4] and/or
end system stack-based (e.g., [RFC2767]) protocol translation
systems. Figure 10 shows the NAT - Protocol Translation
(NAT-PT)-equivalent translation in the VET router, and Figure 11
shows the "Bump-In-the-Stack" (BIS)-equivalent translation in end
systems ([RFC2767]). These examples address scenarios not mentioned
in RFC 4852.
IPv4 App A IPv4 App B
_____________ _____________
|_TCP or UDP__| |_TCP or UDP__|
|____IPv4_____| |____IPv4_____|
______|______ _______|_____
/ \ / \
| IPv4-Only | | IPv4-Only |
| Site 1 | | Site 2 |
\_____________/ \_____________/
______|______ ______|_______
|____IPv4_____| _____________ |____IPv4_____|
|NAT-PT-equiv_| / \ |NAT-PT-equiv_|
|_TCP or UDP__| | Internet | |_TCP or UDP__|
|____IPv6_____| | (RANGER) | |____IPv6_____|
|__VET/SEAL___| \_____________/ |__VET/SEAL___|
\_______________/ \___________/
Figure 10. Translation in Routers
In Figure 10, an IPv4 application on end system A operates normally,
and the end system sends IPv4 packets on the IPv4-only site network.
The IPv4 packets are received by an Enterprise Border Router (EBR)
that translates them into IPv6 packets by a NAT-PT-equivalent
process. The EBR then encapsulates the packets into IPv4 and sends
them across the RANGER-enabled Internet to Site 2 where they are
received and decapsulated by an EBR for Site 2. The EBR uses NAT-PT-
equivalent translation to translate the resulting IPv6 packet back to
an IPv4 packet that is delivered across the Site 2 IPv4-only network
to an IPv4 application on end system B.
IPv4 App A IPv4 App B
_____________ _____________ _____________
|_TCP or UDP__| / \ |_TCP or UDP__|
|____BIS______| | Internet | |____BIS______|
|____IPv6_____| | (RANGER) | |____IPv6_____|
|__VET/SEAL___| \_____________/ |__VET/SEAL___|
\_______________/ \___________/
Figure 11. BIS-Style Translation in Dual-Stack End Systems
Figure 11 shows the simplified approach using a BIS translation
process within dual-stack end systems ([RFC2767]). In this case, the
IPv4 application on dual-stack end system A forms an IPv4 payload,
which is then transformed into an IPv6 packet within the end system
protocol stack itself. The IPv6 packet can then be encapsulated and
sent across the Internet to be decapsulated and sent to the dual-
stack end system hosting IPv4 application B. The BIS-equivalent
process on end system B reverses the translation, yielding an IPv4
packet for consumption by the IPv4-only application.
Other issues besides IP protocol translation may arise during
IPv4-IPv6 transition; [RFC4038] points out issues including
IPv4/IPv6-capable applications running on IPv4-only protocol stacks,
DNS responses that include addresses of both IP versions, and the
difficulty of supporting multiple application versions. It also
advises that applications be converted to dual support as a preferred
solution. These issues are outside the scope of this document.
4.5. Mobility and MANET
4.5.1. Global Mobility Management
Ubiquitous wireless access enables connection to network
infrastructure nearly anywhere. Vehicles and even persons can host
networks that move around with them. For example, commercial
aircraft networks include requirements for nomadic networks, local
mobility, and global mobility where the connection point between
airplane and ground station can move from one continent to another.
Mobile networks need to be able to use provider-independent (PI) as
well as provider-aggregated (PA) address prefixes. Some applications
such as voice require rapid or seamless connection handoffs -- also
known as session survivability. Internet routing should not be
unduly disrupted by mobility, so movement of mobile nodes or edge
networks should not cause large ripples of routing protocol traffic,
especially in the DFZ.
When a RANGER enterprise network is overlaid on the Internet, mobile
nodes or mobile routers (that connect arbitrarily complex edge
networks or enterprise networks) can move between different points of
attachment while remaining reachable and without creating excessive
routing churn. In a commercial airline scenario, an aircraft with a
mobile router would move between ground station points of attachment
(that may be on different continents) without the readdressing of its
onboard networks. Figure 12 shows an aircraft transiting between
four different access points: two that are part of Air Communications
Service Provider (ACSP) 1, one in ACSP2, and the last directly to the
Air Navigation Service Provider (ANSP). ACSP1 and ACSP2 in this
example might be on different continents, so a traditional Mobile IP
Home Agent scheme [RFC3775] [RFC5944] would result in very
inefficient paths for one ACSP or the other. The aero enterprise
network is an overlay that spans both continents and allows efficient
paths by providing multiple entry and exit points (only one, R2, is
shown).
Aircraft - - - - - - ,.- - - - - -.- - ->
. , . . +------+
. , . . | IPv6 |
. , . . |Server|
" ." " " ", "" " " ." " " " " .? " " " " " | S1 |
" . , . . " +--+---+
" . , . . " |
" . ... . . . . . +----+ " |
" . . . . . =+ X3 + " |
" . v +--- + . v . . v +----+ ? |
" . e =+ Y1 + . e . . e . +----+ +--------+
" . t +----+ . t +----+ . t . =+-R2-+==+Internet|
" . 1 . . 2 =+ X2 + . 3 . +----+ +--------+
" . . . +----+ . . " |
" . . . . . . . " +------+
" <ACSP1> <ACSP2> <ANSP> " | IPv4 |
" " |Server|
" - - vet4 - - " | S2 |
" " " " " " " " " " " " " "" " " " " " " | S2 |
<-- Aero Enterprise Network --> +------+
Figure 12. Commercial Airplane Mobility
When the plane moves between ground stations that are located within
the ACSP1 enterprise network, no routing or mapping changes need be
made outside ACSP1. Moreover, if link-layer multiplexing (as
mentioned in Section 3 above) is used, then the VET interface network
layer is unaware of the movement. When the point of access moves to
ACSP2, no changes are made outside the aero enterprise network. When
the aircraft moves between ground stations of the same parent
enterprise network (as indicated by the two different links from the
aircraft to ACSP1 in Figure 12), the aircraft announces its PI
prefixes at its new point of attachment and withdraws them from the
old. The worldwide Internet sees no change, and mapping-system churn
is confined to ACSP1, since the prefixes need not be announced or
withdrawn within the parent aero enterprise network; i.e., the churn
is isolated to lower tiers of the recursive hierarchy. This can be
contrasted with the deprecated mobility solution previously fielded
by Connexion, which propagated disruptive BGP changes into the
Internet routing system to support mobile onboard networks.
4.5.2. First-Responder Mobile Ad Hoc Networks (MANETs)
Many emerging network scenarios require autoconfiguration of Mobile
Ad hoc Networks (MANETs). Where first responders need networking for
communications and coordination between teams, RANGER allows each
team or agency to quickly stand up a network and then use the
autoconfiguration described in [RFC5558] to coordinate address/prefix
autoconfiguration and discover border routers needed for teams and
agencies to interconnect.
For example, Figure 13 shows how police units arriving on a scene
with no network infrastructure can create a wireless network using
vehicle-mounted 802.11 hotspots with one or more cellular, 802.16, or
satellite links in order to reach the Internet. In this example, the
California Highway Patrol sets up an incident management center with
a satellite link to the Internet and vet1 serving network L1. The
Los Angeles County Sheriff team sets up network L1.1 at their field
headquarters, and the Altadena police force creates the L1.2 network
with their mobile units. R2 is the router that serves as an EBG for
border routers X3 and X4, which connect networks L1.2 and L1.1,
respectively. X3 serves vet3, and X4 serves vet2.
In like manner, the Angeles National Forest creates enterprise
network F1, with the San Gabriel Ranger District setting up
enterprise network F1.1 and the Fire Response Team Enterprise Network
F1.2. R1 and R2 discover one another and become peer EBRs across the
Internet by means of manual configuration. In network L1, individual
PI address prefixes are announced from L1.2 and L1.1 to L1, and R2
advertises them to the satellite ISP. R1 receives a PA prefix from
its WiMAX provider and delegates parts of the prefix to X1 and X2.
R2 also runs an IGP with R1, advertising the PI prefixes to R1 and
learning the PA prefixes there.
+------+
| IPv6 |
|Server|
" " " " " " " "" " " " " " " " " " " " " " " " | S1 |
" Law Enforcement Enterprise Network " +--+---+
" 2001:DB8:10::/56 (PI) ----------------> " |
" . . . . . . . +--- + . . . . " |
" . =+ X3 +===========. . " +-----+-------+
" . +----+ v +--- + . v +----+ | +
" . | V += e . . . . e =+ R2 +==+ |
" . +-+--+ t . . +----+ t +----+ | |
" . | 3 . . vet2 + X4 += 1 . " | |
" . H1 . . +----+ . " | |
" . . . . . . . . . . . . . . " | |
" <L1.2> <L1.1> <L1> " | |
" 10/8 10/8 10/8 " | |
" " " " " " " " " " " " " " "" " " " " " " " | Internet |
| |
" " " " " " " "" " " " " " " " " " " " " " " " | |
" USDA Forest Service Enterprise Network " | |
" <----------------- 2001:DB8::/40 (PA) " | |
" . . . . . . . +--- + . . . . " | |
" . =+ X1 +===========. . " | |
" . +----+ v +--- + . v +----+ | |
" . | J += e . . . . e =+ R1 +==+ |
" . +-+--+ t . . +----+ t +----+ | |
" . | 6 . . vet5 + X2 += 4 . " +-----+-------+
" . H2 . . +----+ . " |
" . . . . . . . . . . . . . . " +--+---+
" <F1.2> <F1.1> <F1> " | IPv4 |
" 10/8 10/8 10/8 " |Server|
" " " " " " " " " " " " " " "" " " " " " " " | S2 |
+--+---+
Figure 13. First-Responder MANET
4.5.3. Tactical Military MANETs
Military networks reflect well-defined policy requirements that
differ in many ways from civilian networks. The military's
information security requirements result in information being labeled
into specific classifications. The Bell-LaPadula model
[BELL-LaPADULA] provides a mechanism to extend information security
policy into networked environments. This extension creates
communications security (COMSEC), whose routing and addressing
elements are cleanly supported by RANGER concepts.
Figure 3 shows that RANGER supports creation of a VET interface
between the enterprise-interior (network) interface of two Enterprise
Border Routers (EBR) located within separate enterprise networks, A
and B. When this concept is applied to enterprise networks operating
above the subnetwork level of the IP Topology Hierarchy, then this
VET interface uses IP-in-IP encapsulation. This corresponds with a
popular COMSEC approach (IPsec -- [RFC4301]). When this same RANGER
concept is applied to enterprise networks operating at the subnetwork
level of the IP Topology Hierarchy, then this corresponds to an older
form of COMSEC (Link Layer Encryption). When the same RANGER concept
is applied to enterprise networks being singleton EBR nodes (i.e.,
the interface level of the IP Topology Hierarchy), then this
corresponds to a third military COMSEC alternative (Link Encryption).
The previous paragraph shows the flexibility of the RANGER
architecture to describe COMSEC approaches in terms of IP Topology
Hierarchy structured relationships. The power of the RANGER
architecture becomes apparent when one recognizes that each of the
entities in Figure 3 may themselves be simple or complex network
structures operating at any specific level of the IP Topology
Hierarchy. (Complex structures refer to architectures that have been
extended by RANGER recursion.) For example, the commons in the
figure may itself be an interface, a subnetwork, an autonomous
system, or an Internet. Enterprise networks A and B can be a single
end system, a subnetwork, an autonomous system, or an Internet.
Tactical military MANETs differ from traditional networks in many
ways, the most obvious being the high mobility of tactical
deployments and self-forming-network attributes of MANETs themselves.
Because each networked tactical entity supports a radio/router, the
numbers of routers within military MANETs can be orders of magnitude
more numerous (denser) than traditional civilian networks. This
means that even small deployments have comparatively large router
populations when compared to non-MANET deployments. Larger router
populations directly create greater sensitivity to protocol
scalability issues. Router scalability issues are further
exacerbated because IP protocols react unfavorably to signal
intermittence, which effectively dampens and constrains router
scaling even when mitigation techniques are employed. Signal
intermittence itself is a characteristic of mobility and the radio
signal propagation attributes of local deployment environments (e.g.,
such issues as terrain, foliage, buildings, weather, distance, etc.).
War fighting also encourages war fighters to locate into more
defensible terrain features, many of which naturally reduce radio
signal propagation, further increasing the probability of signal
intermittence.
RANGER recursion enables MANETs that naturally encourage route
aggregation and scaling through simple "plug and play" hierarchical
arrangements that parallel organizational structures and do not
entail complex manual configurations. For example, a MANET
autonomous system may benefit from RANGER recursion by being
physically comprised of enterprise networks that are autonomous
systems themselves. This relationship can be recursively extended
vertically as deep as required in order to create route aggregation
between entities having common mission assignments at differing
levels of abstraction. Since MANET routing is an active research
topic, it is helpful to realize that these structures may or may not
use routing protocols similar to their civilian IP Topology Hierarchy
peers. For example, because of the behavior of BGP within highly
mobile environments, the Exterior Gateway Protocol (EGP) used to link
ASs may or may not be BGP and, if it is BGP, it may have unusual
timer settings. However, whatever IGP and EGP is used, RANGER
constructs can increase route aggregation between entities sharing
common mission assignments to enable route scaling.
Tactical military MANETs often have requirements to communicate with
stationary infrastructures. By localizing mobility into an
enterprise network, the specific mobility-friendly protocols can then
be localized and their aggregation results presented to the
stationary network using a protocol supported by the stable network.
This also reduces the impact of mobility upon routing and addressing
systems as reported to the stationary infrastructure. Mobility-
induced route fluctuations (e.g., routing flaps) can still occur, but
their impact can be dampened if RANGER constructs are used to
localize them in lower tiers of the IP Topology Hierarchy. For
example, enterprise network A in Figure 3 can be a military MANET,
and enterprise network B may be a stationary military entity. Recall
that enterprise networks A and B interface at a specific IP Topology
Hierarchy level, but they may be physically extended by RANGER
mechanisms. For example, enterprise network A can be a MANET
enterprise that is physically a network-of-networks Internet that
interfaces to enterprise network B as if it were an autonomous
system. This gives enterprise network B a more stable and aggregated
view of the enterprise network A Internet than would be the case if
it were directly aware of A's various sub-enterprise components.
Another key distinctive feature of tactical military networks is
that, because radio networks operate at a different classification
level than the data they convey, tactical military networks have
several orders of magnitude more COMSEC devices than do equivalently
sized stationary military deployments (i.e., the number of COMSEC
devices is a function of the number of mobile war-fighting entities).
This can create significant scalability issues within the overlay
COMSEC network relationships themselves. COMSEC scaling problems are
manifested in several dimensions. It is important to recognize,
however, that just as RANGER recursion was used vertically to create
IP Topology enterprise-within-enterprise structures in order to
improve routing aggregation and scaling, so RANGER recursion allows
for authorization of route-optimized transactions between peer
enterprises (within the same IP Topology Hierarchy level) to improve
COMSEC aggregation and scaling of the network overlay system. The
RANGER use of VET also combines with the Subnetwork Encapsulation and
Adaptation Layer (SEAL) to provide robust packet identification and
maximum transmission unit (MTU) link adaptation services over
tunnels. These capabilities protect against both source address
spoofing and black holes caused by MTU limitations.
4.6. Provider Concerns
Network providers must have a way to support the protocol transitions
and network types mentioned above and still remain reliable and
financially sound. The RANGER architecture provides ways to support
general Internet Service Providers (ISPs), cellular operator
networks, and specialized networks such as the Aeronautical
Telecommunications Network (ATN).
4.6.1. ISP Networks
Internet service provider networks provide a commons for the
connection of Customer Premises Equipment (CPE) routers [CPE-RTRS]
that connect arbitrarily complex customer networks. This is true
whether the ISP permits direct customer-to-customer communications,
or whether all communications are forwarded through ISP provider-edge
equipment.
The ISP commons must potentially support hundreds of thousands of CPE
routers (or more); hence the ISP may be obliged to assign private
IPv4 address allocations (i.e., instead of public) as RLOCs for CPE
routers. This gives rise to a "nested NATs" scenario, which can
increase the overall brittleness brought on by NAT traversal.
To address this brittleness, the ISP can deploy "Carrier-Grade NATs"
(CGNs) [INCR-CGN] that provide a second level of RLOC address
translation on the path from the CPE to the Internet. When the CGNs
are also configured as EBGs, CPE routers can discover them as default
routers for reaching EID-based services using the EBG discovery
mechanisms specified in VET.
"Scenarios and Analysis for Introducing IPv6 into ISP Networks"
[RFC4029] discusses both ISP backbone network and customer connection
transition considerations; however, this document considers router-
to-router tunneling use cases. Therefore the ISATAP mechanism (which
only supports host-to-router or host-to-host tunneling) is not
mentioned as a candidate technology. Early point solutions (e.g.,
the Tunnel Setup Protocol (TSP) [RFC5572], the Simple IPv6-in-IPv4
Tunnel Establishment Procedure (STEP) [STEP], etc.) were recommended.
This document suggests that RANGER, VET, and SEAL would also be
suitable solutions in these networks.
4.6.2. Cellular Operator Networks
[RFC4215] provides a (dated) "Analysis on IPv6 Transition in Third
Generation Partnership Project (3GPP) Networks". It envisions an
extended period of support for both IPv4 and IPv6 protocols in the
operator network. User Equipment (UE) uses the Packet Data Protocol
(PDP) context to establish tunnels through the operator network to a
Gateway General Packet Radio Service (GPRS) Support Node (GGSN).
RANGER could be used in 3GPP transition; when the UE uses IPv6, and
the PDP context is established across an IPv4 provider network, the
UE can configure itself as an EBR and contact the GGSN (as a RANGER
EBG) through VET tunneling.
Other [RFC4215] scenarios examine IPv4-only UEs, IPv6-only UEs, and
various combinations of IPv4 and IPv6 within the operator network.
Also to be considered are scenarios in which the UE is configured as
a router or bridge that connects an end system such as a laptop
computer. In that case, the UE could be the first-hop router/bridge
into the cellular provider network, and the laptop computer could be
configured as an EBR in the RANGER model. Again, the GGSN or a
device reachable through the GGSN could serve as a RANGER EBG.
4.6.3. Aeronautical Telecommunications Network (ATN)
The Aeronautical Telecommunications Network (ATN) is currently based
on the OSI and IPv4 protocols and is deployed only in limited areas.
The future ATN under consideration within the civil aviation industry
will be IPv6-based. The IP variant of ATN is expected to take the
form of a worldwide enterprise network that internally comprises an
aeronautical-only Internet that has additional external interfaces to
the global Internet. Within the ATN, there may be many Air
Communications Service Provider (ACSP) and Air Navigation Service
Provider (ANSP) networks that are internally organized either as
autonomous systems or internets within the ATN, i.e., as depicted in
Figure 5. Each of these entities may themselves be further
internally subdivided into lower-tier enterprise networks organized
as regional, organizational, or functional compartments. It is
important to note that while ACSPs and ANSPs within the ATN will
share a common objective of safety-of-flight for civil aviation
services, enterprise networks may have competing business, social, or
political interests that require that components be distinct ASs.
The RANGER principles therefore support collaborative objectives
while allowing very diverse local policy distinctions. In this
manner, entities that do not trust each other can create
collaborative infrastructures to achieve common goals.
Operational associations like this will characterize many future
deployments, including the US Department of Defense's Global
Information Grid (GIG). In particular, although the routing and
addressing arrangements of all enterprise networks require a mutual
level of cooperative active management at a certain level, scaling
issues, security policy differences, free market forces,
organizational differences, political distinctions, or other factors
may create internal competition among entities that otherwise share
common goals. This will require different enterprise networks within
that association to be separated into distinct ASs that are linked
within their own functional Internet relationship.
The ATN illustrates transition from OSI protocols to IPv6. It must
support mobility (see Section 4.5.1), and it serves many government
and private entities that cooperate to provide safe and efficient air
travel while often competing with one another. One possible way to
meet these needs with RANGER is to create an overlay using IP-in-IP
tunneling across the Internet, as illustrated in Figure 14. The aero
overlay forms an enterprise network, so that inner packets from ACSP
and ANSP edge networks that travel between VET interfaces on EBRs see
their passage across the Internet as only one hop.
_...--------------------------------------..._
/ \
( IPv4 Internet )
-...________________________________________...-
| / | \ |
| / | \ |
_...--------------------------------------..._
/ \
( Aero Overlay )
-...________________________________________...-
. . . . . .
. . . . . .
_...-------.._ _...-------.._ _...-------.._
/ \ / \ / \
( ACSP1 ) ( ANSP ) ( ACSP2 )
-...________...- -...________...- -...________...-
Figure 14. Aeronautical Overlay
Each Aeronautical Communications Service Provider (ACSP), and
Aeronautical Navigation Service Provider (ANSP) constitute an
enterprise network recursively nested below the aero overlay.
Relationships between the various enterprise networks can vary from
slight to tight integration. In the example, the ACSP and ANSP might
choose to exchange full routing information for their edge networks
using a coordinated global-scope RLOC address space across which ACSP
and ANSP EBRs can route traffic without further mapping lookups or
re-encapsulation at intermediate EBRs. Other enterprise networks
that have the aero network as a common parent may not have any
knowledge of each other's interior routing but will merely forward
packets on a default route up to the aero overlay.
The ATN is currently an OSI network but is projected to transition to
IPv6 over time. RANGER can bridge OSI networks together across the
IPv4 (or IPv6) Internet, or bridge IPv4 or IPv6 networks across an
OSI network. A pair of EBRs that have IP interfaces on a common
enterprise network (whether it is the Internet, the aero network, or
another parent or child enterprise network) can support
communications between their attached OSI edge networks by looking up
ISO network service access point (NSAP) addresses [IS8348] instead of
IP addresses for RLOC mappings. OSI ConnectionLess Network Protocol
(CLNP) [IS8473] packets can therefore be encapsulated within IPv4 (or
IPv6) headers for transmission across an Internet Protocol enterprise
network. Some OSI networks may transition to IPv6 addressing
[RFC4548] while applications are adapted by using RFC 2126 [RFC2126]
to carry OSI upper layers over TCP/IP, with the resulting IP packets
carried across and between RANGER enterprises in the normal way.
Another approach is to use subnetwork convergence to tunnel OSI
network protocol data units over Internet Protocol networks
[RFC1070].
Figure 15 depicts an ACSP and ANSP connected via an IPv4 aero
overlay. Host H represents a system onboard an aircraft that has a
wireless link to the ACSP, connected via an enterprise-edge network
interface on EBR F within the ACSP enterprise network. H resides on
an IPv6 edge network, and its EID is taken from the ACSP IPv6 prefix.
H needs to send a query to server S in the ANSP enterprise network.
H starts by sending a DNS query to the server at G, and in return it
receives the EID of server S. H then creates an IPv6 packet with
source EID(H) and destination EID(S) and forwards it to its default
router, F. F consults G for a mapping from EID(S) to the appropriate
RLOC. In this case, EBR F encapsulates the IPv6 packet in an IPv6
outer packet and forwards the packet to its default EBG, A. A
decapsulates the packet and looks up the destination EID(S) by
querying the DNS server at EBR B. B returns a mapping with the RLOC
of EBR E. A encapsulates the IPv6 inner packet in an IPv4 outer
packet with source RLOC(A) and destination RLOC(E). The packet is
forwarded via EBRs C and D in the aero overlay until it reaches E,
where it is decapsulated. E consults its cache of EID/RLOC mappings
and finds that the EBR for S is N. E encapsulates the packet in an
IPv6 packet with source RLOC(E) and destination RLOC(N). When the
packet reaches N, it is decapsulated, and the inner IPv6 packet is
forwarded on the edge network to the server, S.
_...--------------------------------------..._
/ (B) (D) \
( Aero Overlay (IPv4) )
-...________________________________________...-
. . .
(A) (C) .
. . .
_...------------------------.._ (E)
/ \ .
/ (F) \ .
( [H] ACSP (IPv6) ) .
\ (G) / .
\...__________________________... .
.
_...------------------------.._
/ \
/ (M) (N) \
( ANSP (IPv6) )
\ [S] /
\...__________________________...
Figure 15. Packet Forwarding for Aeronautical Networks
4.6.4. Unmanaged Networks
"Evaluation of IPv6 Transition Mechanisms for Unmanaged Networks"
[RFC3904] considers four cases for support of IPv6-enabled routers
and end systems connected to an ISP network via a gateway:
a. a gateway that does not provide IPv6 at all;
b. a dual-stack gateway connected to a dual-stack ISP;
c. a dual-stack gateway connected to an IPv4-only ISP; and
d. a gateway connected to an IPv6-only ISP.
Case a is typified by the widespread practice of customer networks
using IPv4-only NAT boxes to connect to their service providers.
RANGER does not address this scenario directly; however, the Teredo
mechanism [RFC4380] can provide a sufficient solution in many cases.
Case d is a scenario that has not yet seen widespread adoption. In
this scenario, the customer network could be configured as IPv6 only,
and the deployment could be considered as an IPv6-only extension to a
RANGER enterprise-edge network. End systems in this scenario would
still require support for legacy IPv4-only applications, and if the
customer network contained IPv4-only routers and end systems the
RANGER encapsulation mechanisms would still apply.
Cases b and c correspond to the scenario of the customer gateway to
the ISP becoming an IPv6 router. In that case, the gateway could
become a RANGER EBR, and the scenario becomes the same as the SOHO
network use cases discussed in Section 4.3. In particular, when
traditional home network IPv4 NAT boxes are updated to also support
IPv6 routing, the NAT box becomes a RANGER EBR.
5. Mapping and Encapsulation Concerns
Mapping and encapsulation concerns related to RANGER have been
discussed in Section 3.7 of [RFC5720]. These include effects of
mapping systems to application traffic, the need to secure the
mapping system, MTU effects, and the ability of legacy Internet
networks to connect to those employing RANGER.
6. Problem Statement and Call for Solutions
The scenarios discussed in this document have not closely examined
future growth of the native IPv6 and IPv4 Internets independently of
any growth in RANGER overlay networking. For example, it is likely
that current-day major Internet services that support millions of
customers simultaneously (e.g., Google, Yahoo, eBay, Amazon, etc.)
will continue to be served best by native Internet routing and
addressing rather than by overlay network arrangements that require
dynamic mapping state coordination. At the same time, however, more
and more small end user networks will wish to use provider-
independent addressing for multihoming via multiple ISPs as well as
support traffic engineering and mobility management.
These requirements call for an overlay network solution that is
compatible with both RANGER and the IPv6 and IPv4 native Internet
routing system without adversely affecting Internet routing scaling.
The solution must avoid the mapping and encapsulation concerns
discussed in Section 3.7 of [RFC5720]; for example, it must provide
generally shortest path routing without imparting unacceptable delays
for initial packets. The solution must further provide mobility
management capabilities for mobile end user networks that can take
advantage of route optimization while requiring no modifications to
end systems. Finally, the solution must be based on a business model
that allows end user networks to obtain Internet access services from
multiple ISPs simultaneously with support for traffic engineering and
fault tolerance.
7. Summary
The Internet today can be considered as a giant enterprise network,
with nodes in the Internet addressed from the public IPv4 address
space as RLOCs. Due to the 32-bit addressing limitations of IPv4,
however, continued expansion has occurred through the widespread
deployment of IPv4 Network Address Translators (NATs) while IPv6 has
yet to see wide adoption.
In many senses, however, this has resulted in a degenerate
manifestation of the network-of-networks model originally envisaged,
e.g., in the Catenet model. Indeed, these NATed domains have the
external appearance of being a simple host within the global Internet
RLOC space even though they may be proxying for arbitrarily large
networks of end systems. The end result is a loss of transparency in
the end-to-end model; it is no longer true that any node in the
Internet can directly address any other node.
RANGER enables a true network-within-network (or enterprise-within-
enterprise) framework. This is true even across a wide array of
deployment scenarios as documented here, and even for networks-
within-networks that may be recursively nested to an arbitrary depth.
RANGER therefore brings a unifying architecture applied consistently
across all layers of recursion, rather than a mixed bag of point
solutions that may or may not be mutually compatible. When coupled
with an overlay network solution that supports coexistence with the
IPv6 and IPv4 native Internet routing systems, a unified future
Internet architecture is possible.
8. Security Considerations
Security considerations are addressed in [RFC5720], [RFC5558], and
[RFC5320]. While the RANGER architecture does not in itself address
security considerations, it proposes an architectural framework for
functional specifications that do. Security concerns with tunneling,
along with recommendations that are compatible with the RANGER
architecture, are found in [TUNNEL-SEC]. Security considerations for
specific use cases are discussed there.
9. Acknowledgements
This work was inspired by the original architectural principles of
the Internet supplemented with "lessons learned" by many peers from
actual Internet deployments and experience developing encapsulation
protocols. The editors acknowledge that they are furthering work
initiated by many.
10. References
10.1. Normative References
[RFC0791] Postel, J., "Internet Protocol", STD 5, RFC 791,
September 1981.
[RFC2460] Deering, S. and R. Hinden, "Internet Protocol, Version 6
(IPv6) Specification", RFC 2460, December 1998.
[RFC5720] Templin, F., "Routing and Addressing in Networks with
Global Enterprise Recursion (RANGER)", RFC 5720, February
2010.
10.2. Informative References
[APT] Jen, D., Meisel, M., Massey, D., Wang, L., Zhang, B., and
L. Zhang, "APT: A Practical Transit Mapping Service",
Work in Progress, November 2007.
[BEHAVE-v6v4]
Baker, F., Li, X., Bao, C., and K. Yin, "Framework for
IPv4/IPv6 Translation", Work in Progress, August 2010.
[BELL-LaPADULA]
Bell, D. and L. LaPadula, "Secure Computer Systems:
Mathematical Foundations and Model", October 1974.
[CATENET] Pouzin, L., "A Proposal for Interconnecting Packet
Switching Networks", May 1974.
[CPE-RTRS] Singh, H., Beebee, W., Donley, C., Stark, B., and O.
Troan, Ed., "Basic Requirements for IPv6 Customer Edge
Routers", Work in Progress, December 2010.
[GROW-VA] Francis, P., Xu, X., Ballani, H., Jen, D., Raszuk, R.,
and L. Zhang, "FIB Suppression with Virtual Aggregation",
Work in Progress, August 2010.
[HUSTON-END]
Huston, G., "The End of the (IPv4) World is Nigher!",
July 2007.
[IEN48] Cerf, V., "The Catenet Model for Internetworking", July
1978.
[INCR-CGN] Jiang, S., Guo, D., and B. Carpenter, "An Incremental
Carrier-Grade NAT (CGN) for IPv6 Transition", Work in
Progress, March 2009.
[IPv4POOL] Hain, T., "The IPv4 Address Pool Projection", April 2009.
[IS8348] International Organization for Standardization,
International Electrotechnical Commission, "Open Systems
Interconnection -- Network service definition", 2002.
[IS8473] International Organization for Standardization,
International Electrotechnical Commission, "Protocol for
providing the connectionless-mode network service:
Protocol specification", 1998.
[LISP] Farinacci, D., Fuller, V., Meyer, D., and D. Lewis,
"Locator/ID Separation Protocol (LISP)", Work in
Progress, March 2009.
[RADIR-PROB-STATE]
Narten, T., "On the Scalability of Internet Routing",
Work in Progress, February 2010.
[RFC1035] Mockapetris, P., "Domain names - implementation and
specification", STD 13, RFC 1035, November 1987.
[RFC1070] Hagens, R., Hall, N., and M. Rose, "Use of the Internet
as a subnetwork for experimentation with the OSI network
layer", RFC 1070, February 1989.
[RFC1122] Braden, R., Ed., "Requirements for Internet Hosts -
Communication Layers", STD 3, RFC 1122, October 1989.
[RFC1380] Gross, P. and P. Almquist, "IESG Deliberations on Routing
and Addressing", RFC 1380, November 1992.
[RFC1753] Chiappa, N., "IPng Technical Requirements Of the Nimrod
Routing and Addressing Architecture", RFC 1753, December
1994.
[RFC1918] Rekhter, Y., Moskowitz, B., Karrenberg, D., de Groot, G.,
and E. Lear, "Address Allocation for Private Internets",
BCP 5, RFC 1918, February 1996.
[RFC1955] Hinden, R., "New Scheme for Internet Routing and
Addressing (ENCAPS) for IPNG", RFC 1955, June 1996.
[RFC2126] Pouffary, Y. and A. Young, "ISO Transport Service on top
of TCP (ITOT)", RFC 2126, March 1997.
[RFC2131] Droms, R., "Dynamic Host Configuration Protocol",
RFC 2131, March 1997.
[RFC2529] Carpenter, B. and C. Jung, "Transmission of IPv6 over
IPv4 Domains without Explicit Tunnels", RFC 2529, March
1999.
[RFC2767] Tsuchiya, K., Higuchi, H., and Y. Atarashi, "Dual Stack
Hosts using the "Bump-In-the-Stack" Technique (BIS)",
RFC 2767, February 2000.
[RFC2775] Carpenter, B., "Internet Transparency", RFC 2775,
February 2000.
[RFC3194] Durand, A. and C. Huitema, "The H-Density Ratio for
Address Assignment Efficiency An Update on the H ratio",
RFC 3194, November 2001.
[RFC3315] Droms, R., Ed., Bound, J., Volz, B., Lemon, T., Perkins,
C., and M. Carney, "Dynamic Host Configuration Protocol
for IPv6 (DHCPv6)", RFC 3315, July 2003.
[RFC3775] Johnson, D., Perkins, C., and J. Arkko, "Mobility Support
in IPv6", RFC 3775, June 2004.
[RFC3904] Huitema, C., Austein, R., Satapati, S., and R. van der
Pol, "Evaluation of IPv6 Transition Mechanisms for
Unmanaged Networks", RFC 3904, September 2004.
[RFC4029] Lind, M., Ksinant, V., Park, S., Baudot, A., and P.
Savola, "Scenarios and Analysis for Introducing IPv6 into
ISP Networks", RFC 4029, March 2005.
[RFC4038] Shin, M-K., Ed., Hong, Y-G., Hagino, J., Savola, P., and
E. Castro, "Application Aspects of IPv6 Transition",
RFC 4038, March 2005.
[RFC4057] Bound, J., Ed., "IPv6 Enterprise Network Scenarios",
RFC 4057, June 2005.
[RFC4192] Baker, F., Lear, E., and R. Droms, "Procedures for
Renumbering an IPv6 Network without a Flag Day",
RFC 4192, September 2005.
[RFC4215] Wiljakka, J., Ed., "Analysis on IPv6 Transition in Third
Generation Partnership Project (3GPP) Networks",
RFC 4215, October 2005.
[RFC4271] Rekhter, Y., Ed., Li, T., Ed., and S. Hares, Ed., "A
Border Gateway Protocol 4 (BGP-4)", RFC 4271, January
2006.
[RFC4301] Kent, S. and K. Seo, "Security Architecture for the
Internet Protocol", RFC 4301, December 2005.
[RFC4380] Huitema, C., "Teredo: Tunneling IPv6 over UDP through
Network Address Translations (NATs)", RFC 4380, February
2006.
[RFC4472] Durand, A., Ihren, J., and P. Savola, "Operational
Considerations and Issues with IPv6 DNS", RFC 4472, April
2006.
[RFC4548] Gray, E., Rutemiller, J., and G. Swallow, "Internet Code
Point (ICP) Assignments for NSAP Addresses", RFC 4548,
May 2006.
[RFC4795] Aboba, B., Thaler, D., and L. Esibov, "Link-local
Multicast Name Resolution (LLMNR)", RFC 4795, January
2007.
[RFC4852] Bound, J., Pouffary, Y., Klynsma, S., Chown, T., and D.
Green, "IPv6 Enterprise Network Analysis - IP Layer 3
Focus", RFC 4852, April 2007.
[RFC4862] Thomson, S., Narten, T., and T. Jinmei, "IPv6 Stateless
Address Autoconfiguration", RFC 4862, September 2007.
[RFC5214] Templin, F., Gleeson, T., and D. Thaler, "Intra-Site
Automatic Tunnel Addressing Protocol (ISATAP)", RFC 5214,
March 2008.
[RFC5320] Templin, F., Ed., "The Subnetwork Encapsulation and
Adaptation Layer (SEAL)", RFC 5320, February 2010.
[RFC5558] Templin, F., Ed., "Virtual Enterprise Traversal (VET)",
RFC 5558, February 2010.
[RFC5572] Blanchet, M. and F. Parent, "IPv6 Tunnel Broker with the
Tunnel Setup Protocol (TSP)", RFC 5572, February 2010.
[RFC5579] Templin, F., Ed., "Transmission of IPv4 Packets over
Intra-Site Automatic Tunnel Addressing Protocol (ISATAP)
Interfaces", RFC 5579, February 2010.
[RFC5887] Carpenter, B., Atkinson, R., and H. Flinck, "Renumbering
Still Needs Work", RFC 5887, May 2010.
[RFC5944] Perkins, C., Ed., "IP Mobility Support for IPv4,
Revised", RFC 5944, November 2010.
[RFC6115] Li, T., Ed., "Recommendation for a Routing Architecture",
RFC 6115, February 2011.
[STEP] Savola, P., "Simple IPv6-in-IPv4 Tunnel Establishment
Procedure (STEP)", Work in Progress, January 2004.
[TUNNEL-SEC]
Krishnan, S., Thaler, D., and J. Hoagland, "Security
Concerns With IP Tunneling", Work in Progress, October
2010.
Authors' Addresses
Steven W. Russert (editor)
1078 Ridge Crest Dr.
Wenatchee, WA 98801
USA
EMail: russerts@hotmail.com
Eric W. Fleischman (editor)
Boeing Research & Technology
P.O. Box 3707 MC 7L-49
Seattle, WA 98124
USA
EMail: eric.fleischman@boeing.com
Fred L. Templin (editor)
Boeing Research & Technology
P.O. Box 3707 MC 7L-49
Seattle, WA 98124
USA
EMail: fltemplin@acm.org