Transcript COM594: Mobile Technologies

COM594: Mobile Technologies
Location-Identifier Separation
In the current Internet TCP/IP Protocol Stack, the IP address functions
simultaneously as:
• A Routing Locator (an identifier with a topological meaning)
and
• An Endpoint Identifier
Application Layer
Transport Layer
IP-address,, port
(Endpoint Identifier)
Internet Layer
IP-address
(Routing Locator)
Data Link Layer
Physical Layer
Location-Identifier Separation
• When a host changes its point of
attachment to the Internet, its IP address
must change as well
• Therefore all transport sessions will break
• There have been various ‘workarounds’ to
implement mobility on the Internet
Location-Identifier Separation
• More fundamental approaches aim to
separate Routing Locators and Endpoint
Identifiers to remove all identification
related functionality from topology related
information such as IP addresses.
• Current implementations adopt IP as the
communications endpoint.
• New approaches have become known as
Location-Identifier (L.I.) Separation.
Location-Identifier Separation
• These more fundamental approaches
require the redesign of the Internet
protocol stack:
• Such proposals are, by definition radical,
and will be difficult to implement.
(Recall IPv4 vs Ipv6)
• The ideas have emerged from the Routing
Research Group (RRG) of the Internet
Research Task Force (IRTF)
L.I Separation
• The lack of L.I. Separation causes
problems beyond simply mobility:
• A key issue is user location privacy
• When
– Identifiers are long lived, and
– A publicly available mapping exists between
identifiers and locators,
• it is possible to determine the location of a host
and thus, the user using it
• Without the user’s permission, or knowledge
Privacy
• Invasion of privacy is increasingly
becoming a criminal offence!
• It is important that new mechanisms, by
default, do not reveal the location of a
particular host to unknown observers.
Proposed Solutions
• Four proposed solutions have exercised
the RRG:
• The Host Identity Protocol (HIP)
• Network Address Translation for IPv6 to
IPv6 (NAT66)
• Identifier-Locator Network Protocol (ILNP)
• Location-Identifier Separation ProtocolMobile Node (LISP-MN)
Fundamental Approaches to LocationIdentifier Separation
• Most approaches to LI Separation fall into
two broad categories:
– Those that introduce an extra layer to hold the
original endpoint identifiers
– Those that split the IPv6 address space into a
part that has topological meaning, and a part
that is used to identify the host.
Case Study
• All four approaches have their advocates
and detractors.
• None of them are ‘perfect’.
• All are ‘work in progress’
• We will briefly review LISP-MN as this has
gained significant industrial support in
recent years.
Location-Identifier Separation Protocol – Mobile Node
(LISP-MN)
• The LISP-NM Protocol enables a mobile
node to roam across network whilst
retaining its IP address.
• During hand-off, sessions may ‘pause’,
and some data loss is possible.
• The key issue however is that sessions
are not dropped.
• So they do not have to be set up again
LISP-MN
• LISP-MN aims to make it possible for
mobile devices to roam while keeping TCP
sessions alive and to be simultaneously
connected to two different networks.
(Multihomed).
• LISP-MN is based on a LISP
infrastructure:
LISP
• LISP implements a Map-and-Encap
scheme.
• Packets are encapsulated at the border
router of the sender domain: The Ingress
Tunnel Router. (ITR)
• Packets are decapsulated at the border
router of the receiver domain: The Egress
Tunnel Router (ETR)
Encapsulation
• By this mechanism, core routing (routing
between domains) is independent of the
encapsulated endpoint identifiers.
• This also optimizes routing for the
topological characteristics of the core
network.
• LISP adds an extra Internet layer below
the existing one:
LISP Stack
Application Layer
Identifier
Transport Layer
Identifier
Internet Layer
Identifier
Internet Layer
Locator
Data Link Layer
Physical Layer
LISP Stack
15
LISP Transmission
1. The Host looks up the correspondent host in a DNS
and gets an Endpoint Identifier;
2. Host makes a packet with it source Endpoint Identifier
and the Destination Endpoint Identifier
3. Packet is sent to the ITR which encapsulates it with the
Routing Locator of the ITR as the source, and the
Routing Locator of an ETR as the target. (This requires
a mapping mechanism)
4. The packet is transmitted over the Internet to the ETR
5. The ETR decapsulates the packet and sends it to the
destination Endpoint Identifier
Typical LISP Scenario
RLOC ITR1:
10.0.0.0/8
RLOC ETR1:
12.0.0.0/8
EID:
1.0.0.0/8
EID:
2.0.0.0/8
Internet Core
EID:
1.0.0.1
1.0.0.1 -> 2.0.0.2
Host EID 1.0.0.1 wants
To send to Host EID
2.0.0.2
The packet
Arrives at
ITR2
RLOC ITR2:
11.0.0.0/8
ITR2 encapsulates
The packet with source
1.0.0.1 and Destination
2.0.0.2 in a packet
RLOC ETR2:
With source 11.0.0.1
13.0.0.0/8
And destination
12.0.0.2
11.0.0.1 -> 12.0.0.2
11.0.0.1 -> 12.0.0.2
1.0.0.1 -> 2.0.0.2
1.0.0.1 -> 2.0.0.2
ITR2 does a DNS on 2.0.0.2
and gets13.0.0.2 and 12.0.0.2:
The latter has priority
EID:
2.0.0.2
1.0.0.1 -> 2.0.0.2
ETR! Forwards the packet
To EID 2.0.0.2
ETR1 receives the packet
And decapsulates it.
17
LISP-MN
• LISP-MN leverages the mapping
infrastructure of LISP to support mobile
devices
• This happens by turning the mobile device
into a LISP ITR and ETR for itself
• The mobile device sends map requests
• All packets originating at the mobile device
are LISP encapsulated
Map Servers
• The mobile device can answer directly to
incoming Map requests, or it can
designate its map server as a proxy
• Map Servers have similar behaviour to
Home Agents in Mobile IP
• Unlike mobile IP, the actual data never
flows through these servers.
• They just answer to the mapping requests.
• Also, home agents never provide mapping
information because that is left to the
mobile node
Example: EID 1.0.0.1 wants to send a packet to EID 1.0.0.2
Mobile host 1.0.0.2 has lost its Wi Fi connection but still has GSM
1. Mobile node updates the Mapping Server to indicate that it is accessible via 13.0.0.2, but not 12.0.0.2
2. The packet arrives at ITR2 (Which has Routing Locator 11.0.0.1)
3. ITR2 Looks up Routing Locators corresponding with EID1.0.0.2 and finds 13.0.0.2
4. ITR2 encapsulates the packet and forwards as normal over the Internet core
5. The mobile host receives the packet and decapsulates it.
WiFi:
12.0.0.0/8
Mapping Server
RLOC ITR1:
10.0.0.0/8
Domain EID:
1.0.0.0/8
WiFi
RLOC
Host
12.0.0.2
--------
13.0.0.2
1.0.0.2
Internet Core
3G
Source EID:
1.0.0.1
1.0.0.1 -> 1.0.0.2
RLOC ITR2:
11.0.0.0/8
3G:
13.0.0.0/8
Dest EID:
1.0.0.2
11.0.0.1 -> 13.0.0.2
11.0.0.1 -> 13.0.0.2
11.0.0.1 -> 13.0.0.2
1.0.0.1 -> 1.0.0.2
1.0.0.1 -> 1.0.0.2
1.0.0.1 -> 1.0.0.2
20
Benefits and Challenges
• LISP implements the separation of
Routing Locators and Endpoint Identifiers
without the need for changes at the host.
• The address in the core network is
independent from that at the edge, so for
example, the core network could use IPv6,
whereas the edge network would use IPv4
and vice-versa.
Benefits and Challenges
• By turning the mobile node into a LISTsite-in-a-box, the MN can change their
point of attachment without breaking
transport session.
• Unfontunatelly, the latter does require
changes at the host, undoing one of the
advantages of LISP.
Benefits and Challenges
• The largest obstacle appears to be the
requirement for an operational LISP
infrastructure. Until LISP is widely
deployed, the benefits of LISP-MN are
small.
Useful References
• CISCO Demo LISP_MN
– http://bit.ly/oYa2IE
– http://www.cisco.com/c/en/us/products/ios-nxos-software/locator-id-separation-protocollisp/index.html
– https://lispmob.org/
– http://lisp.cisco.com
• LISP Mobile Project
(this is just a link of interest)
– http://www.lispmob.org/