Home Agent Reliability Protocol draft-ietf-mip6-hareliability-00.txt
Shared by: xiaopangnv
-
Stats
- views:
- 1
- posted:
- 11/8/2011
- language:
- English
- pages:
- 10
Document Sample


Home Agent Reliability
Protocol
draft-ietf-mip6-hareliability-00.txt
HA Reliability DT
Status
2006 March DT was formed
2006 June
draft-ietf-mip6-hareliability-00.txt was submitted
Any comments are appreciated.
Goal of this work
No reliability scheme is provided to HA, though
MIP6 and NEMO rely on home agent availability
Hardware based solutions are still possible, but a protocol
based solution is also required
Providing redundancy and reliability to home agent
without the mobile node being aware of it
Two modes are proposed in the draft
Home Agent Virtual Switch
Home Agent Hard Switch
Home Agent Virtual Switch
Assumptions
The IPsec states are
backup
synchronized among a
2001::a 2001::a
redundancy home agent set
(HA1 and HA2) HA1 HA2
The HA address is shared 2) SA and Binding
States Exchange
among the redundancy home
agent set
1) MN establishes IPsec SA with 1) BU
HA and sends BU
2) HA exchange binding and
IPsec SA states with backup-HA
3) When the HA fails, backup-HA
automatically takes over the How to synchronize IPsec is out of scope
failed HA
Home Agent Hard Switch
Assumptions
IPsec SA states are not synchronized
among a redundancy home agent set backup
Each HA has distinct IP address 2001::a 2001::b
1) MN pre-establishes IPsec SAs with all HA1 HA2
the HAs 3) Binding States
2) MN sends BU to one of HA as Exchange
RFC3775
3) Binding States are exchanged among
HAs
4) If HA is failed, backup-HA sends a HA 2)BU 5)BU
switch message to MNs by using the pre-
established SA.
4)HA switch
5) MN then re-associates with the backup
HA by:
sending BU to backup-HA
update binding information and tunnel 1) IPsec SA pre-establishment
end points by trigger of the received HA to both HAs
switch message
Protocol Operation between home agents
(active)
HELLO messages
exchange HA information HA1 HA2
HELLO
heartbeat for failure detection
Switch Over Req/Rep
messages (active)
are used by a standby HA to
become an active HA HA1 HA2
Switchover
(active)
Switch Back Req/Rep
messages
are used by an active HA to (active)
become an inactive HA
The receiver HA will be active HA1 HA2
SwitchBack
(active)
HA Discovery
For HA virtual switch, it follows bootstrapping
scheme
For HA hard switch, MN must acquires multiple HA
addresses in order to establish SA beforehand.
Split scenario: MN query DNS SRV records with service
name and DNS returns multiple HA addresses
Integrated scenario: DHCP server replies with multiple HA
addresses.
No technical changes, but only operational changes
are described.
Messages and Options
New Mobility Header Messages
Home Agent Hello Request Message
Home Agent Hello Message
State Synchronization Request Message
State Synchronization Message
Home Agent SwitchOver Request Message
Home Agent SwitchOver Reply Message
Home Agent SwitchBack Request Message
Home Agent SwitchBack Reply Message
New Mobility Options
Binding Cache Information Option
AAA Information Option
Vendor Specific Information Option .
To be addressed
Which states should be synchronized among HAs
binding states, AAA information, vendor specific
information
Optimized HA switch procedure
Can MN updates binding states when a HA switch
message is received, instead of sending BU?
RFC3776 states that IPsec SA is updated when a BU is
received
May need to update draft-ietf-mip6-ha-switch-00.txt
Conclusion
This draft addresses HA reliability solution providing
Inter HAs messages such as HELLO, SwitchBack, SwichOver
messages
State Synchronization scheme
Failure Detection scheme
Failure Recovery scheme
Multiple HA discovery (for Hard Switch Mode only)
The design team is still working on the solution
please send comments to MIP6 list!!!
Get documents about "