Download This critique of LISP concerns LISP+ALT

Two critiques of the LISP-ALT scalable routing proposal
for the ITRF Routing Research Group
2010-01-19 Robin Whittle
First a 747 word version. Then chopping it back to meet the
500 word limit for the "critique" section of the RRG report.
747 words
This critique concerns LISP+ALT. LISP+NERD could scale
to ~10^7 EIDs, but its full-database ITRs would be more
expensive and so less numerous than the caching ITRs with
local full-database query servers which are used in APT and
Ivip.
ALT is a mapping distribution system with globally
distributed query servers: ETRs and Map Servers. The ALT
network is an overlay built with existing tunnel and router
elements. A test network has been built with relative ease
and there are several efforts to write interoperable
implementations of ITRs, ETRs, Map Servers and Map
Resolvers.
A fundamental problem with any global query server
network such as ALT is that the delays inherent in this
approach (with frequently long paths and greater risk of lost
queries or responses) mean that ITRs will drop or
significantly delay the initial packets of many new sessions.
ITRs drop the packet(s) they have no mapping for. After the
mapping arrives, the ITR waits for a resent packet (assuming
the sending host is not trying instead to contact another host
in a different EID prefix) and will then tunnel that packet
correctly. These "initial packet delays" reduce performance
and so create a major barrier to voluntary adoption on wide
enough basis to solve the routing scaling problem.
ALT’s delays are compounded by its structure being
"aggressively aggregated" according to address, without
regard to the geographic or topological location of the routers.
So the tunnels between ALT routers will often span large
geographic distances and traverse many Internet routers.
Therefore, the many levels to which a query typically
ascends in the ALT hierarchy before descending towards its
destination router will too often involve very long geographic
paths and so worsen delays and packet loss rates.
No solution has been proposed for these problems or for the
contradiction between the need for high aggregation while
making the ALT structure robust against single points of
failure. Initial packet delays can only be made insignificant
with NERD or local full-database query servers.
For LISP’s ITRs to perform multihoming service restoration,
they must determine reachability of end-user networks via
two or more ETRs. The individual efforts of large numbers
of ITRs are inefficient and potentially burdensome on the
ETRs.
Testing reachability of the ETRs is complex and costly - and
insufficient. ITRs cannot test network reachability via each
ETR, since the ITRs have no address of a device in that
network. So ETRs must test network reachability and
convey this to ITRs.
LISP involves complex communication between ITRs and
ETRs, with UDP and variable-length LISP headers in all
traffic packets. The ITR's algorithm for solving the PMTUD
problems caused by encapsulation is incomplete and may be
expensive to implement securely.
The advantage of LISP+ALT is that its ability to handle
billions of EIDs is not constrained by the need to transmit or
store the mapping to any one location. Such numbers,
beyond a few tens of millions of EIDs, will only result if the
system is used for Mobility. Yet the concerns just mentioned
about ALT’s structure arise from the millions of ETRs which
would be needed just for non-mobile networks. (Map
Servers may reduce total path lengths somewhat.)
In LISP’s mobility approach each MN needs an RLOC
address to be its own ETR, meaning the MN cannot be
behind NAT. This double address use is unsuitable for IPv4.
Lisp-mn requires instant mapping changes being sent to all
relevant ITRs every time the MN gets a new address - which
LISP cannot achieve. However, LISP could support the TTR
Mobility architecture which does not require mapping
changes to be frequent or instantly achieved.
In order to enforce ISP filtering of incoming packets by
source address, LISP ITRs would have to implement the
same filtering on each decapsulated packet. This is extremely
expensive at high data rates for large numbers of prefixes and is normally done with TCAM hardware.
LISP monolithically integrates multihoming failure detection
and restoration decision-making processes into the core-edge
separation scheme itself. End-user networks must rely on the
necessarily limited capabilities which are built into every ITR.
These functions could be externalised and made the
responsibility of end-user networks if LISP was able to
distribute mapping in real-time to all ITRs which need it.
However this is not practical without full database local
query servers.
LISP-ALT may be able to solve the routing scaling problem,
but alternative approaches would be superior because they
eliminate the initial packet delay problem and give end-user
networks real-time control over ITR tunneling.
Chopping it back . . .
LISP-ALT uses a mapping distribution system with globally
distributed query servers: ETRs and Map Servers.
A fundamental problem with any global query server
network is that the frequently long paths and greater risk of
packet loss cause ITRs to drop or significantly delay the
initial packets of many new sessions. ITRs drop the
packet(s) they have no mapping for. After the mapping
arrives, the ITR waits for a resent packet and will tunnel that
packet correctly. These "initial packet delays" reduce
performance and so create a major barrier to voluntary
adoption on wide enough basis to solve the routing scaling
problem.
ALT’s delays are compounded by its structure being
"aggressively aggregated", without regard to the geographic
location of the routers. The tunnels between ALT routers
will often span intercontinental distances and traverse many
Internet routers.
The many levels to which a query typically ascends in the
ALT hierarchy before descending towards its destination will
often involve excessively long geographic paths and so
worsen initial packet delays.
No solution has been proposed for these problems or for the
contradiction between the need for high aggregation while
making the ALT structure robust against single points of
failure.
For LISP’s ITRs to perform multihoming service restoration,
they must determine reachability of end-user networks via
two or more ETRs. The individual efforts of large numbers
of ITRs are inefficient and may overburden ETRs.
Testing reachability of the ETRs is complex and costly - and
insufficient. ITRs cannot test network reachability via each
ETR, since the ITRs have no address of a device in that
network. So ETRs must report network un-reachability to
ITRs.
LISP involves complex communication between ITRs and
ETRs, with UDP and variable-length LISP headers in all
traffic packets.
The advantage of LISP+ALT is that its ability to handle
billions of EIDs is not constrained by the need to transmit or
store the mapping to any one location. Such numbers,
beyond a few tens of millions of EIDs, will only result if the
system is used for Mobility. Yet the concerns just mentioned
about ALT’s structure arise from the millions of ETRs which
would be needed just for non-mobile networks.
In LISP’s mobility approach each MN needs an RLOC
address to be its own ETR, meaning the MN cannot be
behind NAT. Mapping changes must be sent instantly to all
relevant ITRs every time the MN gets a new address - which
LISP cannot achieve.
In order to enforce ISP filtering of incoming packets by
source address, LISP ITRs would have to implement the
same filtering on each decapsulated packet. This may be
prohibitively expensive.
LISP monolithically integrates multihoming failure detection
and restoration decision-making processes into the core-edge
separation scheme itself. End-user networks must rely on the
necessarily limited capabilities which are built into every ITR.
LISP-ALT may be able to solve the routing scaling problem,
but alternative approaches would be superior because they
eliminate the initial packet delay problem and give end-user
networks real-time control over ITR tunneling.
497 words
LISP-ALT uses a mapping distribution system with globally
distributed query servers: ETRs and Map Servers.
A fundamental problem with any global query server
network is that the frequently long paths and greater risk of
packet loss cause ITRs to drop or significantly delay the
initial packets of many new sessions. ITRs drop the
packet(s) they have no mapping for. After the mapping
arrives, the ITR waits for a resent packet and will tunnel that
packet correctly. These "initial packet delays" reduce
performance and so create a major barrier to voluntary
adoption on wide enough basis to solve the routing scaling
problem.
ALT’s delays are compounded by its structure being
"aggressively aggregated", without regard to the geographic
location of the routers. The tunnels between ALT routers
will often span intercontinental distances and traverse many
Internet routers.
The many levels to which a query typically ascends in the
ALT hierarchy before descending towards its destination will
often involve excessively long geographic paths and so
worsen initial packet delays.
No solution has been proposed for these problems or for the
contradiction between the need for high aggregation while
making the ALT structure robust against single points of
failure.
For LISP’s ITRs to perform multihoming service restoration,
they must determine reachability of end-user networks via
two or more ETRs. The individual efforts of large numbers
of ITRs are inefficient and may overburden ETRs.
Testing reachability of the ETRs is complex and costly - and
insufficient. ITRs cannot test network reachability via each
ETR, since the ITRs have no address of a device in that
network. So ETRs must report network un-reachability to
ITRs.
LISP involves complex communication between ITRs and
ETRs, with UDP and variable-length LISP headers in all
traffic packets.
The advantage of LISP+ALT is that its ability to handle
billions of EIDs is not constrained by the need to transmit or
store the mapping to any one location. Such numbers,
beyond a few tens of millions of EIDs, will only result if the
system is used for Mobility. Yet the concerns just mentioned
about ALT’s structure arise from the millions of ETRs which
would be needed just for non-mobile networks.
In LISP’s mobility approach each MN needs an RLOC
address to be its own ETR, meaning the MN cannot be
behind NAT. Mapping changes must be sent instantly to all
relevant ITRs every time the MN gets a new address - which
LISP cannot achieve.
In order to enforce ISP filtering of incoming packets by
source address, LISP ITRs would have to implement the
same filtering on each decapsulated packet. This may be
prohibitively expensive.
LISP monolithically integrates multihoming failure detection
and restoration decision-making processes into the core-edge
separation scheme itself. End-user networks must rely on the
necessarily limited capabilities which are built into every ITR.
LISP-ALT may be able to solve the routing scaling problem,
but alternative approaches would be superior because they
eliminate the initial packet delay problem and give end-user
networks real-time control over ITR tunneling.