Transcript Slayt 1 - BSDConTR

SCTP: next generation signalling protocol
and FreeBSD 7
Murat Balaban
Director, R & D
endersys ltd.
http://www.enderunix.org/murat/
SCTP Intro
• new IP transport protocol, existing at an
equivalent level as UDP and TCP
• provide transport layer functions to all of
the main Internet applications
• IETF standart
• RFC 2960 – RFC 3758
Like TCP
•
•
•
•
•
provides a reliable transport service
In-order delivery
connection-oriented
Congestion avoidance
Slow start
Unlike TCP
• provides a number of functions that are
considered critical for signaling transport
• at the same time can provide transport
benefits to other applications requiring
additional performance and reliability.
SCTP Core Features
• Multi-streaming
• Multi-homing
Multi-streaming
• Name is derived from the multi-streaming
function
• In contrast, TCP provides a single stream
of data and ensures that delivery of that
stream takes place with perfect sequence
preservation
• For a number of applications, this
characteristic of strict sequence
preservation is not truly necessary
Multi-streaming contd.
• Signalling
Multiple streams can carry different
signalls of varying importance
Multiple calls can be placed on different
streams whithin the same “association”
Multi-streaming contd.
• Web page objects
Objects can be placed on different
independent streams, which results in a
better perceived performance.
Multi-streaming: How?
By creating independence between data
transmission and data delivery.
In particular, each DATA “chunk” (or PDU) in the
protocol uses two sets of sequence numbers, a
Transmission Sequence Number that governs
the transmission of messages and the detection
of message loss, and the Stream ID/Stream
Sequence Number pair, which is used to
determine the sequence of delivery of received
data.
Multi-homing
• the ability for a single SCTP endpoint to support
multiple IP addresses
• Benefit: potentially greater survivability of the
session in the presence of network failures
• redundant LANs can be used to reinforce the
local access, while various options are possible
in the core network to reduce the dependency of
failures for different addresses
Multi-homing
App-3
App-2
App-1
OS
NI-1
NI-2
160.15.82.20
161.10.8.221
NI-3
10.1.61.11
Multi-homing contd.
• Use of addresses with different prefixes
can force routing to go through different
carriers, for example, while route-pinning
techniques or even redundant core
networks can also be used if there is
control over the network architecture and
protocols.
Multi-homing: load sharing?
• Nope.
• Only for redundancy
Multi-homing: load sharing?
• Nope.
• Only for redundancy
Multi-homing: How?
• SCTP endpoints can exchange lists of
addresses during initiation of the association
• A single port number is used across the entire
address list at an endpoint for a specific session.
• to reduce the potential for security problems, it is
required that some response messages be sent
specifically to the source address in the
message that caused the response
Other features
• a unicast protocol, and supports data
exchange between exactly 2 endpoints,
although these may be represented by
multiple IP addresses.
• provides reliable transmission
• Full-duplex
Other features contd.
• message oriented and supports framing of
individual message boundaries. In
comparison, TCP is stream oriented and
does not preserve any implicit structure
within a transmitted byte stream.
• Rate-adaptive
• Unordered data delivery option
Other features contd.
• Security cookie against connection flood
attack (SYN flood)
• Built-in heartbeat (reachability check)
• Extensibility (TLV chunks)
Association Establishment
• To establish this state, both sides go through a
specific set of exchanges
– TCP uses a 3-way handshake (SYN, SYN/ACK, ACK)
– SCTP uses a 4-way handshake
• In TCP, the communication relationship between
two endpoints is called a “connection”
• In SCTP, this is called an “association” this is
because it is a broader concept than a single
connection (i.e. multi-homing)
Setting Up an Association
Endpoint A
Endpoint Z
INIT
INIT-ACK
Association
Is Up
COOKIE-ECHO
*
*
COOKIE-ACK
* -- User data can be attached
Association
Is Up
SCTP States I
[Rcv INIT]
Gen Cookie
Send INIT-ACK
CLOSED
[ASSOCIATE]
Create TCB
Send INIT
Start init timer
[rcv valid COOKIE-ECHO]
Create TCB
Send COOKIE-ACK
COOKIE_WAIT
[rcv INIT-ACK]
Send COOKIE-ECHO
Stop init timer
Start cookie timer
COOKIE_ECHOED
ESTABLISHED
[rcv COOKIE-ACK]
Stop cookie timer
SCTP States II
ESTABLISHED
[SHUTDOWN]
Check outstanding
data chunks
SHUTDOWNPENDING
[No More Outstanding
data chunks]
Send SHUTDOWN
Start shutdown timer
NEXT-SLIDE
[rcv SHUTDOWN]
Check outstanding
data chunks
SHUTDOWNPENDING
[No More Outstanding
data chunks]
Send SHUTDOWN-ACK
Start shutdown- timer
NEXT-SLIDE
SCTP States III
FROM STATES-2
FROM STATES-2
SHUTDOWNSENT
SHUTDOWNACK-SENT
[rcv SHUTDOWN-ACK]
send SHUTDOWN_COMPETE
Stop shutdown timer
Delete TCB
[rcv SHUTDOWN-COMPLETE]
Stop shutdown timer
delete TCB
CLOSED
SCTP Header with IP
IP Header
SCTP Common Header
Chunk 1
...
Chunk N
Common Header
Source Port
Destination Port
Verification Tag
CRC-32c Checksum
SCTP Chunks
Chunk Type
Chunk Flags
Chunk Length
Chunk Data
• Chunk Type: 8-bit value indicating the type of chunk
• Chunk Flags: 8-bit flags, defined on per chunk type basis
• Chunk Length: 16-bit length in bytes, including the chunk
type, chunk flags, and chunk length fields.
–Note that chunks are padded to 32-bit boundaries within
an SCTP packet. Any padding bytes (0x00) used are
NOT included in the chunk length
SCTP Chunk Types
• There are 20 chunk types currently defined in SCTP (including nonRFC/Internet Draft extensions):
–(1) DATA (0x00)
–(2) INITIATION [INIT] (0x01)
–(3) INITIATION-ACKNOWLEDGMENT [INIT-ACK] (0x02)
–(4) SELECTIVE-ACKNOWLEDGMENT [SACK] (0x03)
–(5) HEARTBEAT (0x04)
Cookie Echo Chunk
Type=a
Flags=0
Length=variable
State Cookie from INIT-ACK
DATA Chunk
Type=0x00
Flags=UBE
Length=variable
TSN Value
Stream Identifier
Stream Sequence Num
Payload Protocol Identifier
Variable Length User Data
Flag Bits ‘UBE’ are used to indicate:
–U – Unordered Data
–B – Beginning of Fragmented Message
–E – End of Fragmented Message
• A user message that fits in one chunk would have both the B and E bits
set
FreeBSD ready?
• Not yet.
• -CURRENT
• Planned for 7.0-RELEASE
• Thanks to Randal Steward
-- Thanks --
Q&A
Murat Balaban
EnderUNIX Yazılım Geliştirme Takımı
[email protected]
http://www.enderunix.org/murat/