Professional Documents
Culture Documents
Level
USER GUIDE
© Ericsson GmbH 2018, 2019. All rights reserved. No part of this document may
be reproduced in any form without the written permission of the copyright owner.
Disclaimer
The contents of this document are subject to revision without notice due to
continued progress in methodology, design and manufacturing. Ericsson shall
have no liability for any error or damage of any kind resulting from the use of
this document.
Trademark List
All trademarks mentioned herein are the property of their respective owners.
These are shown in the document Trademark Information.
Contents
1 Revision Information 1
2 Overview 3
2.1 Scope 3
2.2 Introduction 3
2.2.1 Documentation Structure 3
3 General 5
3.1 Concepts 5
3.1.1 Application Server 5
3.1.2 Application Server Process 7
3.1.3 ESP 7
3.1.4 IP Load Balancing 8
3.1.5 IP-SEP 8
3.1.6 IPSP 8
3.1.7 Remote Host 8
3.1.8 Routing Context 8
3.1.9 Routing Key 8
6 How to Use 53
6.1 SCTP Parameter Settings 53
6.1.1 Parameters for Path Handling 53
6.1.2 Timers and Counters for Path Handling 55
6.1.3 Buffers and Windows 60
6.1.4 Protocol Data Unit 63
6.1.5 Number of Streams 64
6.1.6 SCTP Bundling 65
6.1.7 Zero Receiver Window Supervision 65
6.1.8 Tailoring of SCTP Parameters 66
6.2 Network Configuration Examples 73
6.2.1 Configuration in M3UA Network 73
6.2.2 IP-STP in SPX Connectivity 78
Glossary 83
Reference List 89
1 Revision Information
Other than editorial changes, this document has been revised as follows:
— Rev. A
• ‘‘System Improvement’’: Introduction of User Guide Managing the Stream
Control Transmission Protocol on MSC-S level.
• ‘‘System Improvement’’: Consolidation of Concepts section.
2 Overview
2.1 Scope
This document provides guidance for the transport of Signaling over SCTP/IP
in CNCS.
2.2 Introduction
This document guides the reader through the network-level aspects of configuring
Signaling over SCTP/IP and provides network design guidelines.
Section 4 on page 9 describes the basics of Signaling over SCTP/IP using concrete
scenarios.
By the time readers finish this document, they will have reached a good
understanding of the network level aspects. The next step is then to understand
the node level aspects for which this User Guide provides a rich reference list.
CNCS Level
User Guide Managing and Optimizing SS7 Congestion Control and SIGTRAN
Performance
Description Core Network Overview
User Guide Managing Security on Core Network Level
User Guide Managing Primary Rate Access on Core Network Level
MSC-S Level
User Guide Managing IP Integration
User Guide Managing the Stream Control Transmission Protocol
User Guide MSC-S Configuration Management
User Guide Message Transfer Part 3 User Adaptation Layer
User Guide M3UA Relay (SPX specific document)
User Guide Message Transfer Part 2 User Peer-to-Peer Adaptation Layer
User Guide Primary Rate Access in MSC Server
User Guide Gateway Control Protocol over SCTP
User Guide SGs-Interface Configuration and Enabling of SMS over
SGs-Interface and LTE to CS Fallback
M-MGw Level
User Guide for Signalling, Reference [42]
User Guide for Physical Interfaces, Reference [40]
User Guide for Primary Rate Access, Reference [41]
User Guide for Virtual Media Gateway Handling, Reference [43]
3 General
3.1 Concepts
This section lists the concepts specific for this document. See Description Concepts
and Terminology on Circuit Switched Core Network Level, User Guide Message
Transfer Part 3 User Adaptation Layer, and User Guide for Signalling, Reference
[42], for other CN and SS7 over IP-related abbreviations.
The following concepts are described in the Concepts and Terminology on Circuit
Switched Core Network Level:
— Signaling Mode
— SS7-SEP
An M3UA Application Server (AS) is a logical entity in an IP-SEP that receives and
sends MTP3 User messages with an MTP routing label corresponding to a specific
Routing Key.
IP-SEP
AS
Application
M3UA
SCTP
IP
IP-SEP
AS 1 AS 2 AS N
Application Application Application
1 2 N
M3UA
SCTP
IP
IP-SEP 1 IP-SEP 2
AS 1 AS 2
Application Application
M3UA M3UA
SCTP SCTP
IP IP
Figure 3 ASs in IPSP-IPSP Configuration
An AS in an IP-SEP is, for example, a switch element handling all call processing
for a signaling relation, identified by an SS7 Signaling Point Code (SPC). Another
example is a database element, handling all HLR transactions, which is identified
by a specific SS7 SPC.
3.1.1.2 IUA AS
3.1.3 ESP
An E-SIGTRAN Signaling Process (ESP) is a 3GPP TS 29.202, Reference [44],
compliant M3UA signaling process.
IP Load Balancing is a mechanism that allows a node to expose one or more virtual
IP addresses to external nodes, while enabling distributed node architecture
with multiple components sharing the virtual IP addresses for their external
connections. IP Load Balancing is only needed for distributed systems. The IP
Load Balancer (IPLB) distributes incoming IP datagrams to the application blades.
3.1.5 IP-SEP
An IP-SEP is a Signaling End Point (SEP) residing in the IP network, which uses
adaptation layer protocols such as M3UA and M2PA to transport application
signaling messages over SCTP.
3.1.6 IPSP
An IPSP is an IETF RFC 4666, Reference [47] compliant M3UA signaling
process which defines the behavior over an SCTP association of an AS in an
IP-SEP⇔IP-SEP communication scenario.
The Routing Context is used to determine an M3UA AS for incoming MTP3 User
messages. Use of the Routing Context is mandatory when an SCTP association is
configured to carry traffic for multiple ASs.
The Routing Key is used for routing MTP3 User messages in an SGW or IP-SEP.
SMS-SC/ gsmSCF
SCP HLR/AuC FNR LCS
SMS-C /CCN
INAP CAP
MAP
MME
SGsAP MAP, CAP, BSSAP,
BICC/ RANAP, ISUP, BICC and INAP
SGSN BSSAP+ ISUP
BSSAP STP/SGW/SRP
BSC MSC-S MSC-S
RANAP
PABX POI
M- M- ISUP
MGw MGw
CNCS Nodes VM
External nodes such as HLR/AuC, FNR, LCS, gsmSCF/CCN, SCP, and SMS Center
which do not convey user plane traffic can be connected directly to the MSC-S
using TDM, ATM, or IP transport.
Nodes such as BSC, RNC, SGSN, Voicemail (VM), and POI which convey signaling
and user plane traffic over TDM or ATM are preferably connected to MSC-S
through the M-MGw that in this case acts as SGW. This has the advantage that
signaling and user plane traffic types are directed to the common M-MGw node
simplifying cabling and hardware used.
One of the major benefits that Signaling over SCTP/IP offers, compared to TDM,
or ATM signaling, is that the available bandwidth is better utilized because it can
be accessed by all logical SCTP links and only when data packets are transmitted.
New high-speed IP interfaces connect the nodes directly to a site LAN or site
routers in the multi service IP network. Thanks to the bandwidth capabilities
of the IP interfaces, much less equipment needs to be deployed and signaling
capacity management is simplified.
IP-SEP
IP-SEP M3UA
IP SGW ATM/TDM
IP-SEP
A hierarchical structure can also be introduced within large networks for better
scalability. This means that any configuration changes in one SEP access segment
does not necessarily require reconfiguration in other SEP access segments.
Moreover, it can be used to centralize certain functions, such as Global Title
Translation (GTT), FNR, accounting, policing and Generic Message Screening and
rerouting (STP functions, SPC-based routing).
The different layers are now introduced which provide the foundation of
transporting Signaling Application Protocols over IP according to the framework
architecture defined in IETF RFC 2719, Reference [45].
Adaptation Layers
SCTP
IP
Figure 6 IP Signaling Transport Components
GCP and SGsAP are exemptions to the transport of Signaling Application protocols
as previously described: As described in Section 4.4 on page 23 and Section 4.3.1
on page 15, CNCS offers the option to transport GCP either directly over SCTP
or over M3UA. As described in Section 4.5 on page 24, CNCS transports SGsAP
only directly over SCTP.
Aside from these exemptions, all other Signaling Application Protocols are
transported on a dedicated adaptation layer.
Suitable for the transport of ISUP, SCCP Users, TUP, BICC, GCP Signaling
Application Protocols over the IP network.
Allows for full MTP3 message handling and network management capabilities
between any two IP-SEPs.
Suitable for the transport of ISDN Q.921 user messages (Q.931) over the
IP network.
All adaptation layers use the services of the underlying reliable common SCTP
protocol defined in IETF RFC 4960, Reference [46], which is a transport protocol
that provides error control, sequence control and flow control functions required
for the reliable transport of signaling messages over IP networks.
4.1 IP Layer
The IP network provides IP connectivity through some IP routers. As shown
in Figure 7, the IP network is typically divided in several sites hosting various
network nodes (in Figure 7: MSC-S, M-MGw, and HLR) and a common IP
transport backbone.
SITE SITE
MSC-S MSC-S
M-MGw Router
M-MGw
Switch Router Router Switch
Router IP
HLR HLR
The basic site infrastructure includes site routers and LAN switches. This site
equipment provides the connectivity, aggregation, and traffic separation. The
LAN switch provides Ethernet interfaces and the site router provides a wide range
of interface types towards the IP backbone.
The site routers and the LAN switches are expected to be duplicated to achieve
redundant site configuration. It is also advisable to use a pair of Gigabit Ethernet
(GE) or 10GE links between LAN switches to achieve a link redundant solution.
Optionally, the site infrastructure can include Firewalls and Security Gateways
to protect the network nodes from malicious attacks and to provide encryption
for enhanced security.
An overview of the solution for IP and Ethernet connectivity and basic resilience
for the transport of signaling over SCTP/IP is given in the User Guide Managing IP
Transport in the Core Network.
SCTP provides reliable transfer of user messages between peer SCTP Users.
It is designed to allow the transport of signaling traffic over IP networks and
operates on top of the IP layer. It performs this service based on multiple Streams
within an SCTP Association, established between two SCTP EPs. SCTP offers a
sequenced delivery of user messages within multiple Streams with an option to
bypass the sequenced delivery for individual messages. The advantage of using
multiple Streams within an SCTP association is that while one Stream can be
blocked waiting for the next in-sequence user message, the delivery from other
Streams can proceed.
IP-1 IP-3
SCTP MH EP SCTP MH EP
IP-2 SCTP Association IP-4
MSC-S M-MGw
The details about how IP Path management is implemented in CNCS nodes are
further detailed in Section 6.1.1.1 on page 53.
The scenario depicted in Figure 8 is the more common scenario, but in some
networks it is possible that only connectivity of the direct IP Paths (IP1⇔IP3
and IP2⇔IP4) is available as shown in Figure 9. This is referred to as Reduced
Connectivity scenario.
IP-1 IP-3
SCTP MH EP SCTP MH EP
IP-2 SCTP Association IP-4
MSC-S M-MGw
One of the basic functions provided by SCTP is congestion control. Under adverse
network conditions, congestion situations arise which can lead to partial or, in the
worst case, full network outages. SCTP provides mechanisms to adapt to such
conditions avoiding message loss and spreading congestion into the network. The
basic idea behind SCTP congestion control is to regulate the SCTP transmission
rate. SCTP congestion control is applied to the SCTP association and not to
individual Streams.
The User Guide Managing and Optimizing SS7 Congestion Control and SIGTRAN
Performance provides the reader with an overview of how SCTP congestion
control is implemented in CNCS nodes and guidance for an accurate configuration
of the SCTP Congestion Control functions.
Another aspect considered in this document is the Stream setting. The number
of Streams to be supported by an SCTP association is specified by the SCTP user
at SCTP association establishment, or by values of MIS and MOS parameters,
depending on the SCTP O&M model used. Both SCTP EPs include in the INIT and
INIT ACK chunks the number of Streams the sender wants to create (number
Figure 10 shows the different Signaling Application Protocols that are possible
users of the M3UA layer.
(*)
M3UA
SCTP
LL
Figure 11 depicts a scenario where the IP-SEP, in this case an MSC-S, is connected
through two M-MGws to several SS7-SEPs residing in a TDM- or ATM-based
network, in this case HLRs residing in a TDM-based network. Figure 11 shows
the use of M-MGws but the traffic scenario is valid also if instead of M-MGw,
STP/SGWs from other vendors are used.
M-MGw HLR
IP TDM
MSC-S
M-MGw HLR
MAP MAP
TCAP TCAP
NIF
SCCP SCCP
M3UA M3UA MTP3 MTP3
SCTP SCTP MTP2 MTP2
LL_ip LL_ip LL_tdm LL_tdm
Figure 11 Multiple Gateway Scenario
The two M-MGws act as SGWs for the conversion between TDM and IP transport
technologies. The M-MGws are transparent to the Signaling Application Protocol.
The M-MGw pair can also be replaced by an SGW pair of any other vendor. This
scenario provides node level redundancy avoiding single point of failure.
The scenario applies to the case where the SS7-SEPs support Quasi-Associated
Signaling Mode; the SS7-SEPs are provisioned with the M-MGw and MSC-S SPCs.
From the SS7-SEPs viewpoint, the two M-MGws are seen as STPs.
Two signaling links between the HLRs and the pair of M-MGws are created
following the physical links. The signaling interfaces are used in load-sharing
mode. On the IP network side, one or two SCTP Associations are defined between
the MSC-S and each M-MGw.
If M-MGws are used as illustrated in Figure 11, only direct routing must be used.
Even if there is a link between the M-MGws, no routing between them must
be used because the M-MGw does not support M3UA Relay, although it does
not prevent such configuration. In other words, even if an SCTP association is
available between the M-MGws no routing of the previously mentioned protocols
must be used over that SCTP association, otherwise it can result in circular routing
of application messages between the two M-MGws and traffic loss.
Figure 12 illustrates the scenario where the IP-SEP, in this case an MSC-S, is
connected through one M-MGw to one SS7-SEP residing in a TDM- or ATM-based
network, in this case a POI residing in a TDM-based network.
POI
TDM
MSC-S M-MGw
NIF
ISUP
M3UA M3UA MTP3 ISUP
SCTP SCTP MTP3
LL_tdm
LL_ip LL_ip LL_tdm
Figure 12 Single Gateway Scenario
The M-MGw acts as SGW to convert between TDM or ATM and IP transport
technologies. The M-MGw can also be replaced by an SGW of any other vendor.
The M-MGw is transparent to the Signaling Application Protocols.
This scenario applies to the case where the SS7-SEP supports ‘‘Associated
Signaling Mode’’ and it requires the activation of ‘‘Associated Signaling Mode in
SGW’’ feature in M-MGw. The feature is supported for ANSI, ITU, TTC, and CCSA
standards. The SS7-SEP is connected to the MSC-S through the M-MGw so that
the MSC-S is seen as adjacent node in the SS7-SEP node.
This scenario does not provide node-level redundancy and therefore has the
disadvantage that it creates a single point of failure.
It is also possible to connect an SS7 STP node to the M-MGw using the Associated
Signaling Mode feature so that the MSC-S is seen as adjacent node in the SS7 STP
as shown in Figure 13. This scenario is only useful to save additional SPCs of the
M-MGw, for instance in international networks.
Recommendation:
In cases where the SPCs are available, it is recommended to use two M-MGws
that act as STP/SGW.
Acts as SPC=A
towards SPC=C
NIF
ISUP
M3UA M3UA MTP3 ISUP ISUP
SCTP SCTP MTP3 MTP3
LL_tdm
LL_ip LL_ip LL_tdm LL_tdm
Figure 13 Associated Signaling Mode in SGW, SS7 STP
The M-MGw is seen as ordinary SGW (also performing the STP function) from
MSC-S but the SS7-SEP or SS7 STP is configured as if the signaling links were
terminated in the SP of the MSC-S.
Recommendation:
Configure several signaling links in the link set towards the SS7-SEP and to
provide redundancy and load sharing. Distribution of the signaling links from
the SS7-SEP and STP to two redundant M-MGws must not be used because
such configuration results with disturbed MTP3 management procedures.
Recommendation:
IP gsmSCF
MSC-S
CAP CAP
TCAP TCAP
SCCP SCCP
M3UA M3UA
SCTP SCTP
LL_ip LL_ip
Figure 14 Peer-to-Peer over M3UA
Ericsson IP-STPs support M3UA relay function, which is illustrated in Figure 15.
MSC-S IP-STP
IP
MSC-S IP-STP
BICC
M3UA M3UA
SCTP SCTP SCTP
LL_ip LL_ip LL_ip
Figure 15 Single M3UA Relay Stage
For the configuration of M3UA Relay, see User Guide M3UA Relay (SPX specific
document).
Recommendation:
The reason for the previous recommendation is that with M3UA Relay MTP-level
functions such as point code mapping, accounting and policing are not available.
(*)
MTP3/MTP3b
M2PA
SCTP
LL
Recommendation:
M2PA connections are recommended for the signaling between nodes with
‘‘IP-STP’’ function.
Use redundant M2PA links between IP-STPs deployed as mated IP-STP pair.
Use at least two M2PA-based links per link set between the nodes with
‘‘IP-STP’’ function.
The reason for the previous recommendation is that there is a one-to-one relation
between an M2PA link and an SCTP association.
Note: Use of multiple SCTP associations between nodes has no adverse effect
on the nodes or SCTP performances as long as the used number of SCTP
associations can be supported by the hardware resources.
The use of a single M2PA-based link in a link set between the IP-STPs is not
encouraged because it introduces head-of-line blocking by retransmissions on
SCTP level, as M2PA does not use multiple SCTP Streams for traffic messages.
Recommendation:
User Guide Managing and Optimizing SS7 Congestion Control and SIGTRAN
Performance provides dimensioning guidelines for M2PA links.
Figure 17 shows the Signaling Application Protocol that is the user of the IUA
Layer.
Q.931
IUA
SCTP
LL
Figure 17 IUA Protocol Stack
The PRA is physically connected to the M-MGw where the user plane (B-Channels)
is handled. The D-Channel signaling messages are backhauled on the IUA
protocol stack between M-MGw and MSC-S. The B-Channels are controlled from
the MSC-S over the existing GCP interface to the M-MGw.
IP TDM
PABX
MSC-S M-MGw
NIF
Q.931
IUA IUA
SCTP SCTP Q.921 Q.931
LL LL_ip Q.921
The Q.931 messages from the PBX are carried by the LAPD (Q.921) and
backhauled on the IUA/SCTP/IP stack towards the MSC-S by the ‘‘Nodal
Interworking Function’’ (NIF). This means the Q.931 protocol is transparent
for the M-MGw. In the M-MGw, all D-Channels associated with one MSC-S are
backhauled over one SCTP association. IUA and GCP traffic use different SCTP
associations. The association for GCP traffic is not shown in Figure 18.
One PBX can communicate with one MSC-S. One MSC-S can communicate with
several PBXs connected to one or several M-MGws.
Note: The STC (described in ITU-T Q.2150.3, Reference [52]) is not a real
protocol layer as it does not add protocol control information to
the Message Signal Unit (MSU). It is however used from a protocol
perspective to hide the underlying layers from GCP.
GCP
STC
SCTP
LL
Figure 19 GCP over SCTP Protocol Stack
In CNCS, the only communication scenario which involves GCP over SCTP is
between the MSC-S and the M-MGw as illustrated in Figure 20.
IP
MSC-S M-MGw
GCP GCP
STC STC
SCTP SCTP
LL-ip LL-ip
Figure 20 GCP over SCTP
In this scenario, only Multi-Homed SCTP Associations must be used. The reason is
that there is no redundancy supported at GCP level.
Only one MH SCTP association can be defined for one remote virtual M-MGw.
If more than one virtual M-MGw exists on a physical M-MGw, one MH SCTP
association must be configured for each virtual M-MGw.
SGsAP
SCTP
LL
Figure 21 SGsAP over SCTP Protocol Stack
In CNCS, the only communication scenario which involves SGsAP over SCTP is
between the MSC-S and the MME as illustrated in Figure 22.
MME
IP
MSC-S
SGsAP SGsAP
SCTP SCTP
LL-ip LL-ip
Figure 22 SGsAP over SCTP
— SCCP level
— SGsAP level
When the Primary Path becomes inactive, the Failover to a Secondary Path
occurs. If Switchback Mode is set to ‘‘Immediate’’ or ‘‘Exponential Switchback’’,
the Primary Path is used for data transmission whenever it is available or after
some successful heartbeats.
However, if switchback mode is set to ‘‘No Switchback’’, the Primary Path is used
for data transmission only at establishment of the association and it remains in
use until it fails. At that point, data transmission is permanently migrated to a
Secondary Path.
The IP Failover guarantees zero packet loss for failures in the IP network.
Note: However, ‘‘zero packet loss’’ is only guaranteed for IP backbone or LAN
infrastructure failures but not for failures of interface boards, in which
case all messages in the send buffer are lost.
Ericsson nodes are MH and details about IP Path Management can be found in
Section 6.1.1 on page 53 and in Section 6.1.2 on page 55.
Recommendation:
The architecture of the remote node might impose use of multiple MH SCTP
Associations for M3UA connectivity. For instance, MSC-S BC requires two MH
SCTP Associations.
GCP over SCTP and IUA signaling can use only one SCTP Association from
MSC-S to an M-MGw. See User Guides MSC-S Configuration Management, and
Managing and Optimizing SS7 Congestion Control and SIGTRAN Performance
for more details.
There are situations where CNCS nodes communicate with nodes that support
Single-homed SCTP EPs only. Node that supports Single-homed SCTP EPs only
can communicate with nodes that support Multi-homed SCTP EPs, such as CNCS
nodes, that is, it can use the different IP Paths provided by the nodes that support
Multi-homed SCTP EPs according to the previous description.
Recommendation:
In this case, CNCS nodes must use MH SCTP EPs for associations to such nodes
according to the previous recommendation.
There are situations where CNCS nodes communicate with nodes of other vendors
that do not provide sufficient redundancy on transport level alone and rely on
redundancy mechanisms of layers above SCTP. For instance, such nodes can rely
on multiple M3UA signaling processes in their transport redundancy solutions.
Recommendation:
In this case, an MH SCTP association is used from CNCS nodes to each M3UA
process at the remote node.
Each M2PA link uses its dedicated SCTP Association, which can be either SH or MH.
Recommendation:
More details on M3UA level redundancy are provided by User Guide Message
Transfer Part 3 User Adaptation Layer and User Guide for Signalling, Reference
[42].
Recommendation:
Use load-sharing redundancy mode for signaling sent to more than one SGW or
IP-STP.
Recommendation:
Recommendation:
Redundancy on M3UA level can be also needed when CNCS nodes use MH SCTP
associations depending on the node architecture (see Section 4.6.2 on page 25).
Global Title Routing Cases (GTRCs) must be configured in dominant mode (no load
sharing defined) for traffic destined to a pair of redundant SCCP nodes working in
active-standby mode, so that Primary Signaling Point (PSP) designates the active
node while Secondary Signaling Point (SSP) designates the standby node.
For the redundant peer nodes working in load sharing mode, see Section 4.6.8.1
on page 29.
For further details, see Adaptation Direction Traffic Data: Handling of Global
Title Routing Cases.
For further details, see User Guide SGs-Interface Configuration and Enabling of
SMS over SGs-Interface and LTE to CS Fallback.
For more information see User Guide Message Transfer Part 3 User Adaptation
Layer and User Guide for Signalling, Reference [42].
The benefits of load sharing are more equal use of the nodes and an in-service
performance impact reduced to a portion of the traffic if there is a failure. Load
sharing can be configured on different levels in the signaling network.
The principles for load sharing depend on the protocol used, but the load sharing
mechanisms in all layers are based on the principle of considering SS7 parameters
such the Signaling Link Selection (SLS) field for selecting the routes and links
(for MTP), SCTP associations (for M3UA), or Destination Point Codes (DPCs) (for
SCCP).
Recommendation:
Note: The feature is supported for ITU, CCSA, and TTC standards.
— Basic
— Extended (for CL traffic only and available through SCCP GTRE feature)
Recommendation:
Use SCCP load sharing instead of load sharing on MTP3/M3UA level in cases
when multiple nodes in the network share the SPC value.
Example:
HLR-FEs can be configured to share a same SPC value to load share TCAP
dialogs among HLR-FEs on M3UA/MTP3 level. Avoid such configurations.
For further details, see Adaptation Direction Traffic Data: Handling of Global
Title Routing Cases.
The SLS field is provided by the users of MTP3 or M3UA similarly to SCCP to
guarantee in-sequence delivery. It is used by the M3UA and MTP3 layers as a
basis for the load sharing scheme.
In the MTP3 ITU-T and China standards, load sharing can be performed between
up to four routes towards a destination.
In ANSI, the MSC-S can perform load sharing between up to two routes, while the
M-MGw is able to load share over a maximum of four routes.
The M3UA load sharing scheme is node-dependent. MSC-S offer three load
sharing schemes which differ in the maximum number of SCTP associations to
load share over and the algorithm used for traffic distribution over these SCTP
associations. With the predefined pattern scheme, it is possible to load share over
a maximum of 16 SCTP associations, while for the other two schemes it is possible
to load share over a maximum of 64 SCTP associations.
For more information about the M3UA load-sharing schemes, see User Guide
Message Transfer Part 3 User Adaptation Layer.
For other MTP3 standards, see Function Specification CCITT7 Signalling System
No. 7, Message Transfer Part.
For more information about load-sharing schemes in the M-MGw see User Guide
for Signalling, Reference [42].
Recommendation:
To achieve equal use of signaling routes in a node, the number of routes of the
same priority to a destination is recommended to be a power of 2.
In MSC-S, at load sharing over two routes in MTP3 networks (ITU-T, China and
TTC standards), it is possible to control the SLS bit used for load sharing in a
signaling route set.
The selection of the SLS bit that is used as a basis for load sharing is important by
an intermediate STP node scenario as the one depicted in Figure 23, because it
can cause uneven load distribution in non-adjacent STPs when on a transfer path
through the network (route) several nodes (for instance SEPs and their adjacent
STPs/SGWs) use the same SLS bit for load sharing.
STP/SGW STP/SGW
A C
MSC-S MSC-S
SEP SEP
STP/SGW STP/SGW
B D
Figure 23 Intermediate STP Node Scenario
Recommendation:
At load sharing on two MTP3 routes, it is recommended to set the SLS bit
differently in the following cases:
— a: On link sets at SS7-SEPs
For example, the load sharing bit could be configured to value 3 in SS7-SEPs
(previously item a), and 2 at the first STP/SGWs nodes on the route to a
destination (previously item b).
Recommendation:
Recommendation:
At load sharing on more than two MTP3 routes, for instance when a SS7-SEP is
connected to four STPs and all routes through the STPs have the same priority,
the MTP3 load sharing has to be set to use the extended load sharing option.
In MSC-S, at load sharing over up to four routes in M3UA networks, for each
MTP3-User protocol it is possible to configure an SLS bit pattern for load sharing
of the protocol messages by a node.
Recommendation:
See User Guide Message Transfer Part 3 User Adaptation Layer for
recommendations on use of predefined patterns at M3UA load-sharing in MSC-S.
Note: At each STP/SGW, the M3UA and MTP3 routes to a SEP defined on a link
between mated STPs (STP A - STP B and STP C - STP D in Figure 23)
must be assigned lower priority than routes from the STP/SGW to the
SEP. This secures that links between mated STPs are used for alternative
routing only at failure of direct routes of higher priority to SEPs.
Load sharing on SGsAP level is achieved by defining two SCTP associations that
are bound to the same remote host object (representing the remote MME node).
Load distribution and association selection based on availability are automatically
applied by an internal function on SCTP User level per remote host.
On CNCS side, an adaptive load sharing mechanism is applied, that is, if one SCTP
association is congested then the load sharing method shifts more traffic to the
non-congested association.
For further details, see User Guide SGs-Interface Configuration and Enabling of
SMS over SGs-Interface and LTE to CS Fallback.
For more information on the specifics of the architecture and configuration, see
the Function Specification Cluster Overview and User Guide MSC-S Configuration
Management, respectively.
— SUA Gateway for SCCP (MAP, CAP, RANAP, BSSAP, BSSAP+ and INAP) traffic
— M3UA Relay for trunk signaling traffic (ISUP and BICC) or GCP over M3UA
Remote M3UA peer must have SCTP Associations configured to both SPXs for
redundancy reasons. Exception to this rule is the SCTP association for GCP/M3UA
traffic used for PRA.
M2PA traffic is used when SPXs are configured as STP or SRP between two
external nodes.
IUA traffic is terminated on one of the SPXs only and remote IUA peer must have
SCTP association configured to one SPX.
GCP traffic that is used for PRA and is defined between M-MGw and the SPX, can
be defined over SCTP/IP or M3UA/SCTP/IP stack.
The MSC-S can be configured as one of the following two configurations or both
of them concurrently:
In case of multiple SPC configuration, MSC-S uses a dedicated own SPC for GCP
over M3UA and trunk signaling traffic (ISUP, BICC, and China TUP). MSC-S uses
another dedicated own SPC for SCCP (MAP, CAP, RANAP and BSSAP) signaling
traffic. Figure 24 illustrates the point code allocation in MSC-S for multiple SPC
configuration.
SPX1
M3UA MTP3
MTP3- SCCP- SPC-C SPC-C
Users Users
SPC-B SPC-A
MTP3 Traffic
M3UA Traffic
SPX2
SCCP
M3UA MTP3
SPC-D SPC-D
In case of single SPC configuration, MSC-S can use one own SPC for GCP over
M3UA, SCCP (MAP, CAP, RANAP BSSAP, BSSAP+ and INAP) and trunk (ISUP,
BICC, and China TUP) signaling traffic. Figure 25 illustrates the point code
allocation in MSC-S BC for single SPC configuration.
Note: When PRA is required for a single SPC, two additional SPCs can be needed
for communication between SPX and blades. Additional SPC can be
needed in M-MGw, if GCP over M3UA is used. More information can be
found in User Guide MSC-S Configuration Management.
SPX1
M3UA MTP3
MTP3- SCCP- SPC-A SPC-A
Users Users
SPC-A SPC-A
MTP3 Traffic
M3UA Traffic
SPX2
SCCP
M3UA MTP3
SPC-A SPC-A
MTP3 Traffic
SPC for SCCP, Trunk,
SPC-A
GCP and MTP3 Traffic
M3UA Traffic
Figure 25 SPC Representation in MSC-S BC (Single SPC Configuration)
MSC-S allows the use of multiple own SPCs towards a same DPC for trunk
signaling traffic (ISUP, BICC, and China TUP).
Each SPX must have configured its own SPC, but these SPCs can remain hidden
from the other nodes when M3UA is used for transport of signaling traffic to and
from external nodes.
When MSC-S takes part in several SS7/SIGTRAN networks, then the dedicated
own SPC must be configured for each NI of those networks. For example, if trunk
signaling is used in several networks then a dedicated own SPC must be used for
trunk signaling in each network. Similarly, dedicated own SPCs are needed for
SCCP signaling in multiple networks.
See User Guide MSC-S Configuration Management for further details on use of
point codes in MSC-S and related configuration.
For further information about the signaling capabilities in MSC-S, see Function
Specification IP Based Functions on CP, Protocol Stack.
Path Diversity in egress direction is achieved by the fact that the IP addresses of
the SPXs are allocated to a separate physical interface.
GCP or SGsAP over SCTP traffic does not transit through the SPXs.
Path Diversity in egress direction is achieved by the fact that the MSC Blades VIP
addresses are associated to a separate physical interface.
All nodes connected to an MSC-S must be connected to each SPX using at least one
SCTP association. That is, at MSC-S, two sets of SCTP associations are configured
for M3UA connectivity – one set of associations at each SPX. For example, when
two MSC-Ss are connected to each other, they use at least four SCTP associations,
where each SPX of an MSC-S is connected to both SPXs of the peer MSC-S.
MSC-S supports ASP, SGP, IPSP, or ESP process types where ASP, SGP, and IPSP
are compliant with IETF RFC 4666, Reference [47], while ESP complies with
3GPP TS 29.202, Reference [44]. The process type is configured on a per SCTP
association basis. All process types can coexist on an MSC-S. Although the process
type has to be configured according to the role of a node in M3UA network –
the MSC-S usually acts as IP-SEP – it must be chosen based on the peer node
capabilities.
There are no restrictions on process type (ASP, SGP, IPSP, or ESP) configured in a
node but it is more natural to follow M3UA IETF RFC 4666, Reference [47], and
configure IPSP process type for IP-SEP to IP-SEP communication, and ASP for
IP-SEP to M3UA SGW communication. See User Guide MSC-S Configuration
Management, for recommendations on process type configuration.
Recommendation:
Use the ESP process when communicating to M-MGw or any other Ericsson
node, see 3GPP TS 29.202, Reference [44].
Note: ESP process type can be used for communication to nodes of other
vendors which follow 3GPP TS 29.202, Reference [44].
Use of ESP process is valid for M3UA relations configured up to R14.0, which
means that ESP process type is already in use, and there are no reasons to modify
configuration from ESP to another process type.
Recommendation:
Recommendation:
Recommendation:
On SCTP associations where the MSC-S acts as M3UA SGW or as SGW with
M3UA Relay towards nodes of other vendors, it is recommended to use the SGP
process type for connectivity from each SPX to IP-SEP nodes of other vendors.
If ESP process type is configured for an SCTP association, only a single NI can
be used.
For further information about the M3UA layer configuration, see User Guides
Message Transfer Part 3 User Adaptation Layer, and M3UA Relay (SPX specific
document), for further information on process type configuration at M3UA Relay.
For further information about the M3UA layer functionality and capabilities, see
the following Function Specifications:
For further information about the M2PA layer configuration, see User Guide
MSC-S Configuration Management.
For further information about the M2PA layer functionality and capabilities, see
the following SPX specific documents:
For information about the functionality and capabilities for the transport of GCP
over SCTP, see Function Specification Signalling Transport Converter over SCTP.
IUA uses dedicated single SCTP Association between remote peer and one and
only one SPX component of the node.
For information about configuration of PRA, see User Guides Managing Primary
Rate Access on Core Network Level and MSC-S Configuration Management.
M-M G w node
GCP IU A
STC
EET-IPG
T -IP G EET-IPG
T -IP G
1G E thernet 1G E thernet
Figure 26 shows the allocation of two stacks in an M-MGw and GCP over SCTP
and IUA protocols. Signaling over SCTP/IP is supported over FE, GE, and 10GE
interfaces. Different configurations can be applied for the physical connectivity,
see User Guide for Signalling, Reference [42], for further information.
The node is MH where up to two IP addresses can be configured. The IP, SCTP,
and M3UA layers are protected from board failures by having active – stand-by
General Purpose Board (GPB) configuration.
Each stack hosts own MTP3b SPs. Each MTP3b SP, and thus each stack, needs to
manage its own resources (such as signaling link sets, M3UA SCTP Associations)
and these cannot be shared between the SPs.
For further information about the M-MGw signaling architecture, see User Guide
for Signalling, Reference [42].
Two M3UA standard variants are supported in M-MGw. The protocol variant that
is compliant with 3GPP TS 29.202, Reference [44], is described in this document
as it is the default and recommended configuration for M-MGw. M-MGw is also
compliant with IETF RFC 4666, Reference [47], where the M3UA process types
ASP, SGP, and IPSP are supported. Further information about this variant can be
found in User Guide for Signalling, Reference [42].
Recommendation:
M-M G w n o de
S tack 1 S tack 2
SP SP SP
3-111 2-111 2-444
SR SR SR SR SR SR
M 3U A M 3U A M 3U A M 3U A M 3U A M 3U A
SCTP SCTP
IP host IP host
IP 1 IP 2 IP 3 IP 4
ET-IPG
E T -IP G ET-IPG
E T -IP G
1G E thernet 1G E thernet
For further information about the M3UA layer configuration, functionality and
capabilities, see User Guide for Signalling, Reference [42].
The site structure provides redundant connections for each pair of communicating
nodes within and between sites and enables Path Diversity when the nodes reside
in different sites.
S ite 1 S ite 2
A A
B AB AB B M S C -S
M S C -S
1 .. n 1 .. n
Switch A
R o u te r R o u te r Switch A
A A A
A
B IP B
A AB AB A
M-MGw M-MGw
B B
1 .. n 1 .. n
Switch B R o u te r R o u te r Switch B
A B B A
H LR B B H LR
1 .. n 1 .. n
Each site hosts the nodes that communicate using Signaling over SCTP/IP. Figure
28 depicts MSC-Ss, HLRs, and PNFs M-MGws. The nodes are connected to the site
L2 infrastructure which consists of a pair of L2 (Ethernet) switches denoted as
Switch A and Switch B in Figure 28. The L2 switches are connected to the pair of
the Site Routers (Router A and Router B). The Site Routers enable communication
between the sites through the IP backbone.
The nodes are connected to the site L2 infrastructure with two groups of physical
interfaces (referred to as Interface Group A and Interface Group B).
Within the site, and from the perspective of each site pair in the network, the IP
signaling transport infrastructure is logically divided into two parts:
Node connectivity on Ethernet and IP levels is not discussed in this document. For
further information, see User Guides: Managing IP Transport in the Core Network
and MSC-S Configuration Management.
a Two IP subnets over one (or two) L2 VLAN. One Ethernet interface of the
node interface board or router pair belonging to one IP subnet.
b One IP subnet over one L2 VLAN. All node interfaces and router interfaces
are in one subnet.
The selection of one of the previous options must be combined with the selection
of the redundancy option described in Section 5.1.2 on page 45 and also by
considering the intra-site traffic.
Since the IP signaling transport infrastructure is logically divided into two parts,
it is assumed that the IP backbone infrastructure can be associated to Network
Part A or Network Part B (or providing equivalent redundancy level). The required
two direct IP Paths between each two communicating nodes can be then easily
identified: one is within the Network Part A and the second within the Network
Part B.
If two Single-Homed SCTP Associations are configured between the nodes (that
is, SCTP-application level redundancy is used), one SCTP Association is configured
within Network Part A and the second SCTP association is configured within
Network Part B, and they should use different IP Paths.
Recommendation:
These two approaches are described thoroughly in Section 5.1.2.1 on page 46 and
Section 5.1.2.2 on page 48 but the fundamental difference is highlighted here.
The first approach leads to a rather strict separation between Network Part A
and Network Part B. Although the advantages of introducing the concept of
Network Parts in network design have been discussed to enable Path Diversity, a
strict separation can lead to loss of connectivity between peer nodes in certain
conditions which are described in Section 5.1.2.1 on page 46.
Recommendation:
The requirement for the existence of separate physical paths does not exclude
connectivity between these different paths. On the contrary, the SCTP
implementation works best when there is full connectivity between all involved
IP addresses.
Figure 29 shows as an example the traffic flow between an MSC-S and an M-MGw
in the same site and between M-MGws in different sites. Full and dashed lines
indicate two alternative paths. Similar traffic flows would apply also towards
other nodes that are connected to MSC-S or M-MGw by SCTP.
S ite S ite
A A
M S C -S B B M S C -S
1 .. n 1 .. n
R outer R outer
A Switch A A IP A Switch A A
1
M-MGw B B M-MGw
1 .. n 1 .. n
R outer R outer Switch B
A Switch B B B A
H LR B B H LR
1 .. n 1 .. n
The SCTP layer (if there is one MH SCTP association between the nodes) or
the SCTP-application layer (if there are two Single-Homed SCTP Associations
between the nodes) handles the failure case without assistance from the IP layer
(no IP redundancy applied in the client nodes).
With SCTP-application level redundancy, and during normal operation, both SCTP
associations are used in load sharing mode, this is not the case when SCTP level
redundancy is used where within the MH SCTP association only one path is active
(primary) and the others are standby (alternative).
This risk can become important in large networks, where control of equipment on
different sites is not always coordinated and there is a relatively high chance
that equipment is taken out of operation simultaneously, for example because
of maintenance.
As it can be seen from the previous example, communication loss occurs when
two specific routers are taken out of operation at the same time. For networks
where such an event can be ruled out (that is, networks with central control of
IP infrastructure network-wide) and for networks where conformance with IP
configuration common practice is of high value, the solution described in this
section can be used.
Recommendation:
Traffic from Interface A preferentially sent Router A preferentially attracts and delivers
to other Interface A or Router A, but it can traffic for Interface Group A, but it can also
reach also Interface B or Router B deliver traffic for Interface Group B
S ite S ite
A A
M S C -S B B M S C -S
1 .. n 1 .. n
R outer R outer
A Switch A A IP A Switch A A
2 2
1
M-MGw B B M-MGw
1 .. n 1 .. n
R outer R outer Switch B
A Switch B B B A
H LR B B H LR
1 .. n 1 .. n
The bridge between Network Part A and Network Part B can be realized on L2,
basically placing all client node interfaces on the same IP subnetwork, so that they
can directly communicate over the L2 Switches ((2) in Figure 30) and configuring
appropriate routing entries in the routers. In this case, more attention has to be
paid to ensure Path Diversity.
Path Diversity can be achieved by configuring the IP layer in a way that client
nodes interfaces are placed in the same IP subnetwork thus allowing direct
communication between Network Part A and Network Part B within the same site,
but to configure the routers so that they see two separate IP-subnetworks.
When Path Diversity is combined with IP redundancy, the chance that in case
of failure the path between communicating nodes is found is higher. Moreover,
this network design is capable of handling a broader spectrum of failure cases.
This can be illustrated with the same fault scenario example as discussed in
the previous section (simultaneous failure of both the Router A in Site 1 and
the Router B in Site 2): IP connectivity between the two remaining routers is
used to run SCTP traffic (for example Router B on the left side takes over the
incoming/outgoing traffic that was handled by Router A before the failure).
In addition, the MSC-S can use IP Address Migration since both Interface A and
Interface B of the MSC-S are in the same IP subnetwork. This increases resilience
if there is a board or cable failure in the node as complement to other equipment
in the site failure.
The entire IP backbone has to be divided into several ‘‘logical’’ networks in such a
way that the logical networks can be designed independently. This means that
on top of the shared IP infrastructure, additional measures must be taken so that
different logical networks interfere with each other as little as possible.
2. OAM Network
3. Data Network
Typical techniques to achieve this network separation are MPLS and BGP/MPLS
IP Virtual Private Networks (VPNs).
Note: The typical mechanisms to provide Path Diversity are MPLS and routing
protocols (metrics).
— For IUA: User Guide for Primary Rate Access, Reference [41]
— For GCP: User Guide for Virtual Media Gateway Handling, Reference [43]
6 How to Use
This section provides several different parameter value sets, encompassing SH,
and MH scenarios. The basic assumption is that transport requirements are
fulfilled. If they are not fulfilled, the set of parameters must be adapted on a
case-by-case basis to the network conditions.
All listed parameters are specified in IETF RFC 4960, Reference [46] unless
differently indicated.
For additional details about other SCTP parameter settings, see User Guide
Managing the Stream Control Transmission Protocol
— IP Path mode
Recommendation:
— No Switchback
— Immediate Switchback
— Exponential Switchback
Recommendation:
The response time of IP Path Failover detection can be improved by enabling the
‘‘Potentially Failed’’ function. Potentially Failed function enables failover to take
place at an earlier phase before PMR is exceeded.
Recommendation:
See User Guide Managing and Optimizing SS7 Congestion Control and SIGTRAN
Performance for more details.
See User Guide Managing and Optimizing SS7 Congestion Control and SIGTRAN
Performance for more details.
As explained further in Table 6, these two parameter values control the number of
retransmission attempts that are done after transmission time-outs.
In the SH case, there is only one SCTP Path to be attempted, thus Assoc.Max.Rtx
equals to Path.Max.Rtx.
In the MH case, more IP Paths can be attempted and the strategy is to give each
IP Path equal chance, thus Assoc.Max.Rtx equals to Path.Max.Rtx * no. of
SCTP Paths (SCTP Path Mode) or equals to Path.Max.Rtx * no. of IP Paths (IP
Path Mode).
In case two Single-Homed SCTP Associations are defined between two nodes,
redundancy is handled on M3UA level with two routes towards the same
destination, each link tied to an SCTP association residing on one interface.
For multi-homing, the network scenario is the one where an MH SCTP association
is defined between two nodes and redundancy is handled at SCTP level.
Recommendation:
The following ranges of SCTP and M3UA parameter values are recommended
to achieve failover times below three seconds and to ensure correct protocol
behavior.
Note: Individual nodes have default parameter values widely covering general
cases. Therefore, for certain parameters different default values to
those listed in Table 6 can be found in node documentation. If there is
a mismatch with the parameter values indicated in this User Guide, the
values stated here apply.
Applying the parameter previous values results in the following fault detection
time (assuming that the RTO in the exact moment of the failure is equal to
RTO.Min):
Note: The factor 2 is actually the back-off factor indicated in the IETF RFC 4960,
Reference [46]. For the CP-based SCTP, this value is calculated from
configuration parameter RTOBF (depending on the SCTP O&M model
used, default value of parameter RTOBF, gives back-off factor lower than 2
(1.57) and this lowers the overall computation for such implementation).
The RTO value is kept for each destination and, assuming that the RTO in the
exact moment of the failure is equal to the RTO.Min, if there is a complete failure
of the remote end, the following applies:
The first retransmission happens after RTO, the second happens after RTO *2
and up to RTO.Max as specified in IETF RFC 4960, Reference [46]. In the worst
case of one single packet to be sent, all the timers run sequentially and the fault
detection time would be > 7 seconds.
However, in the normal case different packets can be sent to a different SCTP
path. This causes the retransmission timer to be started and run in parallel to
another path retransmission timer. This brings the detection time below 6 seconds
and typically in the range 2.5–2.8 seconds.
Note: The factor 2 is actually the back-off factor indicated in the IETF RFC 4960,
Reference [46]. For the CP-based SCTP, the value of back-off factor is
calculated from configuration parameter RTOBF (depending on the SCTP
O&M model used, default value of parameter RTOBF, gives back-off factor
lower than 2 (1.57) and this lowers the overall computation for such
implementation).
Recommendation:
Note: The advertised receiver window and the RTT control the throughput
of an SCTP association. Any sender cannot have more than A_RWND
outstanding data (that is, data not yet acknowledged) towards a given
destination. The acknowledgment comes within RTT. Therefore a peer
can send A_RWND data per RTT interval according to the formula:
Throughput = A_RWND / RTT
RTT is not simply 2*one_way_delay but also includes internal delays such
as scheduling delays. This User Guide simplifies the formula, however,
real calculation take all the contributions into consideration.
RTT = 10 ms
A_RWND = 8 Kbytes
Throughput = 8 Kbytes/10 ms = 800 kbyte/s = 6.4 Mbit/s
RTT = 10 ms
A_RWND = 32 Kbytes
Throughput = 32 Kbytes/10 ms = 3200 kbyte/s = 25.6 Mbit/s
RTT = 100 ms
A_RWND = 64 Kbytes
Throughput = 64 Kbytes/100 ms = 640 kbyte/s = 5.1 Mbit/s
Example 1
Example 2
But also:
RTT = 10 ms
A_RWND = 64 Kbytes
Throughput = 64 Kbytes/10 ms = 6400 kbyte/s = 51 Mbit/s
As seen, the last setting could cause bursts of signaling toward one node, thus the
changing of A_RWND must be considered carefully. Even though 51 Mbit/s could be
considered a high value, it is only the theoretical throughput over the transport
network. In reality, the speed is limited by the consumption speed of the receiver.
The consumption speed is how fast the receiver SCTP layer can deliver data to
its users.
If the calculated throughput is too low, given the current maximum value for
A_RWND, multiple parallel SCTP associations can be used. In this case SCTP rules
must be observed: at least one of the values for the tuple source IP address,
remote IP address, and port number must be changed.
During congestion situations (that is, when A_RWND is zero) the sending node
probes the receiving node to check if the congestion situation has ceased. One
SCTP packet is sent. The receiver sends a SACK: if congestion situation persists,
the SACK does not step the Cumulative TSN Ack Point, otherwise the TSN is
stepped up and communication resumes.
MSC-S and M-MGw send spontaneous SACK when A_RWND changes from 0 to a
higher value. 0 here is any value in the range [0 - MTU size], that is whenever
the amount of free receive buffer space exceeds 1 Maximum Transmission Unit
(MTU), a spontaneous SACK is generated.
In M-MGw, the spontaneous SACK is generated when 50% threshold for free
receiver window has been crossed.
The size of the SCTP packet is referred to as Protocol Data Unit (PDU) in this
document.
The following relationship expressed in octets (Bytes) must be kept when IPv4
is used:
Protocol data unit size + IP header size <= IP MTU size ⇒
- IPSec header size
Where:
— The IPsec header size can vary but a typical value is 64 octets (IPsec with
AH/ESP tunnel, Authentication Header (AH) MD5, and ESP 3DES).
If the IPsec header size is 64 octets, IP header size is 20 octets and the IP MTU
size is 1500, the PDU size must be set as follows:
Protocol data unit <= 1500 - 64 - 20 = 1416 octets.
Table 8 shows the parameter names used in nodes for setting the previously
explained PDU size.
SCTP can specify at SCTP association startup time the number of Streams to be
supported by the SCTP association. This number is negotiated with the remote
end.
Depending on the SCTP O&M model used, maximum number of Streams allowed
by the SCTP User can be set for selected SCTP EP and SCTP associations without
impacting the signaling traffic over other SCTP users.
Recommendation:
In some nodes (depending on the SCTP O&M model used) the Stream setting
is global for all associations, therefore if more protocols have to be deployed
(M3UA, GCP, and so on, except the M2PA) the highest needed Stream number
must be used.
Proper planning is needed because if the addition of a new protocol is done when
the node is already in operation, depending on the SCTP O&M model used, the
existing connections must be undefined before changing the Stream number.
Recommendation:
If the Stream setting is global for all associations, MIS and OS or MOS, depending
on the SCTP O&M model used, must be set to the maximum allowed value in
MSC-S and M-MGw. The SCTP protocol then negotiates the number of Streams.
Recommendation:
Note: SCTP bundling denotes the encapsulation of more than one SCTP MSU
into one SCTP packet. SCTP bundling can reduce the number of packets
thus can increase throughput and reduce the load (on the nodes that
concentrate traffic similarly to MSC-S). Therefore, it is recommended
that peer nodes also use SCTP bundling on SCTP associations towards
the MSC-S.
The timer is started at entering into zero window condition and stopped at
termination of zero window condition.
SCTP ungracefully closes the association (ABORT is sent with cause code ‘‘User
Initiated Abort’’) at the expiration of this timer.
Recommendation:
Set the Zero Receiver Window Supervision timer to the following value:
4 * RTO.Max * Assoc.Max.Rtx.
When the IP network does not fulfill such transport characteristics, SCTP protocol
can be adapted by tuning certain parameters. This adaptation has consequences,
for example, for performance and fault detection.
When longer delays (and also likely larger delay variations) occur then a trade-off
between sensitivity to detect packet loss and the ability for fast detection of SCTP
association loss must be considered. Increasing the round-trip timer thresholds
leads to less sensitivity against long delays but also increases failure detection
time and delay for application messages when retransmissions occur.
By just considering longer delays, sensitivity is ranked higher than long application
delays and failure detection times, based on the assumption that the IP network
has only minor packet loss even with longer delays. For high packet loss, see
Section 6.1.8.2 on page 69.
Recommendation:
RTO.Min > 2*average_one_way_delay
(and in any case not lower then 100ms).
Note: Even for the case where no delay exists (for example LAN), the lower
limit is still set to 100 ms. Below that value, possible delay variation can
cause too many spurious time-outs.
Note: Even for the case where no delay exists (for example LAN), the upper
limit is still set to 400 ms. Above that value, possible delay variation can
cause too many spurious time-outs.
Example 3
Example 4
The values of RTO, RTO.Min, and RTO.Max strictly govern the fault detection time
of an SCTP association, with the values of AMR and PMR. In the two previous
examples, and generally speaking in all cases where a longer RTT is needed, the
fault detection time is longer accordingly. In other words, in case of failure it takes
longer for SCTP to realize that the SCTP association has to be turned down.
Although the operator should know the expected minimum and maximum packet
delays from network design, it is recommended to fine-tune RTO.Init, RTO.Min,
and RTO.Max parameters based on delay measurements.
— The value of Smoothed SCTP Round-Trip Time (SRTT) can be obtained from
the result printout of command IHSSP.
For a given A_RWND value, the throughput decreases when the RTT increases.
Therefore, it is important to adapt A_RWND to the increased delay.
Some examples:
RTT = 400 ms
A_RWND = 8 Kbytes
Throughput = 8 Kbytes/400 ms = 20 kbyte/s = 160 kbit/s
Example 5
RTT = 600 ms
A_RWND = 8 Kbytes
Throughput = 8 Kbytes/600 ms = 13,3 kbyte/s = 106 kbit/s
Example 6
Example 7
Example 8
Here, node-specific limitations can apply (for example the A_RWND limit in MSC-S is
64K) thus a lower throughput is achieved. Node documentation must be consulted
for the possible values.
Throughput limitation can cause buffer exhaustion of the SCTP send buffer and
congestion indication sent from SCTP to SCTP user as the individual node can still
try to send at higher speed than SCTP can reach.
Note: Thus careful dimensioning must be done and in the extreme case more
SCTP associations must be defined.
Moderate increase (for example 25%) of BUF (and THR accordingly) can be done
to compensate the lower throughput after a packet loss (slow start procedure in
SCTP). Still, congestion cannot be avoided in all cases, but that is a consequence
of the packet drop.
Note: The amount of available memory is specific to the node; therefore any
change in this area must be done by considering the available resources
because of the potential impact on the number of sustainable SCTP
associations.
SCTP handles packet retransmission, thus it is able to deal with packet loss.
Note: There are two types of packet loss. One is random single packet loss
caused by Random Early Deletion algorithms frequently used by routers
to prevent congestion. The second one can be characterized by bursts
of packet loss which, for example, can be caused by (frequently) failing
equipment.
SCTP detects packet loss by expiration of RTO for any packet that is retransmitted.
A high packet loss ratio has impacts on SCTP association throughput. In these
conditions, the slow-start procedure of SCTP reduces the performance as it
considers the IP network congested.
Stability analysis of SCTP association needs to consider how long the error bursts
can be. Engineer RTO timers so that the error bursts are detected fairly exactly,
specifically detecting the end of the error burst well.
Since RTO.Min and RTO.Max are limiting the time period between two probes to
detect that the error burst has ended again, it is suggested to decrease RTO.Max
and RTO.Min to a lower value. Reducing SACK_delay also reduces the RTT and
thus improves throughput.
However, this must be balanced by an increasing of AMR and PMR, otherwise the
SCTP association can go down because of spurious time-out.
A combination of high packet loss ratio and long delay makes things worse
because the slow start period is longer (for example it takes longer for the SCTP
association to resume its maximum throughput).
If a combination of packet loss and long delay exists, RTO.Max must be set to the
maximum given by the formula for long delay, and RTO.Min set to a low value
(for example 50 ms). In this case, RTO is governed by the actual calculation of
RTT made by SCTP without imposing any strict MIN limit. This can allow early
detection of packet loss by SCTP if the average delay improves in the IP network.
For the long delay case, throughput limitation affects buffering so the same
considerations apply.
SCTP over satellite connection can be considered as subset of the ‘‘long delay’’ use
case described in Section 6.1.8.1 on page 66, where one-way delays are typically
in the range of >300 ms (RTT in the range of 1 second).
Note: Even for the case where no delay exists (for example LAN), the lower
limit is still set to 100 ms. Below that value, possible delay variation can
cause too many spurious time-outs.
Note: Even for the case where no delay exists (for example LAN), the upper
limit is still set to 400 ms. Above that value, possible delay variation can
cause too many spurious time-outs.
Example 9
As explained earlier, long delay reduces the throughput to a low value. Therefore,
A_RWND must be set to the highest possible value and SACK_delay must be set
to the lowest possible value.
For further information, see and User Guide for Signalling, Reference [42] for
M-MGw.
6.1.8.5 Bundling
Note: The maximum number of MSUs in one SCTP packet is limited by the
maximum size of the SCTP packet (see Section 6.1.4 on page 63).
In Example 10 the effect of LBT on the number of transmitted packets per second
is presented:
Example 10
As illustrated, by configuring the bundling timer to the proper value the number of
packets in the network can be reduced.
In Table 10 the effect of LBT=50ms is shown for different values of MSU rate and
MSU length. Cells where the number of octets received from SCTP User exceed
MTU (see Section 6.1.4 on page 63) are grey filled. In these cases, it is likely that
SCTP transmits fully seized (MTU) packet before LBT expiry.
Recommendation:
HLR
SGW 1
TDM POI
MSC-S
Compact
SGW 2
IP
TDM
SPX2
M-MGw BSC
1
SPX1
MSC-S
BC
ATM
M-MGw
2 RNC
Figure 31 Configuration in M3UA Network
— The M-MGws, SGWs of other vendors and the HLR are connected to the
Compact MSC-S using one Multi-Homed SCTP Association. If the SGWs are
SH, two Single-Homed SCTP Associations are needed between each SGW
and Compact MSC-S .
— The M-MGws, SGWs of other vendors and the HLR are connected to the
MSC-S BC using two MH SCTP associations, each one towards one of the two
SPXs. If the SGWs are SH, two SH SCTP associations are needed between
each SGW and each SPX in MSC-S BC.
— The same MH SCTP association between MSC-S DB and M-MGw and MSC-S
BC and M-MGw can be shared for all signaling application protocols.
IETF RFC 4666, Reference [47], recommends that one endpoint takes the
role of ‘‘client’’ and the other endpoint takes role of ‘‘server’’. Therefore,
the following scenarios are possible:
— SCTP association definition from the MH SCTP EP in each SPX of the MSC-S
to each peer SCTP EPs in HLR, SGWs, and M-MGws
— Own SPC definition. If multiple NIs are used, one own SPC for each NI value
is needed.
— Definition of SPC of the peer nodes (HLR, RNC, and POI) reachable by IP
network only.
— Route definition.
• Two routes in load-sharing mode from each SPX of the MSC-S to the POI
through the two SGWs. The load-sharing scheme is according to the
predefined pattern. On these routes, MSC-S can be configured to behave
as either ESP or ASP depending on M3UA protocol support in the SGWs.
ASP can always be used in this case, but it can complicate configuration.
If ESP is used, then the M3UA behavior must be configured. In this case,
the behavior towards SGWs of other vendors must be configured as
‘‘peer’’.
Recommendation:
• Two routes in load sharing mode from each SPX of the MSC-S to the
RNC through the two M-MGw. Load sharing scheme is according to the
predefined pattern so that each M-MGw receives an equal amount of odd
and even SLS values. MSC-S is configured to behave as ESP.
• One route from each SPX of the MSC-S to BSC connected through one
M-MGw. MSC-S is configured to behave as ESP. The M3UA behavior
must be configured and it has to be ‘‘peer’’.
• One route from each SPX of the MSC-S to HLR. MSC-S is configured to
behave as ESP. The M3UA behavior must be configured. The behavior
can be configured as either ‘‘peer’’ or ‘‘server’’.
• If multiple signaling stacks are used, own SPC(s) is needed for each
signaling stack. Recommendations for distribution of traffic across
multiple stacks are provided in User Guide for Signalling, Reference [42].
• If multiple NIs are used within a signaling stack, one own SPC for each
NI value is needed.
• Signaling Route Sets for Own SPC ‘‘Radio Access Network’’ (SPX1 and
SPX2 of MSC-S).
Note: The routing configuration towards BSC and RNC is out of the
scope of this document. See User Guide for Signalling, Reference
[42], for further information.
• Signaling Route Sets for Own SPC ‘‘Core Network’’ (SPX1 and SPX2 of
MSC-S).
• Signaling Route Sets for Own SPC ‘‘Associated Signaling Mode in SGW’’
(SPX1 and SPX2 of MSC-S).
Note: The routing configuration towards BSC and RNC is out of the
scope of this document. See User Guide for Signalling, Reference
[42], for further information.
— SCTP association definition from the SCTP EP in M-MGw towards peer SCTP
EPs in SPXs in MSC-S
The following SCTP associations are needed for each own SPC:
• Local port number and endpoint instance. This defines the local SCTP EP.
Recommendation:
Recommended as MH.
Recommendation:
— Route definition
• Two routes in load sharing mode to the MSC-S through the two SPXs.
Note: The name BSC is used to highlight the fact that there can be
several BSCs connected to the M-MGw. Different BSCs can be
controlled by MSC-Ss of different architecture, and the mapping
is done accordingly. In Figure 31, these BSCs are called ‘‘BSC’’
for editorial convenience.
Note: This SPC can be used only for traffic to or from BSC. SEP or STP/SGW
signaling traffic to or from other nodes (M-MGw and RNC) must not be
transferred through this SPC.
M2PA STP/SGW
3 A
TP C
M
IP-STP
SS7-SEP
PA
MT
M2
P3
M2PA
UA
M3
M2
PA
IP-SEP
M3 B M2PA STP/SGW
UA D
IP-STP
Recommendation:
Only consider using M3UA if IP-STP C and IP-STP D do not support M2PA.
The link between IP-STP A and IP-STP B is not always needed, as detailed in
Section 6.2.2.1 on page 79 and Section 6.2.2.2 on page 79.
Recommendation:
Use several redundant M2PA links between IP-STP A and IP-STP B when this is
needed.
Only consider M3UA if one IP-STP A or B is mated to the IP-STP of another vendor
which does not support M2PA and no SS7-SEP is present in the access segment.
The presence of SS7-SEP in the access segment excludes the possibility to use
M3UA because an Ericsson IP-STP does not allow switchback from TDM/ATM to
IP type of bearer if the TDM/ATM link goes out of service because of hardware
failure.
Recommendation:
The routes defined through the redundant M2PA links must be assigned lower
priority so that they are not used in load-sharing mode with the M3UA routes.
In case protection against double hardware failure is wanted, the use of redundant
M2PA links between IP-STP A and IP-STP B depends on whether these nodes
support SRP or FNR functions.
In case IP-STP A and IP-STP B do not support SRP or FNR functions, then the
use of redundant M2PA links between IP-STP A and IP-STP B does not bring
any additional benefit because at double hardware failure on a link in the access
segment, MTP3 Network Management messages between the IP-STPs trigger
network rerouting towards the link on the access segment which is still in service.
Recommendation:
The routes defined through the redundant M2PA links must be assigned lower
priority so that they are not used in load-sharing mode with the M3UA routes.
The function ‘‘IP to TDM/ATM failover’’ must be enabled in SPXs if the M2PA link
between IP-STP A and IP-STP B is used.
IETF RFC 4666, Reference [47], recommends that one endpoint takes the
role of ‘‘client’’ and other endpoint takes role of ‘‘server’’. Therefore, the
following scenarios are possible:
The prerequisite for initialization of M2PA links is to configure the following parts
at MTP3 level:
For a detailed description of SCTP and M2PA configuration in IP-STP, see User
Guide Message Transfer Part 2 User Peer-to-Peer Adaptation Layer.
Note: If the redundant M2PA links between IP-STP A and IP-STP B are
used, then the definition of SPC in IP-SEP requires additionally the
definition of IP-SEP as reachable through both M3UA and MTP3
networks with preference for the M3UA network.
• Routing Key. In this case the Routing Key consists of the IP-SEP DPC the
AS is associated to.
The definition of the AS at the IP-SEP automatically defines a route to the IP-SEP
DPC. If this is the first route towards the IP-SEP DPC, the following data is defined:
If this is the first AS created in the node, the following data is defined:
For a detailed description of SCTP and M3UA configuration in IP-STP, see User
Guide Message Transfer Part 3 User Adaptation Layer.
Glossary
3GPP CAP
3rd Generation Partnership Project CAMEL Application Part
AH CCN
Authentication Header Cost Control Node
AMR CCSA
Association Maximum Retransmissions China Communications Standards Association
ANSI CL
American National Standards Institute Connectionless
AoIP CN
A-Interface over IP Core Network
AS CP
Application Server Central Processor
ASP CS
Application Server Process Circuit Switched
ATM CNCS
Asynchronous Transfer Mode Circuit Switched Core Network
AuC DES
Authentication Center Data Encryption Standard
AXE DNS
Public Telephone Exchange System Domain Name System
BGP DPC
Border Gateway Protocol Destination Point Code
BICC DSCP
Bearer Independent Call Control Differentiated Services Code Point
BSC DTA
Base Station Controller Destination Transport Address
BSP DTP
Blade Server Platform Data Transfer Path
BSSAP EGEM
Base Station System Application Part Enhanced Generic Ericsson Magazine
BSSAP+ EP
Base Station System Application Part + End Point
E-SIGTRAN GTT
Ericsson SIGTRAN Global Title Translation
ESP HLR
E-SIGTRAN Signaling Process Home Location Register
ET-IP HLR-FE
Exchange Terminal - IP Home Location Register Front End
ET-IPG IETF
Exchange Terminal - Internet Protocol Internet Engineering Task Force
Gateway
INAP
FE Intelligent Network Application Protocol
Fast Ethernet
IP
FNR Internet Protocol
Flexible Numbering Register
IP-FE
FQDN IP Front End
Fully Qualified Domain Name
IPLB
GCP IP Load Balancer
Gateway Control Protocol
IPsec
GE Internet Protocol Security
Gigabit Ethernet
IP-SEP
GEM Signaling End Point
Generic Ericsson Magazine
IPSP
GESB IP Server Process
Gigabit Ethernet Switching Board
IPv4
GMSC-S Internet Protocol version 4
Gateway MSC Server
IPU
GPB Instruction Processor Unit
General Purpose Processor Board
ISDN
GPRS Integrated Services Digital Network
General Packet Radio Service
ISUP
gsmSCF ISDN User Part
GSM Service Control Function
ITU
GTRC International Telecommunication Union
Global Title Routing Case
ITU-T
GTRE International Telecommunication Union
Global Title Routing Extension Telecommunication Standardization Sector
IUA MPT
ISDN Q.921 User Adaptation Layer Ministry of Posts and Telecommunications of
the People's Republic of China
KPI
Key Performance Indicator MSC
Mobile Switching Center
L2
Layer 2 MSC-S
MSC Server
LL
Link Layer MSC-S BC
MSC Server Blade Cluster
LAN
Local Area Network MSU
Message Signal Unit
LAPD
Link Access Procedures - channel D MTP
Message Transfer Part
LBT
Local Bundling Timer MTP2
Message Transfer Part Level 2
LCS
Location Services MTP3
Message Transfer Part Level 3
LSP
Label-Switched Path MTP3b
Message Transfer Part Level 3 with Broadband
LTE Enhancements
Long Term Evolution
MTU
M2PA Maximum Transmission Unit
MTP2 User Peer-to-Peer Adaptation Layer
NI
M3UA Network Indicator
MTP3 User Adaptation Layer
NIF
MAP Nodal Interworking Function
Mobile Application Part
O&M
MH Operation and Maintenance
Multi-Homed
PBX
MME Private Branch Exchange
Mobility Management Entity
PDU
M-MGw Protocol Data Unit
Ericsson Media Gateway for Mobile Networks
PFMR
MPLS Potentially Failed Maximum Retransmissions
Multiprotocol Label Switching
PLMN
Public Land Mobile Network
PMR SGW
Path Maximum Retransmissions Signaling Gateway
POI SH
Point of Interconnect Single-Homed
PRA SIGTRAN
Primary Rate Access Signaling Transport over IP
PSP SLS
Primary Signaling Point Signaling Link Selection
PSS SMS
Path Selection Scheme Short Message Service
QoS SMS-SC
Quality of Service Short Message Service Center
RANAP
SP
Radio Access Network Application Part
Signaling Point
RNC
Radio Network Controller SPC
Signaling Point Code
RTO
Retransmission time-out SPX
Signaling Proxy
RTT
Round-Trip Time SPU
Signal Processor Unit
SCCP
Signaling Connection Control Part SRP
Signaling Relay Point
SCF
Service Control Function SR
Site Router
SCP
Service Control Point SS7
Signaling System No. 7
SCTP
Stream Control Transmission Protocol SS7oIP
Signaling System No. 7 over IP
SEP
Signaling End Point SSP
Secondary Signaling Point
SGP
Signaling Gateway Process STC
Signaling Transport Converter
SGsAP
SGs Application Part STP
Signaling Transfer Point
SGSN
Serving GPRS Support Node
SUA
Signaling Connection Control Part User
Adaptation Layer
TCAP
Transaction Capabilities Application Part
TCP
Transmission Control Protocol
TDM
Time Division Multiplexing
TMT
Traffic Mode Type
TSC
Transit Switching Center
TSC-S
TSC Server
TTC
Telecommunication Technology Committee
(Japan)
TUP
Telephony User Part
ULP
Upper Layer Protocol
VIP
Virtual IP
VLR
Visitor Location Register
VM
Voicemail
VPN
Virtual Private Network
VR
Virtual Router
WCDMA
Wideband Code Division Multiple Access
Reference List
Standard References
[44] 3GPP, 29.202, Signalling System No. 7 (SS7) signalling transport in core
network; Stage 3
[47] IETF, RFC 4666, Signaling System 7 (SS7) Message Transfer Part 3 (MTP3)
- User Adaptation Layer (M3UA)
[48] IETF, RFC 4165, Signaling System 7 (SS7) Message Transfer Part 2 (MTP2)
- User Peer-to-Peer Adaptation Layer (M2PA)
[49] IETF, RFC 4233, Integrated Services Digital Network (ISDN) Q.921-User
Adaptation Layer
[50] IETF, RFC 4594, Configuration Guidelines for DiffServ Service Classes