| SAM Research Group | M. Waehlisch |
| Internet-Draft | link-lab & FU Berlin |
| Intended status: Experimental | T C. Schmidt |
| Expires: February 02, 2013 | HAW Hamburg |
| S. Venaas | |
| cisco Systems | |
| August 03, 2012 |
A Common API for Transparent Hybrid Multicast
draft-irtf-samrg-common-api-06
Group communication services exist in a large variety of flavors, and technical implementations at different protocol layers. Multicast data distribution is most efficiently performed on the lowest available layer, but a heterogeneous deployment status of multicast technologies throughout the Internet requires an adaptive service binding at runtime. Today, it is difficult to write an application that runs everywhere and at the same time makes use of the most efficient multicast service available in the network. Facing robustness requirements, developers are frequently forced to use a stable, upper layer protocol provided by the application itself. This document describes a common multicast API that is suitable for transparent communication in underlay and overlay, and grants access to the different multicast flavors. It proposes an abstract naming by multicast URIs and discusses mapping mechanisms between different namespaces and distribution technologies. Additionally, it describes the application of this API for building gateways that interconnect current multicast domains throughout the Internet, and reports on an implementation of the programming interface including a service middleware. This document is a product of the Scalable Adaptive Multicast Research Group (SAM) Research Group.
This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.
Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet- Drafts is at http:/⁠/⁠datatracker.ietf.org/⁠drafts/⁠current/⁠.
Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress."
This Internet-Draft will expire on February 02, 2013.
Copyright (c) 2012 IETF Trust and the persons identified as the document authors. All rights reserved.
This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (http:/⁠/⁠trustee.ietf.org/⁠license-⁠info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License.
Currently, group application programmers need to make the choice of the distribution technology that the application will require at runtime. There is no common communication interface that abstracts multicast transmission and subscriptions from the deployment state at runtime, nor has been the use of DNS for group addresses established. The standard multicast socket options [RFC3493], [RFC3678] are bound to an IP version by not distinguishing between naming and addressing of multicast identifiers. Group communication, however, is commonly implemented in different flavors such as any source (ASM) vs. source specific multicast (SSM), on different layers (e.g., IP vs. application layer multicast), and may be based on different technologies on the same tier as with IPv4 vs. IPv6. It is the objective of this document to provide for programmers a universal access to group services.
Multicast application development should be decoupled of technological deployment throughout the infrastructure. It requires a common multicast API that offers calls to transmit and receive multicast data independent of the supporting layer and the underlying technological details. For inter-technology transmissions, a consistent view on multicast states is needed, as well. This document describes an abstract group communication API and core functions necessary for transparent operations. Specific implementation guidelines with respect to operating systems or programming languages are out of scope of this document.
In contrast to the standard multicast socket interface, the API introduced in this document abstracts naming from addressing. Using a multicast address in the current socket API predefines the corresponding routing layer. In this specification, the multicast name used for joining a group denotes an application layer data stream that is identified by a multicast URI, independent of its binding to a specific distribution technology. Such a group name can be mapped to variable routing identifiers.
The aim of this common API is twofold:
Multicast technologies may be of various peer-to-peer kinds, IPv4 or IPv6 network layer multicast, or implemented by some other application service. Corresponding namespaces may be IP addresses or DNS naming, overlay hashes, or other application layer group identifiers like <sip:*@peanuts.org>, but also names independently defined by the applications. Common namespaces are introduced later in this document, but follow an open concept suitable for further extensions.
This document also discusses mapping mechanisms between different namespaces and forwarding technologies and proposes expressions of defaults for an intended binding. Additionally, the multicast API provides internal Interfaces to access current multicast states at the host. Multiple multicast protocols may run in parallel on a single host. These protocols may interact to provide a gateway function that bridges data between different domains. The application of this API at gateways operating between current multicast instances throughout the Internet is described, as well. Finally, a report is presented on an implementation of the programming interface including a service middleware.
This document represents the consensus of the SAM Research Group. It has been reviewed by the Research Group members active in the specific area of work. In addition, this document has been comprehensively reviewed by people who are not "in" the Research Group but are expert in the area.
The following generic use cases can be identified that require an abstract common API for multicast services:
On a very high-level, the common multicast API provides the application programmer with one single interface to manage multicast content independent of the technology underneath. Considering the following simple example in Figure 1: A multicast source S is connected via IPv4 and IPv6. It distributes one piece of multicast content (e.g., a movie). Receivers are connected via IPv4/v6 and overlay multicast respectively.
+-------+ +-------+ / +-------+
| S | | R1 | | R3 |
+-------+ +-------+ +-------+
v6| v4| |v4 |OLM
| | / |
| ***| *** ***/ ** *** /*** *** ***
\* |* ** /** * * /* ** ** *
*\ \_______/_______*__v4__+-------+ * / *
*\ IPv4/v6 * | R2 |__OLM__ *_/ Overlay Mcast *
* \_________________*__v6__+-------+ * *
* ** ** ** * * ** ** ** *
*** *** *** *** *** *** *** ***
Figure 1: Source S sends the same multicast content to all interfaces
Using the current socket API, the application programmer needs to decide on the IP technologies at coding time. Additional distribution techniques, such as overlay multicast, must be individually integrated into the application. For each technology, the application programmer needs to create a separate socket and to initiate a dedicated join or send. As the current socket API does not distinguish between group name and group address, the content will be delivered multiple times to the same receiver (cf., R2). Whenever the source distributes content via a technology that is not supported by the receivers or its Internet Service Provider (cf., R3), a gateway is required. Gateway functions rely on a coherent view of the multicast group states.
The common multicast API simplifies programming of multicast applications as it abstracts content distribution from specific technologies. In addition to calls which implement receiving and sending of multicast data, it provides service calls to grant access to internal multicast states at the host.
The API described in this document defines a minimal set of interfaces for the system components at the host to fulfill group communication. It is open to the implementation to provide additional convenience functions for the programmer.
The implementation of content distribution for the example shown in Figure 1 may then look like:
//Initialize multicast socket
MulticastSocket m = new MulticastSocket();
//Associate all available interfaces
m.addInterface(getInterfaces());
//Subscribe to multicast group
m.join(URI("opaque://news@cnn.com"));
//Send to multicast group
m.send(URI("opaque://news@cnn.com"),message);
The gateway function for R2 can be implemented by the service calls similar to:
//Initialize multicast socket
MulticastSocket m = new MulticastSocket();
//Check (a) host is designated multicast node for this interface
// (b) receivers exist
for all this.getInterfaces() {
if(designatedHost(this.interface) &&
childrenSet(this.interface,
URI("opaque://news@cnn.com")) != NULL) {
m.addInterface(this.interface);
}
}
while(true) {
m.send(URI("opaque://news@cnn.com"),message);
}
Multi-resolution multicast adjusts the multicast stream to consider heterogeneous end devices. The multicast data (e.g., available by different compression levels) is typically announced using multiple multicast addresses, which are unrelated to each other. Using the common API, multi-resolution multicast can be implemented transparently by an operator with the help of Name-to-Address mapping, or by systematic naming in a subscriber-centric perspective.
This document uses the terminology as defined for the multicast protocols [RFC2710],[RFC3376],[RFC3810],[RFC4601],[RFC4604]. In addition, the following terms will be used.
The default use case addressed in this document targets at applications that participate in a group by using some common identifier taken from some common namespace. This Group Name is typically learned at runtime from user interaction like the selection of an IPTV channel, from dynamic session negotiations like in the Session Initiation Protocol (SIP) [RFC3261], but may as well have been predefined for an application as a common Group Name. Technology-specific system functions then transparently map the Group Name to Group Addresses such that
This document considers two reference scenarios that cover the following hybrid deployment cases displayed in Figure 2:
+-------+ +-------+
| Member| | Member|
| Foo | | G |
+-------+ +-------+
\ /
*** *** *** ***
* ** ** ** *
* *
* MCast Tec A *
* *
* ** ** ** *
*** *** *** ***
+-------+ +-------+ |
| Member| | Member| +-------+
| G | | Foo | | IMG |
+-------+ +-------+ +-------+
| | |
*** *** *** *** *** *** *** ***
* ** ** ** * * ** ** ** *
* * +-------+ * *
* MCast Tec A * --| IMG |-- * MCast Tec B * +-------+
* * +-------+ * * - | Member|
* ** ** ** * * ** ** ** * | G |
*** *** *** *** *** *** *** *** +-------+
Figure 2: Reference scenarios for hybrid multicast, interconnecting group members from isolated homogeneous and heterogeneous domains.
The group communication API consists of four parts. Two parts combine the essential communication functions, while the remaining two offer optional extensions for an enhanced monitoring and management:
Multicast applications that use the common API require assistance by a group communication stack. This protocol stack serves two needs:
A general initiation of a multicast communication in this setting proceeds as follows:
The abstract multicast socket describes a group communication channel composed of one or multiple Interfaces. A socket may be created without explicit Interface association by the application, which leaves the choice of the underlying forwarding technology to the group communication stack. However, an application may also bind the socket to one or multiple dedicated Interfaces, which predefines the forwarding technology and the Multicast Namespace(s) of the Group Address(es).
Applications are not required to maintain mapping states for Group Addresses. The group communication stack accounts for the mapping of the Group Name to the Group Address(es) and vice versa. Multicast data passed to the application will be augmented by the corresponding Group Name. Multiple multicast subscriptions thus can be conducted on a single multicast socket without the need for Group Name encoding at the application side.
Hosts may support several multicast protocols. The group communication stack discovers available multicast-enabled Interfaces. It provides a minimal hybrid function that bridges data between different Interfaces and Multicast Domains. Details of service discovery are out of scope of this document.
The extended multicast functions can be implemented by a middleware as conceptually visualized in Figure 3.
*-------* *-------*
| App 1 | | App 2 |
*-------* *-------*
| |
*---------------------* ---|
| Middleware | |
*---------------------* |
| | |
*---------* | |
| Overlay | | \ Group Communication
*---------* | / Stack
| | |
| | |
*---------------------* |
| Underlay | |
*---------------------* ---|
Figure 3: A middleware for offering uniform access to multicast in underlay and overlay
Applications use Group Names to identify groups. Names can uniquely determine a group in a global communication context and hide technological deployment for data distribution from the application. In contrast, multicast forwarding operates on Group Addresses. Even though both identifiers may be identical in symbols, they carry different meanings. They may also belong to different Multicast Namespaces. The Namespace of a Group Address reflects a routing technology, while the Namespace of a Group Name represents the context in which the application operates.
URIs [RFC3986] are a common way to represent Namespace-specific identifiers in applications in the form of an abstract meta-data type. Throughout this document, all Group Names follows a URI notation with the syntax defined in Section 4.2.1. Examples are, ip://224.1.2.3:5000 for a canonical IPv4 ASM group at UDP port 5000, sip://news@cnn.com for an application-specific naming with service instantiator and default port selection.
An implementation of the group communication stack can provide convenience functions that detect the Namespace of a Group Name or further optimize service instantiation. In practice, such a library would provide support for high-level data types to the application, similar to some versions of the current socket API (e.g., InetAddress in Java). Using this data type could implicitly determine the Namespace. Details of automatic Namespace identification or service handling are out of scope of this document.
Namespace identifiers in URIs are placed in the scheme element and characterize syntax and semantic of the group identifier. They enable the use of convenience functions and high-level data types while processing URIs. When used in names, they may indicate an application context, or facilitate a default mapping and a recovery of names from addresses. They characterize its type, when used in addresses.
Compliant to the URI concept, namespace-schemes can be added. Examples of schemes are generic or inherited from applications.
The multicast communication paradigm requires all group members to subscribe to the same Group Name, taken from a common Multicast Namespace, and thereby to identify the group in a technology-agnostic way. Following this common API, a sender correspondingly registers a Group Name prior to transmission.
At communication end points, Group Names require a mapping to Group Addresses prior to service instantiation at its Interface(s). Similarly, a mapping is needed at gateways to translate between Group Addresses from different namespaces consistently. Two requirements need to be met by a mapping function that translates between Multicast Names and Addresses.
In general, mapping can be complex and need not be invertible. A mapping can be realized by embedding smaller in larger namespaces or selecting an arbitrary, unused ID in a smaller target namespace. For example, it is not obvious how to map a large identifier space (e.g., IPv6) to a smaller, collision-prone set like IPv4 (see [I-D.venaas-behave-v4v6mc-framework][I-D.venaas-behave-mcast46], [RFC6219] ). Mapping functions can be stateless in some contexts, but may require states in others. The application of such functions depends on the cardinality of the namespaces, the structure of address spaces, and possible address collisions. However, some namespaces facilitate a canonical, invertible transformation to default address spaces.
Some Multicast Namespaces defined in Section 3.4 can express a canonical default mapping. For example, ip://224.1.2.3:5000 indicates the correspondence to 224.1.2.3 in the default IPv4 multicast address space at port 5000. This default mapping is bound to a technology and may not always be applicable, e.g., in the case of address collisions. Note that under canonical mapping, the multicast URI can be completely recovered from any data message received from this group.
Multicast listeners or senders require a Name-to-Address conversion for all technologies they actively run in a group. Even though a mapping applies to the local Multicast Domain only, end points may need to learn a valid Group Address from neighboring nodes, e.g., from a gateway in the collision-prone IPv4 domain. Once set, an end point will always be aware of the Name-to-Address correspondence and thus can autonomously invert the mapping.
Multicast data may arrive at an IMG in one technology, requesting the gateway to re-address packets for another distribution system. At initial arrival, the IMG may not have explicit knowledge of the corresponding Multicast Group Name. To perform a consistent mapping, the group name needs to be acquired. It may have been distributed at source registration, or may have been learned from a neighboring node, details of which are beyond the scope of this document.
In Explicit Multicast (XCAST) [RFC5058], the multicast source explicitly pre-defines the receivers. From a conceptual perspective, XCAST is an additional distribution technology (i.e., a new technology-specific interface) for this API. The instantiation part of the Group Name may refer to multiple receivers. However, from an implementation perspective, this specification defines the syntax of the Group Name according to [RFC3986], which does not support a set of (sub-)identifiers. Extending [RFC3986] is out of scope of this document.
Implementing XCAST without a set representation in the Group Name requires a topology-dependent mapping of the Name to a set of subscribers. Defining these details is out of scope of this document.
This API considers a multi-technology scenario, in which different technologies may have different Maximum Transmission Unit (MTU) sizes. Even if the MTU size between two hosts has been determined, it may change over time either initiated by the network (e.g., path changes) or by the end hosts (e.g., interface change due to mobility).
The design of this API is based on the objective of robust communication and easy application development. The MTU handling and the placement of fragmentation is thus guided by the following observations.
The concept of this API aims at guaranteeing a maximum message size for the application programmer, thereby to handle fragmentation at the interface level, if needed. Nevertheless, the application programmer should be able to determine the technology-specific atomic message size to optimize data distribution or for other reasons.
The maximum message size should take realistic values (e.g., following IP clients) to enable smooth and efficient services. A detailed selection scheme of MTU values is out of scope of this document.
The following description of the common multicast API is described in pseudo syntax. Variables that are passed to function calls are declared by "in", return values are declared by "out". A list of elements is denoted by <>. The pseudo syntax assumes that lists include an attribute which represents the number of elements.
The corresponding C signatures are defined in Appendix Appendix A.
Multicast Names and Multicast Addresses used in this API follow an URI scheme that defines a subset of the generic URI specified in [RFC3986] and is syntactically compliant with the guidelines given in [RFC4395].
The multicast URI is defined as follows:
The parts of the URI are defined as follows:
The Interface denotes the layer and instance on which the corresponding call will be effective. In agreement with [RFC3493], we identify an Interface by an identifier, which is a positive integer starting at 1.
Properties of an Interface are stored in the following data structure:
struct if_prop {
unsigned int if_index; /* 1, 2, ... */
char *if_name; /* "eth0", "eth1:1", "lo", ... */
char *if_addr; /* "1.2.3.4", "abc123", ... */
char *if_tech; /* "ip", "overlay", ... */
};
The following function retrieves all available Interfaces from the system:
getInterfaces(out Interface <ifs>);
It extends the functions for Interface Identification defined in Section 4 of [RFC3493] and can be implemented by:
struct if_prop *if_prop(void);
A membership event is triggered by a multicast state change, which is observed by the current node. It is related to a specific Group Name and may be receiver or source oriented.
event_type {
join_event;
leave_event;
new_source_event;
};
event {
event_type event;
Uri group_name;
Interface if;
};
An event will be created by the group communication stack and passed to applications that have registered for events.
The create call initiates a multicast socket and provides the application programmer with a corresponding handle. If no Interfaces will be assigned based on the call, the default Interface will be selected and associated with the socket. The call returns an error code in the case of failures, e.g., due to a non-operational communication middleware.
createMSocket(in Interface <ifs>,
out Socket s);
The ifs argument denotes a list of Interfaces (if_indexes) that will be associated with the multicast socket. This parameter is optional.
On success a multicast socket identifier is returned, otherwise NULL.
The delete call removes the multicast socket.
deleteMSocket(in Socket s, out Int error);
The s argument identifies the multicast socket for destruction.
On success the out parameter error is 0, otherwise -1.
The join call initiates a subscription for the given Group Name. Depending on the Interfaces that are associated with the socket, this may result in an IGMP/MLD report or overlay subscription, for example.
join(in Socket s, in Uri groupName, out Int error);
The s argument identifies the multicast socket.
The groupName argument identifies the group.
On success the out parameter error is 0, otherwise -1.
The leave call results in an unsubscription for the given Group Name.
leave(in Socket s, in Uri groupName, out Int error);
The s argument identifies the multicast socket.
The groupName identifies the group.
On success the out parameter error is 0, otherwise -1.
The srcRegister call registers a source for a Group on all active Interfaces of the socket s. This call may assist group distribution in some technologies, for example the creation of sub-overlays or may facilitate a name-to-address mapping. Likewise, it may remain without effect in some multicast technologies.
srcRegister(in Socket s, in Uri groupName,
out Interface <ifs>, out Int error);
The s argument identifies the multicast socket.
The groupName argument identifies the multicast group to which a source intends to send data.
The ifs argument points to the list of Interface indexes for which the source registration failed. A NULL pointer is returned, if the list is empty. This parameter is optional.
If source registration succeeded for all Interfaces associated with the socket, the out parameter error is 0, otherwise -1.
The srcDeregister indicates that a source does no longer intend to send data to the multicast group. This call may remain without effect in some multicast technologies.
srcDeregister(in Socket s, in Uri groupName,
out Interface <ifs>, out Int error);
The s argument identifies the multicast socket.
The group_name argument identifies the multicast group to which a source has stopped to send multicast data.
The ifs argument points to the list of Interfaces for which the source deregistration failed. A NULL pointer is returned, if the list is empty.
If source deregistration succeeded for all Interfaces associated with the socket, the out parameter error is 0, otherwise -1.
The send call passes multicast data destined for a Multicast Name from the application to the multicast socket.
It is worth noting that it is the choice of the programmer to send data via one socket per group or to use a single socket for multiple groups.
send(in Socket s, in Uri groupName,
in Size msgLen, in Msg msgBuf,
out Int error);
The s argument identifies the multicast socket.
The groupName argument identifies the group to which data will be sent.
The msgLen argument holds the length of the message to be sent.
The msgBuf argument passes the multicast data to the multicast socket.
On success the out parameter error is 0, otherwise -1. A message that is too long is indicated by an implementation-specific error code (e.g., EMSGSIZE in C).
The receive call passes multicast data and the corresponding Group Name to the application. This may come in a blocking or non-blocking variant.
It is worth noting that it is the choice of the programmer to receive data via one socket per group or to use a single socket for multiple groups.
receive(in Socket s, out Uri groupName,
out Size msgLen, out Msg msgBuf,
out Int error);
The s argument identifies the multicast socket.
The group_name argument identifies the multicast group for which data was received.
The msgLen argument holds the length of the received message.
The msgBuf argument points to the payload of the received multicast data.
On success the out parameter error is 0, otherwise -1. A message that is too long is indicated by an implementation-specific error handling (e.g., EMSGSIZE).
The following calls configure an existing multicast socket.
The getInterface call returns an array of all available multicast communication Interfaces associated with the multicast socket.
getInterfaces(in Socket s,
out Interface <ifs>, out Int error);
The s argument identifies the multicast socket.
The ifs argument points to an array of Interface index identifiers.
On success the out parameter error is 0, otherwise -1.
The addInterface call adds a distribution channel to the socket. This may be an overlay or underlay Interface, e.g., IPv6 or DHT. Multiple Interfaces of the same technology may be associated with the socket.
addInterface(in Socket s, in Interface if,
out Int error);
The s and if arguments identify a multicast socket and Interface, respectively.
On success the value 0 is returned, otherwise -1.
The delInterface call removes the Interface if from the multicast socket.
delInterface(in Socket s, Interface if,
out Int error);
The s and if arguments identify a multicast socket and Interface, respectively.
On success the out parameter error is 0, otherwise -1.
The setTTL call configures the maximum hop count for the socket a multicast message is allowed to traverse.
setTTL(in Socket s, in Int h,
in Interface <ifs>,
out Int error);
The s and h arguments identify a multicast socket and the maximum hop count, respectively.
The ifs argument points to an array of Interface index identifiers. This parameter is optional.
On success the out parameter error is 0, otherwise -1.
The getTTL call returns the maximum hop count a multicast message is allowed to traverse for the socket.
getTTL(in Socket s,
out Int h, out Int error);
The s argument identifies a multicast socket.
The h argument holds the maximum number of hops associated with socket s.
On success the out parameter error is 0, otherwise -1.
The getAtomicMsgSize function returns the maximum message size that an application is allowed to transmit per socket at once without fragmentation. This value depends on the interfaces associated with the socket in use and thus may change during runtime.
getAtomicMsgSize(in Socket s,
out Int return);
On success, the function returns a positive value of appropriate message size, otherwise -1.
The groupSet call returns all multicast groups registered at a given Interface. This information can be provided by group management states or routing protocols. The return values distinguish between sender and listener states.
struct GroupSet {
uri groupName; /* registered multicast group */
int type; /* 0 = listener state, 1 = sender state,
2 = sender & listener state */
}
groupSet(in Interface if,
out GroupSet <groupSet>, out Int error);
The if argument identifies the Interface for which states are maintained.
The groupSet argument points to a list of group states.
On success the out parameter error is 0, otherwise -1.
The neighborSet function returns the set of neighboring nodes for a given Interface as seen by the multicast routing protocol.
neighborSet(in Interface if,
out Uri <neighborsAddresses>, out Int error);
The if argument identifies the Interface for which neighbors are inquired.
The neighborsAddresses argument points to a list of neighboring nodes on a successful return.
On success the out parameter error is 0, otherwise -1.
The childrenSet function returns the set of child nodes that receive multicast data from a specified Interface for a given group. For a common multicast router, this call retrieves the multicast forwarding information base per Interface.
childrenSet(in Interface if, in Uri groupName,
out Uri <childrenAddresses>, out Int error);
The if argument identifies the Interface for which children are inquired.
The groupName argument defines the multicast group for which distribution is considered.
The childrenAddresses argument points to a list of neighboring nodes on a successful return.
On success the out parameter error is 0, otherwise -1.
The parentSet function returns the set of neighbors from which the current node receives multicast data at a given Interface for the specified group.
parentSet(in Interface if, in Uri groupName,
out Uri <parentsAddresses>, out Int error);
The if argument identifies the Interface for which parents are inquired.
The groupName argument defines the multicast group for which distribution is considered.
The parentsAddresses argument points to a list of neighboring nodes on a successful return.
On success the out parameter error is 0, otherwise -1.
The designatedHost function inquires whether this host has the role of a designated forwarder resp. querier, or not. Such an information is provided by almost all multicast protocols to prevent packet duplication, if multiple multicast instances serve on the same subnet.
designatedHost(in Interface if, in Uri groupName
out Int return);
The if argument identifies the Interface for which designated forwarding is inquired.
The groupName argument specifies the group for which the host may attain the role of designated forwarder.
The function returns 1 if the host is a designated forwarder or querier, otherwise 0. The return value -1 indicates an error.
The enableEvents function registers an application at the group communication stack to receive information about group changes. State changes are the result of new receiver subscriptions or leaves as well as of source changes. Upon receiving an event, the group service may obtain additional information from further service calls.
enableEvents();
Calling this function, the stack starts to pass membership events to the application. Each event includes an event type identifier and a Group Name (cf., Section 4.2.3).
The multicast protocol has not to support membership tracking to enable this feature. This function can also be implemented at the middelware layer.
The disableEvents function deactivates the information about group state changes.
disableEvents();
On success the stack will not pass membership events to the application.
The getMaxMsgSize function returns the maximum message size that an application is allowed to transmit per socket at once. This value is statically guaranteed by the group communication stack.
getMaxMsgSize(out Int return);
On success, the function returns a positive value of allowed message size, otherwise -1.
A reference implementation of the Common API for Transparent Hybrid Multicast is available with the HAMcast stack [hamcast-dev], [GC2010], [LCN2012]. This open-source software supports the multicast API (C++ and Java library) for group application development, the middleware as a userspace system service, and several multicast-technology modules. The middleware is implemented in C++.
This API is verified and adjusted based on the real-world experiences gathered in the HAMcast project, and by additional users of the stack.
This document makes no request of IANA.
This draft does neither introduce additional messages nor novel protocol operations.
We would like to thank the HAMcast-team, Nora Berg, Dominik Charousset, Gabriel Hege, Fabian Holler, Alexander Knauf, Sebastian Meiling, Sebastian Woelke, and Sebastian Zagaria, at the HAW Hamburg for many fruitful discussions and for their continuous critical feedback while implementing the common multicast API and a hybrid multicast middleware. We gratefully acknowledge WeeSan, Mario Kolberg, and John Buford for reviewing and their suggestions to improve the document. We would like to thank the Name-based socket BoF (in particular Dave Thaler) for clarifying insights into the question of meta function calls.
This work is partially supported by the German Federal Ministry of Education and Research within the HAMcast project (see http://hamcast.realmv6.org), which is part of G-Lab.
This section describes the C signatures of the common multicast API, which are defined in Section 4.
int createMSocket(int* result, size_t num_ifs, const uint32_t* ifs);
int deleteMSocket(int s);
int join(int msock, const char* group_uri);
int leave(int msock, const char* group_uri);
int srcRegister(int msock,
const char* group_uri,
size_t num_ifs,
const uint32_t *ifs);
int srcDeregister(int msock,
const char* group_uri,
size_t num_ifs,
const uint32_t *ifs);
int send(int msock,
const char* group_uri,
size_t buf_len,
const void* buf);
int receive(int msock,
const char* group_uri,
size_t buf_len,
void* buf);
int getInterfaces(int msock,
size_t* num_ifs,
uint32_t** ifs);
int addInterface(int msock, uint32_t iface);
int delInterface(int msock, uint32_t iface);
int setTTL(int msock, uint8_t value,
size_t num_ifs, uint32_t* ifs);
int getTTL(int msock, uint8_t* result);
int getAtomicMsgSize(int msock);
typedef struct {
char* group_uri; /* registered mcast group */
int type; /* 0: listener state,
1: sender state
2: sender and listener state */
}
GroupSet;
int groupSet(uint32_t iface,
size_t* num_groups,
GroupSet** groups);
int neighborSet(uint32_t iface,
const char* group_name,
size_t* num_neighbors,
char** neighbor_uris);
int childrenSet(uint32_t iface,
const char* group_name,
size_t* num_children,
char** children_uris);
int parentSet(uint32_t iface,
const char* group_name,
size_t* num_parents,
char** parents_uris);
int designatedHost(uint32_t iface,
const char* group_name);
typedef void (*MembershipEventCallback)(int, /* event type */
uint32_t, /* interface id */
const char*); /* group uri */
int registerEventCallback(MembershipEventCallback callback);
int disableEvents();
int getMaxMsgSize();
-- Application above middleware:
//Initialize multicast socket;
//the middleware selects all available interfaces
MulticastSocket m = new MulticastSocket();
m.join(URI("ip://224.1.2.3:5000"));
m.join(URI("ip://[FF02:0:0:0:0:0:0:3]:6000"));
m.join(URI("sip://news@cnn.com"));
-- Middleware:
join(URI mcAddress) {
//Select interfaces in use
for all this.interfaces {
switch (interface.type) {
case "ipv6":
//... map logical ID to routing address
Inet6Address rtAddressIPv6 = new Inet6Address();
mapNametoAddress(mcAddress,rtAddressIPv6);
interface.join(rtAddressIPv6);
case "ipv4":
//... map logical ID to routing address
Inet4Address rtAddressIPv4 = new Inet4Address();
mapNametoAddress(mcAddress,rtAddressIPv4);
interface.join(rtAddressIPv4);
case "sip-session":
//... map logical ID to routing address
SIPAddress rtAddressSIP = new SIPAddress();
mapNametoAddress(mcAddress,rtAddressSIP);
interface.join(rtAddressSIP);
case "dht":
//... map logical ID to routing address
DHTAddress rtAddressDHT = new DHTAddress();
mapNametoAddress(mcAddress,rtAddressDHT);
interface.join(rtAddressDHT);
//...
}
}
}
This section describes the application of the defined API to implement an IMG.
The following procedure describes a transparent mapping of a DVMRP-based any source multicast service to another many-to-many multicast technology, e.g., an overlay.
An arbitrary DVMRP [RFC1075] router will not be informed about new receivers, but will learn about new sources immediately. The concept of DVMRP does not provide any central multicast instance. Thus, the IMG can be placed anywhere inside the multicast region, but requires a DVMRP neighbor connectivity. Thus the group communication stack used by the IMG is enhanced by a DVMRP implementation. New sources in the underlay will be advertised based on the DVMRP flooding mechanism and received by the IMG. Based on this, the event "new_source_event" is created and passed to the application. The relay agent initiates a corresponding join in the native network and forwards the received source data towards the overlay routing protocol. Depending on the group states, the data will be distributed to overlay peers.
DVMRP establishes source specific multicast trees. Therefore, a graft message is only visible to DVMRP routers on the path from the new receiver subnet to the source, but in general not to an IMG. To overcome this problem, data of multicast senders in the overlay may become noticable via the Source Register call, as well as by an IMG that initiates an an all-group join in the overlay using the namespace extension of the API. Each IMG is initially required to forward the data received in the overlay to the underlay, independent of native multicast receivers. Subsequent prunes may limit unwanted data distribution thereafter.
The following procedure describes a transparent mapping of a PIM-SM-based any source multicast service to another many-to-many multicast technology, e.g., an overlay.
The Protocol Independent Multicast Sparse Mode (PIM-SM) [RFC4601] establishes rendezvous points (RP). These entities receive listener subscriptions and source registering of a domain. For a continuous update an IMG has to be co-located with an RP. Whenever PIM register messages are received, the IMG must signal internally a new multicast source using the event "new_source_event". Subsequently, the IMG joins the group and a shared tree between the RP and the sources will be established, which may change to a source specific tree after PIM switches to phase three. Source traffic will be forwarded to the RP based on the IMG join, even if there are no further receivers in the native multicast domain. Designated routers of a PIM-domain send receiver subscriptions towards the PIM-SM RP. The reception of such messages initiates the event "join_event" at the IMG, which initiates a join towards the overlay routing protocol. Overlay multicast data arriving at the IMG will then transparently be forwarded in the underlay network and distributed through the RP instance.
The following procedure describes a transparent mapping of a PIM-SSM-based source specific multicast service to another one-to-many multicast technology, e.g., an overlay.
PIM Source Specific Multicast (PIM-SSM) is defined as part of PIM-SM and admits source specific joins (S,G) according to the source specific host group model [RFC4604]. A multicast distribution tree can be established without the assistance of a rendezvous point.
Sources are not advertised within a PIM-SSM domain. Consequently, an IMG cannot anticipate the local join inside a sender domain and deliver a priori the multicast data to the overlay instance. If an IMG of a receiver domain initiates a group subscription via the overlay routing protocol, relaying multicast data fails, as data is not available at the overlay instance. The IMG instance of the receiver domain, thus, has to locate the IMG instance of the source domain to trigger the corresponding join. In agreement with the objectives of PIM-SSM, the signaling should not be flooded in underlay and overlay.
A solution can be to intercept the subscription at both, source and receiver sites: To monitor multicast receiver subscriptions ("join_event" or "leave_event") in the underlay, the IMG is placed on path towards the source, e.g., at a domain border router. This router intercepts join messages and extracts the unicast source address S, initializing an IMG specific join to S via regular unicast. Multicast data arriving at the IMG of the sender domain can be distributed via the overlay. Discovering the IMG of a multicast sender domain may be implemented analogously to AMT [I-D.ietf-mboned-auto-multicast] by anycast. Consequently, the source address S of the group (S,G) should be built based on an anycast prefix. The corresponding IMG anycast address for a source domain is then derived from the prefix of S.
The following procedure describes a transparent mapping of a BIDIR-PIM-based any source multicast service to another many-to-many multicast technology, e.g., an overlay.
Bidirectional PIM [RFC5015] is a variant of PIM-SM. In contrast to PIM-SM, the protocol pre-establishes bidirectional shared trees per group, connecting multicast sources and receivers. The rendezvous points are virtualized in BIDIR-PIM as an address to identify on-tree directions (up and down). Routers with the best link towards the (virtualized) rendezvous point address are selected as designated forwarders for a link-local domain and represent the actual distribution tree. The IMG is to be placed at the RP-link, where the rendezvous point address is located. As source data in either cases will be transmitted to the rendezvous point link, the BIDIR-PIM instance of the IMG receives the data and can internally signal new senders towards the stack via the "new_source_event". The first receiver subscription for a new group within a BIDIR-PIM domain needs to be transmitted to the RP to establish the first branching point. Using the "join_event", an IMG will thereby be informed about group requests from its domain, which are then delegated to the overlay.
The following changes have been made from draft-irtf-samrg-common-api-05
The following changes have been made from draft-irtf-samrg-common-api-04
The following changes have been made from draft-irtf-samrg-common-api-03
The following changes have been made from draft-irtf-samrg-common-api-02
The following changes have been made from draft-irtf-samrg-common-api-01
The following changes have been made from draft-irtf-samrg-common-api-00
The following changes have been made from draft-waehlisch-sam-common-api-06
The following changes have been made from draft-waehlisch-sam-common-api-05
The following changes have been made from draft-waehlisch-sam-common-api-04
The following changes have been made from draft-waehlisch-sam-common-api-03
The following changes have been made from draft-waehlisch-sam-common-api-02
The following changes have been made from draft-waehlisch-sam-common-api-01