RFC 7285
Application-Layer Traffic Optimization (ALTO) Protocol
Internet Engineering Task Force (IETF) R. Alimi, Ed. Request for Comments: 7285 Google Category: Standards Track R. Penno, Ed. ISSN: 2070-1721 Cisco Systems, Inc. Y. Yang, Ed. Yale University S. Kiesel University of Stuttgart S. Previdi Cisco Systems, Inc. W. Roome Alcatel-Lucent S. Shalunov Open Garden R. Woundy Comcast September 2014 Application-Layer Traffic Optimization (ALTO) ProtocolAbstract
Applications using the Internet already have access to some topology information of Internet Service Provider (ISP) networks. For example, views to Internet routing tables at Looking Glass servers are available and can be practically downloaded to many network application clients. What is missing is knowledge of the underlying network topologies from the point of view of ISPs. In other words, what an ISP prefers in terms of traffic optimization -- and a way to distribute it. The Application-Layer Traffic Optimization (ALTO) services defined in this document provide network information (e.g., basic network location structure and preferences of network paths) with the goal of modifying network resource consumption patterns while maintaining or improving application performance. The basic information of ALTO is based on abstract maps of a network. These maps provide a simplified view, yet enough information about a network for applications to effectively utilize them. Additional services are built on top of the maps. This document describes a protocol implementing the ALTO services. Although the ALTO services would primarily be provided by ISPs, other entities, such as content service providers, could also provide ALTO services. Applications that could use the ALTO services are those that have a choice to which end points to connect. Examples of such applications are peer-to-peer (P2P) and content delivery networks.
Status of This Memo This is an Internet Standards Track document. This document is a product of the Internet Engineering Task Force (IETF). It represents the consensus of the IETF community. It has received public review and has been approved for publication by the Internet Engineering Steering Group (IESG). Further information on Internet Standards is available in Section 2 of RFC 5741. Information about the current status of this document, any errata, and how to provide feedback on it may be obtained at http://www.rfc-editor.org/info/rfc7285. Copyright Notice Copyright (c) 2014 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.Table of Contents
1. Introduction ....................................................6 1.1. Problem Statement ..........................................6 1.1.1. Requirements Language ...............................7 1.2. Design Overview ............................................7 2. Terminology .....................................................7 2.1. Endpoint ...................................................8 2.2. Endpoint Address ...........................................8 2.3. Network Location ...........................................8 2.4. ALTO Information ...........................................8 2.5. ALTO Information Base ......................................8 3. Architecture ....................................................8 3.1. ALTO Services and Protocol Scope ...........................9 3.2. ALTO Information Reuse and Redistribution .................11 4. ALTO Information Service Framework .............................11 4.1. ALTO Information Services .................................12 4.1.1. Map Service ........................................12 4.1.2. Map-Filtering Service ..............................12
4.1.3. Endpoint Property Service ..........................12
4.1.4. Endpoint Cost Service ..............................13
5. Network Map ....................................................13
5.1. Provider-Defined Identifier (PID) .........................13
5.2. Endpoint Addresses ........................................14
5.3. Example Network Map .......................................14
6. Cost Map .......................................................15
6.1. Cost Types ................................................16
6.1.1. Cost Metric ........................................16
6.1.2. Cost Mode ..........................................17
6.2. Cost Map Structure ........................................18
6.3. Network Map and Cost Map Dependency .......................18
6.4. Cost Map Update ...........................................19
7. Endpoint Properties ............................................19
7.1. Endpoint Property Type ....................................19
7.1.1. Endpoint Property Type: pid ........................19
8. Protocol Specification: General Processing .....................19
8.1. Overall Design ............................................19
8.2. Notation ..................................................20
8.3. Basic Operations ..........................................21
8.3.1. Client Discovering Information Resources ...........21
8.3.2. Client Requesting Information Resources ............22
8.3.3. Server Responding to Information Resource Request ..22
8.3.4. Client Handling Server Response ....................23
8.3.5. Authentication and Encryption ......................23
8.3.6. Information Refreshing .............................24
8.3.7. Parsing of Unknown Fields ..........................24
8.4. Server Response Encoding ..................................24
8.4.1. Meta Information ...................................24
8.4.2. Data Information ...................................25
8.5. Protocol Errors ...........................................25
8.5.1. Media Type .........................................25
8.5.2. Response Format and Error Codes ....................25
8.5.3. Overload Conditions and Server Unavailability ......28
9. Protocol Specification: Information Resource Directory .........28
9.1. Information Resource Attributes ...........................29
9.1.1. Resource ID ........................................29
9.1.2. Media Type .........................................29
9.1.3. Capabilities .......................................29
9.1.4. Accepts Input Parameters ...........................29
9.1.5. Dependent Resources ................................30
9.2. Information Resource Directory (IRD) ......................30
9.2.1. Media Type .........................................30
9.2.2. Encoding ...........................................30
9.2.3. Example ............................................32
9.2.4. Delegation Using IRD ...............................35
9.2.5. Considerations of Using IRD ........................37
10. Protocol Specification: Basic Data Types ......................38
10.1. PID Name .................................................38
10.2. Resource ID ..............................................38
10.3. Version Tag ..............................................38
10.4. Endpoints ................................................39
10.4.1. Typed Endpoint Addresses ..........................39
10.4.2. Address Type ......................................39
10.4.3. Endpoint Address ..................................40
10.4.4. Endpoint Prefixes .................................40
10.4.5. Endpoint Address Group ............................41
10.5. Cost Mode ................................................41
10.6. Cost Metric ..............................................42
10.7. Cost Type ................................................42
10.8. Endpoint Property ........................................42
10.8.1. Resource-Specific Endpoint Properties .............43
10.8.2. Global Endpoint Properties ........................43
11. Protocol Specification: Service Information Resources .........43
11.1. Meta Information .........................................43
11.2. Map Service ..............................................43
11.2.1. Network Map .......................................44
11.2.2. Mapping IP Addresses to PIDs for
'ipv4'/'ipv6' Network Maps ........................46
11.2.3. Cost Map ..........................................47
11.3. Map-Filtering Service ....................................50
11.3.1. Filtered Network Map ..............................50
11.3.2. Filtered Cost Map .................................53
11.4. Endpoint Property Service ................................57
11.4.1. Endpoint Property .................................58
11.5. Endpoint Cost Service ....................................61
11.5.1. Endpoint Cost .....................................61
12. Use Cases .....................................................64
12.1. ALTO Client Embedded in P2P Tracker ......................65
12.2. ALTO Client Embedded in P2P Client: Numerical Costs ......66
12.3. ALTO Client Embedded in P2P Client: Ranking ..............67
13. Discussions ...................................................68
13.1. Discovery ................................................68
13.2. Hosts with Multiple Endpoint Addresses ...................68
13.3. Network Address Translation Considerations ...............69
13.4. Endpoint and Path Properties .............................69
14. IANA Considerations ...........................................70
14.1. application/alto-* Media Types ...........................70
14.2. ALTO Cost Metric Registry ................................71
14.3. ALTO Endpoint Property Type Registry .....................73
14.4. ALTO Address Type Registry ...............................75
14.5. ALTO Error Code Registry .................................76
15. Security Considerations .......................................76
15.1. Authenticity and Integrity of ALTO Information ...........77
15.1.1. Risk Scenarios ....................................77
15.1.2. Protection Strategies .............................77
15.1.3. Limitations .......................................77
15.2. Potential Undesirable Guidance from Authenticated ALTO
Information ..............................................78
15.2.1. Risk Scenarios ....................................78
15.2.2. Protection Strategies .............................78
15.3. Confidentiality of ALTO Information ......................79
15.3.1. Risk Scenarios ....................................79
15.3.2. Protection Strategies .............................79
15.3.3. Limitations .......................................80
15.4. Privacy for ALTO Users ...................................80
15.4.1. Risk Scenarios ....................................80
15.4.2. Protection Strategies .............................80
15.5. Availability of ALTO Services ............................81
15.5.1. Risk Scenarios ....................................81
15.5.2. Protection Strategies .............................81
16. Manageability Considerations ..................................81
16.1. Operations ...............................................82
16.1.1. Installation and Initial Setup ....................82
16.1.2. Migration Path ....................................82
16.1.3. Dependencies on Other Protocols and
Functional Components .............................83
16.1.4. Impact and Observation on Network Operation .......83
16.2. Management ...............................................84
16.2.1. Management Interoperability .......................84
16.2.2. Management Information ............................84
16.2.3. Fault Management ..................................84
16.2.4. Configuration Management ..........................84
16.2.5. Performance Management ............................85
16.2.6. Security Management ...............................85
17. References ....................................................85
17.1. Normative References .....................................85
17.2. Informative References ...................................86
Appendix A. Acknowledgments .......................................89
Appendix B. Design History and Merged Proposals ...................90
1. Introduction
1.1. Problem Statement
This document defines the ALTO Protocol, which provides a solution for the problem stated in [RFC5693]. Specifically, in today's networks, network information such as network topologies, link availability, routing policies, and path costs are hidden from the application layer, and many applications benefited from such hiding of network complexity. However, new applications, such as application-layer overlays, can benefit from information about the underlying network infrastructure. In particular, these new network applications can be adaptive; hence, they can become more network efficient (e.g., reduce network resource consumption) and achieve better application performance (e.g., accelerated download rate), by leveraging network-provided information. At a high level, the ALTO Protocol specified in this document is an information-publishing interface that allows a network to publish its network information such as network locations, costs between them at configurable granularities, and endhost properties to network applications. The information published by the ALTO Protocol should benefit both the network and the applications (i.e., the consumers of the information). Either the operator of the network or a third party (e.g., an information aggregator) can retrieve or derive related information of the network and publish it using the ALTO Protocol. To allow better understanding of the goal of the ALTO Protocol, this document provides a short, non-normative overview of the benefits of ALTO to both networks and applications: o A network that provides ALTO information can achieve better utilization of its networking infrastructure. For example, by using ALTO as a tool to interact with applications, a network is able to provide network information to applications so that the applications can better manage traffic on more expensive or difficult-to-provision links such as long-distance, transit, or backup links. During the interaction, the network can choose to protect its sensitive and confidential network state information, by abstracting real metric values into non-real numerical scores or ordinal ranking. o An application that uses ALTO information can benefit from better knowledge of the network to avoid network bottlenecks. For example, an overlay application can use information provided by the ALTO services to avoid selecting peers connected via high- delay links (e.g., some intercontinental links). Using ALTO to
initialize each node with promising ("better-than-random") peers,
an adaptive peer-to-peer overlay may achieve faster, better
convergence.
1.1.1. Requirements Language
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
"SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
document are to be interpreted as described in RFC 2119 [RFC2119].
1.2. Design Overview
The ALTO Protocol specified in this document meets the ALTO
requirements specified in [RFC5693], and unifies multiple protocols
previously designed with similar intentions. See Appendix A for a
list of people and Appendix B for a list of proposals that have made
significant contributions to this effort.
The ALTO Protocol uses a REST-ful (Representational State Transfer
(REST)) design [Fielding-Thesis], and encodes its requests and
responses using JSON [RFC7159]. These designs are chosen because of
their flexibility and extensibility. In addition, these designs make
it possible for ALTO to be deployed at scale by leveraging existing
HTTP [RFC7230] implementations, infrastructures and deployment
experience.
The ALTO Protocol uses a modular design by dividing ALTO information
publication into multiple ALTO services (e.g., the Map service, the
Map-Filtering Service, the Endpoint Property Service, and the
Endpoint Cost Service). Each ALTO service provides a given set of
functionalities and is realized by a set of information resources,
which are announced by information resource directories, to guide
ALTO clients.
2. Terminology
This document uses the following terms defined in [RFC5693]:
Application, Overlay Network, Peer, Resource, Resource Identifier,
Resource Provider, Resource Consumer, Resource Directory, Transport
Address, ALTO Server, ALTO Client, ALTO Query, ALTO Response, ALTO
Transaction, Local Traffic, Peering Traffic, and Transit Traffic.
This document extends the term "ALTO Service" defined in [RFC5693].
In particular, by adopting a modular design, this document allows the
ALTO Protocol to provide multiple ALTO services.
This document also uses the following additional terms: Endpoint Address, Network Location, ALTO Information, and ALTO Information Base.2.1. Endpoint
An endpoint is an application or host that is capable of communicating (sending and/or receiving messages) on a network. An endpoint is typically either a resource provider or a resource consumer.2.2. Endpoint Address
An endpoint address represents the communication address of an endpoint. Common forms of endpoint addresses include IP addresses, Media Access Control (MAC) addresses, and overlay IDs. An endpoint address can be network-attachment based (e.g., IP address) or network-attachment agnostic (e.g., MAC address). Each endpoint address has an associated address type, which indicates both its syntax and semantics.2.3. Network Location
This document uses network location as a generic term to denote a single endpoint or a group of endpoints. For instance, it can be a single IPv4 or IPv6 address, an IPv4 or IPv6 prefix, or a set of prefixes.2.4. ALTO Information
This document uses ALTO information as a generic term to refer to the network information provided by an ALTO server.2.5. ALTO Information Base
This document uses the term ALTO information base to refer to the internal representation of ALTO information maintained by an ALTO server. Note that the structure of this internal representation is not defined by this document.3. Architecture
This section defines the ALTO architecture and the ALTO Protocol's place in the overall architecture.
3.1. ALTO Services and Protocol Scope
Each network region in the global Internet can provide its ALTO services, which convey network information from the perspective of that network region. A network region in this context can be an Autonomous System (AS), an ISP, a region smaller than an AS or ISP, or a set of ISPs. The specific network region that an ALTO service represents will depend on the ALTO deployment scenario and ALTO service discovery mechanism. The ALTO services specified in this document define network endpoints (and aggregations thereof) and generic costs amongst them from the region's perspective. The network endpoints may include all endpoints in the global Internet. We say that the network information provided by the ALTO services of a network region represents the "my-Internet view" of the network region. The "my-Internet view" defined in this document does not specify the internal topology of a network, and hence, it is said to provide a "single-node" abstract topology. Extensions to this document may provide topology details in "my-Internet view". Figure 1 provides an overall picture of ALTO's system architecture, so that one can better understand the ALTO services and the role of the ALTO Protocol. In this architecture, an ALTO server prepares ALTO information, an ALTO client uses ALTO service discovery to identify an appropriate ALTO server, and the ALTO client requests available ALTO information from the ALTO server using the ALTO Protocol. The ALTO information provided by the ALTO server can be updated dynamically based on network conditions, or they can be seen as a policy that is updated on a longer time scale.
+-------------------------------------------------------------------+ | Network Region | | | | +-----------+ | | | Routing | | | +--------------+ | Protocols | | | | Provisioning | +-----------+ | | | Policy | | | | +--------------+\ | | | \ | | | \ | | | +-----------+ \+---------+ +--------+ | | |Dynamic | | ALTO | ALTO Protocol | ALTO | | | |Network |.......| Server | ==================== | Client | | | |Information| +---------+ +--------+ | | +-----------+ / / | | / ALTO SD Query/Response / | | / / | | +----------+ +----------------+ | | | External | | ALTO Service | | | | Interface| | Discovery (SD) | | | +----------+ +----------------+ | | | | +-------------------------------------------------------------------+ | +------------------+ | Third Parties | | | | Content Providers| +------------------+ Figure 1: Basic ALTO Architecture Figure 1 illustrates that the ALTO information provided by an ALTO server may be influenced (at the service provider's discretion) by other systems. In particular, the ALTO server can aggregate information from multiple systems to provide an abstract and unified view that can be more useful to applications. Examples of other systems include (but are not limited to) static network configuration databases, dynamic network information, routing protocols, provisioning policies, and interfaces to outside parties. These components are shown in the figure for completeness but are outside the scope of this specification. Recall that while the ALTO Protocol may convey dynamic network information, it is not intended to replace near-real-time congestion control protocols.
It may also be possible for an ALTO server to exchange network information with other ALTO servers (either within the same administrative domain or another administrative domain with the consent of both parties) in order to adjust exported ALTO information. Such a protocol is also outside the scope of this specification.3.2. ALTO Information Reuse and Redistribution
ALTO information may be useful to a large number of applications and users. At the same time, distributing ALTO information must be efficient and not become a bottleneck. The design of the ALTO Protocol allows integration with the existing HTTP caching infrastructure to redistribute ALTO information. If caching or redistribution is used, the response message to an ALTO client may be returned from a third party. Application-dependent mechanisms, such as P2P Distributed Hash Tables (DHTs) or P2P file sharing, may be used to cache and redistribute ALTO information. This document does not define particular mechanisms for such redistribution. Additional protocol mechanisms (e.g., expiration times and digital signatures for returned ALTO information) are left for future investigation.4. ALTO Information Service Framework
The ALTO Protocol conveys network information through ALTO information services (services for short), where each service defines a set of related functionalities. An ALTO client can request each service individually. All of the services defined in ALTO are said to form the ALTO service framework and are provided through a common transport protocol; messaging structure and encoding; and transaction model. Functionalities offered in different services can overlap. The goals of the ALTO information services defined in this document are to convey (1) network locations, which denote the locations of endpoints at a network, (2) provider-defined costs for paths between pairs of network locations, and (3) network-related properties of endpoints. The aforementioned goals are achieved by defining the Map Service, which provides the core ALTO information to clients, and three additional information services: the Map-Filtering Service, the Endpoint Property Service (EPS), and the Endpoint Cost Service (ECS). Additional information services can be defined in companion documents. Figure 2 gives an overview of the information services. Details of the services are presented in subsequent sections.
.-----------------------------------------. | ALTO Information Services | | .-----------. .----------. .----------. | | | Map- | | Endpoint | | Endpoint | | | | Filtering | | Property | | Cost | | | | Service | | Service | | Service | | | `-----------' `----------' `----------' | | .-------------------------------------. | | | Map Service | | | | .-------------. .--------------. | | | | | Network Map | | Cost Map | | | | | `-------------' `--------------' | | | `-------------------------------------' | `-----------------------------------------' Figure 2: ALTO Information Service Framework4.1. ALTO Information Services
4.1.1. Map Service
The Map Service provides batch information to ALTO clients in the forms of ALTO network maps (network maps for short) and ALTO cost maps (cost maps for short). An ALTO network map (See Section 5) provides a full set of network location groupings defined by the ALTO server and the endpoints contained within each grouping. An ALTO cost map (see Section 6) provides costs between defined groupings. These two maps can be thought of (and implemented) as simple files with appropriate encoding provided by the ALTO server.4.1.2. Map-Filtering Service
Resource-constrained ALTO clients may benefit from the filtering of query results at the ALTO server. This avoids the situation in which an ALTO client first spends network bandwidth and CPU cycles to collect results and then performs client-side filtering. The Map- Filtering Service allows ALTO clients to query an ALTO server on ALTO network maps and/or cost maps based on additional parameters.4.1.3. Endpoint Property Service
This service allows ALTO clients to look up properties for individual endpoints. An example property of an endpoint is its network location (i.e., its grouping defined by the ALTO server). Another example property is its connectivity type such as ADSL (Asymmetric Digital Subscriber Line), Cable, or FTTH (Fiber To The Home).
4.1.4. Endpoint Cost Service
Some ALTO clients may also benefit from querying for costs and rankings based on endpoints. The Endpoint Cost Service allows an ALTO server to return costs directly amongst endpoints.5. Network Map
An ALTO network map defines a grouping of network endpoints. This document uses ALTO network map to refer to the syntax and semantics of how an ALTO server defines the grouping. This document does not discuss the internal representation of this data structure within an ALTO server. The definition of ALTO network maps is based on the observation that, in reality, many endpoints are near by to one another in terms of network connectivity. By treating a group of nearby endpoints together as a single entity, an ALTO server indicates aggregation of these endpoints due to their proximity. This aggregation can also lead to greater scalability without losing critical information when conveying other network information (e.g., when defining cost maps).5.1. Provider-Defined Identifier (PID)
One issue is that proximity varies depending on the granularity of the ALTO information configured by the provider. In one deployment, endpoints on the same subnet may be considered close; while in another deployment, endpoints connected to the same Point of Presence (POP) may be considered close. ALTO introduces provider-defined network location identifiers called Provider-defined Identifiers (PIDs) to provide an indirect and network-agnostic way to specify an aggregation of network endpoints that may be treated similarly, based on network topology, type, or other properties. Specifically, a PID is a string of type PIDName (see Section 10.1) and its associated set of endpoint addresses. As discussed above, there can be many different ways of grouping the endpoints and assigning PIDs. For example, a PID may denote a subnet, a set of subnets, a metropolitan area, a POP, an autonomous system, or a set of autonomous systems. Interpreting the PIDs defined in an ALTO network map using the "single-node" abstraction, one can consider that each PID represents an abstract port (POP) that connects a set of endpoints. A key use case of PIDs is to specify network preferences (costs) between PIDs instead of individual endpoints. This allows cost information to be more compactly represented and updated at a faster time scale than the network aggregations themselves. For example, an
ISP may prefer that endpoints associated with the same POP in a P2P application communicate locally instead of communicating with endpoints in other POPs. The ISP may aggregate endpoints within a POP into a single PID in a network map. The cost may be encoded to indicate that network locations within the same PID are preferred; for example, cost(PID_i, PID_i) == c and cost(PID_i, PID_j) > c for i != j. Section 6 provides further details on using PIDs to represent costs in an ALTO cost map.5.2. Endpoint Addresses
The endpoints aggregated into a PID are denoted by endpoint addresses. There are many types of addresses, such as IP addresses, MAC addresses, or overlay IDs. This document specifies (in Section 10.4) how to specify IPv4/IPv6 addresses or prefixes. Extension documents may define further address types; Section 14.4 of this document provides an IANA registry for endpoint address types.5.3. Example Network Map
This document uses the ALTO network map shown in Figure 3 in most examples.
.------------------------------------------------------------. | An ALTO Network Map | | | | .-----------------------------------. .----------------. | | | NetLoc: PID-1 | | NetLoc: PID-3 | | | | .------------------------------. | | | | | | | 192.0.2.0/24 | | | .-----------. | | | | | .--------------------------. | | | | 0.0.0.0/0 | | | | | | | Endpoint: 192.0.2.34 | | | | `-----------` | | | | | `--------------------------` | | | | | | | `------------------------------` | | | | | | .------------------------------. | | | | | | | 198.51.100.0/25 | | | | | | | | .--------------------------. | | | | | | | | | Endpoint: 198.51.100.100 | | | | | | | | | `--------------------------` | | | | | | | `------------------------------` | | | | | `-----------------------------------` | | | | | | | | .-----------------------------------. | | | | | NetLoc: PID-2 | | | | | | .------------------------------. | | | | | | | 198.51.100.128/25 | | | | | | | `------------------------------` | | | | | `-----------------------------------` `----------------` | `------------------------------------------------------------` Figure 3: Example Network Map6. Cost Map
An ALTO server indicates preferences amongst network locations in the form of path costs. Path costs are generic costs and can be internally computed by a network provider according to its own policy. For a given ALTO network map, an ALTO cost map defines path costs pairwise amongst the set of source and destination network locations defined by the PIDs contained in the network map. Each path cost is the end-to-end cost when a unit of traffic goes from the source to the destination. Since cost is directional from the source to the destination, an application, when using ALTO information, may independently determine how the resource consumer and resource provider are designated as the source or destination in an ALTO query and, hence, how to utilize the path cost provided by ALTO information. For example, if the cost is
expected to be correlated with throughput, a typical application concerned with bulk data retrieval may use the resource provider as the source and the resource consumer as the destination. One advantage of separating ALTO information into network maps and cost maps is that the two types of maps can be updated at different time scales. For example, network maps may be stable for a longer time while cost maps may be updated to reflect more dynamic network conditions. As used in this document, an ALTO cost map refers to the syntax and semantics of the information distributed by the ALTO server. This document does not discuss the internal representation of this data structure within the ALTO server.6.1. Cost Types
Path costs have attributes: o Cost Metric: identifies what the costs represent; o Cost Mode: identifies how the costs should be interpreted. The combination of a cost metric and a cost mode defines an ALTO cost type. Certain queries for ALTO cost maps allow the ALTO client to indicate the desired cost type. For a given ALTO server, the combination of cost type and network map defines a key. In other words, an ALTO server MUST NOT define two ALTO cost maps with the same cost type \ network map pair.6.1.1. Cost Metric
The cost metric attribute indicates what the cost represents. For example, an ALTO server could define costs representing air miles, hop-counts, or generic routing costs. Cost metrics are indicated in protocol messages as strings.6.1.1.1. Cost Metric: routingcost
An ALTO server MUST offer the "routingcost" cost metric. This cost metric conveys a generic measure for the cost of routing traffic from a source to a destination. A lower value indicates a higher preference for traffic to be sent from a source to a destination.
Note that an ISP may internally compute routing cost using any method that it chooses (e.g., air miles or hop-count) as long as it conforms to the semantics.6.1.2. Cost Mode
The cost mode attribute indicates how costs should be interpreted. Specifically, the cost mode attribute indicates whether returned costs should be interpreted as numerical values or ordinal rankings. It is important to communicate such information to ALTO clients, as certain operations may not be valid on certain costs returned by an ALTO server. For example, it is possible for an ALTO server to return a set of IP addresses with costs indicating a ranking of the IP addresses. Arithmetic operations that would make sense for numerical values, do not make sense for ordinal rankings. ALTO clients may handle such costs differently. Cost modes are indicated in protocol messages as strings. An ALTO server MUST support at least one of the following modes: numerical and ordinal. An ALTO client needs to be cognizant of operations when its desired cost mode is not supported. Specifically, an ALTO client desiring numerical costs MAY adjust its behaviors if only the ordinal cost mode is available. Alternatively, an ALTO client desiring ordinal costs MAY construct ordinal costs from retrieved numerical values, if only the numerical cost mode is available.6.1.2.1. Cost Mode: numerical
This cost mode is indicated by the string "numerical". This mode indicates that it is safe to perform numerical operations (e.g., normalization or computing ratios for weighted load-balancing) on the returned costs. The values are floating-point numbers.6.1.2.2. Cost Mode: ordinal
This cost mode is indicated by the string "ordinal". This mode indicates that the cost values in a cost map represent ranking (relative to all other values in a cost map), not actual costs. The values are non-negative integers, with a lower value indicating a higher preference. Ordinal cost values in a cost map need not be unique or contiguous. In particular, it is possible that two entries in a cost map have an identical rank (ordinal cost value). This document does not specify any behavior by an ALTO client in this case; an ALTO client may decide to break ties by random selection, other application knowledge, or some other means.
6.2. Cost Map Structure
A request for an ALTO cost map will either explicitly or implicitly include a list of source network locations and a list of destination network locations. (Recall that a network location can be an endpoint address or a PID.) Specifically, assume that a request specifies a list of source network locations, say [Src_1, Src_2, ..., Src_m], and a list of destination network locations, say [Dst_1, Dst_2, ..., Dst_n]. The ALTO server will return the path cost for each of the m*n communicating pairs (i.e., Src_1 -> Dst_1, ..., Src_1 -> Dst_n, ..., Src_m -> Dst_1, ..., Src_m -> Dst_n). If the ALTO server does not define the path cost for a particular pair, that cost may be omitted. This document refers to this structure as a cost map. If the cost mode is ordinal, the path cost of each communicating pair is relative to the m*n entries.6.3. Network Map and Cost Map Dependency
An ALTO cost map gives path costs between the PIDs defined in an ALTO network map. An ALTO server may modify an ALTO network map at any time, say by adding or deleting PIDs, or even redefining them. Hence, to effectively use an instance of an ALTO cost map, an ALTO client must know which version of the network map defined the PIDs in that cost map. Version tags allow an ALTO client to correlate cost map instances with the corresponding versions of the network maps. Specifically, a version tag is a tuple of (1) an ID for the resource (e.g., an ALTO network map) and (2) a tag (an opaque string) associated with the version of that resource. An ALTO network map distributed by an ALTO server includes its version tag. An ALTO cost map referring to PIDs also includes the version tag for the network map on which it is based. Two ALTO network maps are the same if they have the same version tag. Whenever the content of an ALTO network map maintained by an ALTO server changes, the tag MUST also be changed. Possibilities of setting the tag component include the last-modified timestamp for the network map, or a hash of its contents, where the collision probability is considered zero in practical deployment scenarios.
6.4. Cost Map Update
An ALTO server can update an ALTO cost map at any time. Hence, the same cost map retrieved from the same ALTO server but from different requests can be inconsistent.7. Endpoint Properties
An endpoint property defines a network-aware property of an endpoint.7.1. Endpoint Property Type
For each endpoint and an endpoint property type, there can be a value for the property. The type of an endpoint property is indicated in protocol messages as a string. The value depends on the specific property. For example, for a property such as whether an endpoint is metered, the value is a true or false value. See Section 10.8 for more details on specifying endpoint properties.7.1.1. Endpoint Property Type: pid
An ALTO server MUST define the "pid" endpoint property type for each ALTO network map that it provides. Specifically, each ALTO network map defines multiple PIDs. For an "ipv4"/"ipv6" network map, given an endpoint's IP address, the ALTO server uses the algorithm specified in Section 11.2.2 to look up the PID of the endpoint. This PID is the "pid" property of the endpoint for the network map. See Section 11.4.1.7 for an example.
(page 19 continued on part 2)
