RFC 9344 | CCNinfo | February 2023 |
Asaeda, et al. | Experimental | [Page] |
This document describes a mechanism named "CCNinfo" that discovers information about the network topology and in-network cache in Content-Centric Networks (CCNs). CCNinfo investigates 1) the CCN routing path information per name prefix, 2) the Round-Trip Time (RTT) between the content forwarder and the consumer, and 3) the states of in-network cache per name prefix. CCNinfo is useful to understand and debug the behavior of testbed networks and other experimental deployments of CCN systems.¶
This document is a product of the IRTF Information-Centric Networking Research Group (ICNRG). This document represents the consensus view of ICNRG and has been reviewed extensively by several members of the ICN community and the RG. The authors and RG chairs approve of the contents. The document is sponsored under the IRTF, is not issued by the IETF, and is not an IETF standard. This is an experimental protocol and the specification may change in the future.¶
This document is not an Internet Standards Track specification; it is published for examination, experimental implementation, and evaluation.¶
This document defines an Experimental Protocol for the Internet community. This document is a product of the Internet Research Task Force (IRTF). The IRTF publishes the results of Internet-related research and development activities. These results might not be suitable for deployment. This RFC represents the consensus of the Information-Centric Networking Research Group of the Internet Research Task Force (IRTF). Documents approved for publication by the IRSG are not candidates for any level of Internet Standard; see Section 2 of RFC 7841.¶
Information about the current status of this document, any errata, and how to provide feedback on it may be obtained at https://www.rfc-editor.org/info/rfc9344.¶
Copyright (c) 2023 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 (https://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.¶
In Content-Centric Networks (CCNs), publishers provide the content through the network, and receivers retrieve it by name. In this network architecture, routers forward content requests through their Forwarding Information Bases (FIBs), which are populated by name-based routing protocols. CCN also enables receivers to retrieve content from an in-network cache.¶
In CCN, while consumers do not generally need to know the content forwarder that is transmitting the content to them, the operators and developers may want to identify the content forwarder and observe the routing path information per name prefix for troubleshooting or investigating the network conditions.¶
IP traceroute is a useful tool for discovering the routing conditions in IP networks because it provides intermediate router addresses along the path between the source and the destination, and the Round-Trip Time (RTT) for the path. However, this IP-based network tool cannot trace the name prefix paths used in CCN. Moreover, such IP-based network tools do not obtain the states of the in-network cache to be discovered.¶
Contrace [Contrace] enables end users (i.e., consumers) to investigate path and in-network cache conditions in CCN. Contrace is implemented as an external daemon process running over TCP/IP that can interact with a previous CCNx forwarding daemon (CCNx-0.8.2) to retrieve the caching information on the forwarding daemon. This solution is flexible, but it requires defining the common APIs used for global deployment in TCP/IP networks. ICN (Information-Centric Networking) ping [ICN-PING] and traceroute [ICN-TRACEROUTE] are lightweight operational tools that enable a user to explore the path(s) that reach a publisher or a cache storing the named content. ICN ping and traceroute, however, do not expose detailed information about the forwarders deployed by a network operator.¶
This document describes the specifications of "CCNinfo", an active networking tool for discovering the path and content-caching information in CCN. CCNinfo defines the protocol messages to investigate path and in-network cache conditions in CCN. It is embedded in the CCNx forwarding process and can facilitate with non-IP networks as with the basic CCN concept.¶
The two message types, Request and Reply messages, are encoded in the CCNx TLV format [RFC8609]. The Request-and-Reply message flow, walking up the tree from a consumer toward a publisher, is similar to the behavior of the IP multicast traceroute facility [RFC8487].¶
CCNinfo facilitates the tracing of a routing path and provides 1) the RTT between the content forwarder (i.e., caching router or first-hop router) and consumer, 2) the states of the in-network cache per name prefix, and 3) the routing path information per name prefix.¶
In addition, CCNinfo identifies the states of the cache, such as the metrics for Content Store (CS) in the content forwarder as follows: 1) size of cached Content Objects, 2) number of cached Content Objects, 3) number of accesses (i.e., received Interests) per content, and 4) elapsed cache time and remaining cache lifetime of content.¶
CCNinfo supports multipath forwarding. The Request messages can be forwarded to multiple neighbor routers. When the Request messages are forwarded to multiple routers, the different Reply messages are forwarded from different routers or publishers.¶
Furthermore, CCNinfo implements policy-based information provisioning that enables administrators to "hide" secure or private information but does not disrupt message forwarding. This policy-based information provisioning reduces the deployment barrier faced by operators in installing and running CCNinfo on their routers.¶
The document represents the consensus of the Information-Centric Networking Research Group (ICNRG). This document was read and reviewed by the active research group members. It is not an IETF product and is not a standard.¶
In order to carry out meaningful experimentation with CCNx protocols, comprehensive instrumentation and management information is needed to take measurements and explore both the performance and robustness characteristics of the protocols and of the applications using them. CCNinfo's primary goal is to gather and report this information. As experience is gained with both the CCNx protocols and CCNinfo itself, we can refine the instrumentation capabilities and discover what additional capabilities might be needed in CCNinfo and conversely which features wind up not being of sufficient value to justify the implementation complexity and execution overhead.¶
CCNinfo is intended as a comprehensive experimental tool for CCNx-based networks. It provides a wealth of information from forwarders, including on-path in-network cache conditions as well as forwarding path instrumentation of multiple paths toward content forwarders. As an experimental capability that exposes detailed information about the forwarders deployed by a network operator, CCNinfo employs more granular authorization policies than those required of ICN ping or ICN traceroute.¶
CCNinfo uses two message types: Request and Reply. A CCNinfo user, e.g., consumer, initiates a CCNinfo Request message when they want to obtain routing path and cache information. When an adjacent neighbor router receives the Request message, it examines its own cache information. If the router does not cache the specified content, it inserts its Report block into the hop-by-hop header of the Request message and forwards the message to its upstream neighbor router(s) decided by its FIB. In Figure 1, CCNinfo user and routers (Routers A, B, C) insert their own Report blocks into the Request message and forward the message toward the content forwarder.¶
When the Request message reaches the content forwarder, the content forwarder forms the Reply message; it inserts its own Reply block TLV and Reply sub-block TLV(s) to the Request message. The Reply message is then forwarded back toward the user in a hop-by-hop manner along the Pending Interest Table (PIT) entries. In Figure 2, each router (Routers C, B, and A) forwards the Reply message along its PIT entry, and finally, the CCNinfo user receives a Reply message from Router C, which is the first-hop router for the publisher. Another Reply message from the caching router (i.e., Reply(C)) is discarded at Router B if the other Reply message (i.e., Reply(P)) was already forwarded by Router B.¶
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 [RFC2119] [RFC8174] when, and only when, they appear in all capitals, as shown here.¶
This document follows the basic terminologies and definitions described in [RFC8609]. Although CCNinfo requests flow in the opposite direction to the data flow, we refer to "upstream" and "downstream" with respect to data, unless explicitly specified.¶
CCNinfo Request and Reply messages are encoded in the CCNx TLV format (see [RFC8609] and Figure 3). The Request message consists of a fixed header, Request block TLV (Figure 5), and Report block TLV(s) (Figure 7). The Reply message consists of a fixed header, Request block TLV, Report block TLV(s), Reply block TLV (Figure 9), and Reply sub-block TLV(s) (Figure 10).¶
The PacketType values in the fixed header shown in Figure 3 are PT_CCNINFO_REQUEST and PT_CCNINFO_REPLY (see Table 1). CCNinfo Request and Reply messages are forwarded in a hop-by-hop manner. When the Request message reaches the content forwarder, the content forwarder turns it into a Reply message by changing the Type field value in the fixed header from PT_CCNINFO_REQUEST to PT_CCNINFO_REPLY and forwards it back toward the node that initiated the Request message.¶
Type | Name |
---|---|
0x00 | PT_INTEREST [RFC8609] |
0x01 | PT_CONTENT [RFC8609] |
0x02 | PT_RETURN [RFC8609] |
0x03 | PT_CCNINFO_REQUEST |
0x04 | PT_CCNINFO_REPLY |
Following a fixed header, there can be a sequence of optional hop-by-hop header TLV(s) for a Request message. In the case of a Request message, it is followed by a sequence of Report blocks, each from a router on the path toward the publisher or caching router.¶
At the beginning of PacketPayload TLVs, a top-level TLV type, T_DISCOVERY (Table 2), exists at the outermost level of a CCNx protocol message. This TLV indicates that the Name segment TLV(s) and Reply block TLV(s) would follow in the Request or Reply message.¶
Type | Name |
---|---|
0x0000 | Reserved [RFC8609] |
0x0001 | T_INTEREST [RFC8609] |
0x0002 | T_OBJECT [RFC8609] |
0x0003 | T_VALIDATION_ALG [RFC8609] |
0x0004 | T_VALIDATION_PAYLOAD [RFC8609] |
0x0005 | T_DISCOVERY |
When a CCNinfo user initiates a discovery request (e.g., via the ccninfo command described in Appendix A), a CCNinfo Request message is created and forwarded to its upstream router through the Incoming face(s) determined by its FIB.¶
The Request message format is shown in Figure 4. It consists of a fixed header, Request header block TLV (Figure 5), Report block TLV(s) (Figure 7), Name TLV, and Request block TLV. Request header block TLV and Report block TLV(s) are contained in the hop-by-hop header, as those might change from hop to hop. Request block TLV is encoded in the PacketPayload TLV by content forwarder as the protocol message itself. The PacketType value of the Request message is PT_CCNINFO_REQUEST (Table 1). The Type value of the CCNx Top-Level type is T_DISCOVERY (Table 2).¶
8 bits¶
HopLimit is a counter that is decremented with each hop whenever a Request packet is forwarded. It is specified by the CCNinfo user program. The HopLimit value MUST be decremented by 1 prior to forwarding the Request packet. The packet is discarded if HopLimit is decremented to zero. HopLimit limits the distance that a Request may travel on the network. Only the specified number of hops from the CCNinfo user traces the Request. The last router stops the trace and sends the Reply message back to the CCNinfo user.¶
8 bits¶
ReturnCode is used for the Reply message. This value is replaced by the content forwarder when the Request message is returned as the Reply message (see Section 3.2). Until then, this field MUST be transmitted as zeros and ignored on receipt.¶
Value | Name | Description |
---|---|---|
0x00 | NO_ERROR | No error |
0x01 | WRONG_IF | CCNinfo Request arrived on an interface to which this router would not forward for the specified name and/or function toward the publisher. |
0x02 | INVALID_REQUEST | Invalid CCNinfo Request is received. |
0x03 | NO_ROUTE | This router has no route for the name prefix and no way to determine a route. |
0x04 | NO_INFO | This router has no cache information for the specified name prefix. |
0x05 | NO_SPACE | There was not enough room to insert another Report block in the packet. |
0x06 | INFO_HIDDEN | Information is hidden from this discovery owing to some policy. |
0x0E | ADMIN_PROHIB | CCNinfo Request is administratively prohibited. |
0x0F | UNKNOWN_REQUEST | This router does not support or recognize the Request message. |
0x80 | FATAL_ERROR | In a fatal error, the router may know the upstream router but cannot forward the message to it. |
8 bits¶
The reserved fields in the Value field MUST be transmitted as zeros and ignored on receipt.¶
When a CCNinfo user transmits the Request message, they MUST insert their Request header block TLV (Figure 5) into the hop-by-hop header and Request block TLV (Figure 6) into the message before sending it through the Incoming face(s).¶
Type | Name |
---|---|
0x0000 | Reserved [RFC8609] |
0x0001 | T_INTLIFE [RFC8609] |
0x0002 | T_CACHETIME [RFC8609] |
0x0003 | T_MSGHASH [RFC8609] |
0x0004-0x0007 | Reserved [RFC8609] |
0x0008 | T_DISC_REQHDR |
0x0009 | T_DISC_REPORT |
0x0FFE | T_PAD [RFC8609] |
0x0FFF | T_ORG [RFC8609] |
0x1000-0x1FFF | Reserved [RFC8609] |
16 bits¶
Format of the Value field. The type value of the Request header block TLV MUST be T_DISC_REQHDR.¶
16 bits¶
Length of the Value field in octets.¶
16 bits¶
This field is used as a unique identifier for the CCNinfo Request so that the duplicate or delayed Reply messages can be detected.¶
4 bits¶
Number of skipped routers for a Request. It is specified by the CCNinfo user program. The number of routers corresponding to the value specified in this field are skipped, and the CCNinfo Request messages are forwarded to the next router without the addition of Report blocks; the next upstream router then starts the trace. The maximum value of this parameter is 15. This value MUST be lower than that of HopLimit at the fixed header.¶
12 bits¶
The Flags field is used to indicate the types of the content or path discoveries. Currently, as shown in Table 5, four bits ("C", "O", "F", and "V") are assigned, and the other 8 bits are reserved (MBZ) for the future use. Each flag can be mutually specified with other flags. These flags are set by the CCNinfo user program when they initiate Requests (see Appendix A), and the routers that receive the Requests deal with the flags and change the behaviors (see Section 5 for details). The Flag values defined in this Flags field correspond to the Reply sub-blocks.¶
Flag | Value | Description |
---|---|---|
C | 0 | Path discovery (i.e., no cache information retrieved) (default) |
C | 1 | Path and cache information retrieval |
O | 0 | Request to any content forwarder (default) |
O | 1 | Publisher discovery (i.e., only FHR can reply) |
F | 0 | Request based on FIB's forwarding strategy (default) |
F | 1 | Full discovery request. Request to possible multiple upstream routers specified in FIB simultaneously |
V | 0 | No reply validation (default) |
V | 1 | Reply sender validates Reply message |
Type | Name |
---|---|
0x0000 | T_NAME [RFC8609] |
0x0001 | T_PAYLOAD [RFC8609] |
0x0002 | T_KEYIDRESTR [RFC8609] |
0x0003 | T_OBJHASHRESTR [RFC8609] |
0x0005 | T_PAYLDTYPE [RFC8609] |
0x0006 | T_EXPIRY [RFC8609] |
0x0007-0x000C | Reserved [RFC8609] |
0x000D | T_DISC_REQ |
0x000E | T_DISC_REPLY |
0x0FFE | T_PAD [RFC8609] |
0x0FFF | T_ORG [RFC8609] |
0x1000-0x1FFF | Reserved [RFC8609] |
16 bits¶
Format of the Value field. For the Request block TLV, the type value(s) MUST be T_DISC_REQ (see Table 6) in the current specification.¶
16 bits¶
Length of the Value field in octets.¶
32 bits¶
The Request Arrival Time is a 32-bit NTP timestamp specifying the arrival time of the CCNinfo Request message at the router. The 32-bit form of an NTP timestamp consists of the middle 32 bits of the full 64-bit form, that is, the low 16 bits of the integer part and the high 16 bits of the fractional part.¶
The following formula converts from a timespec (fractional part in nanoseconds) to a 32-bit NTP timestamp:¶
request_arrival_time = ((tv.tv_sec + 32384) << 16) + ((tv.tv_nsec << 7) / 1953125)¶
The constant 32384 is the number of seconds from Jan 1, 1900 to Jan 1, 1970 truncated to 16 bits. ((tv.tv_nsec << 7) / 1953125) is a reduction of ((tv.tv_nsec / 1000000000) << 16), where "<<" denotes a logical left shift.¶
Note that it is RECOMMENDED for all the routers on the path to have synchronized clocks to measure one-way latency per hop; however, even if they do not have synchronized clocks, CCNinfo measures the RTT between the content forwarder and the consumer.¶
variable length¶
This field specifies the node identifier (e.g., node name or hash-based self-certifying name [DCAuth]) or all-zeros if unknown. This document assumes that the Name TLV defined in the CCNx TLV format [RFC8609] can be used for this field and the node identifier is specified in it.¶
A CCNinfo user and each upstream router along the path would insert their own Report block TLV without changing the Type field of the fixed header of the Request message until one of these routers is ready to send a Reply. In the Report block TLV (Figure 7), the Request Arrival Time and Node Identifier values MUST be inserted.¶
16 bits¶
Format of the Value field. For the Report block TLV, the type value(s) MUST be T_DISC_REPORT in the current specification. For all the available types of the CCNx hop-by-hop types, please see Table 4.¶
16 bits¶
Length of the Value field in octets.¶
32 bits¶
Same definition as given in Section 3.1.1.¶
variable length¶
Same definition as given in Section 3.1.1.¶
Specifications of the Name TLV (whose type value is T_NAME) and the Name Segment TLVs are described in [RFC8609], which is followed by CCNinfo. CCNinfo enables the specification of the content name with either a prefix name without chunk number (such as "ccnx:/news/today") or an exact name (such as "ccnx:/news/today/Chunk=10"). When a CCNinfo user specifies a prefix name, they will obtain the summary information of the matched Content Objects in the content forwarder. In contrast, when a CCNinfo user specifies an exact name, they will obtain information only about the specified Content Object in the content forwarder. A CCNinfo Request message MUST NOT be sent only with a scheme name, ccnx:/. It will be rejected and discarded by routers.¶
When a content forwarder receives a CCNinfo Request message from an appropriate adjacent neighbor router, it inserts its own Reply block TLV and Reply sub-block TLV(s) to the Request message and turns the Request into the Reply by changing the Type field of the fixed header of the Request message from PT_CCNINFO_REQUEST to PT_CCNINFO_REPLY. The Reply message (see Figure 8) is then forwarded back toward the CCNinfo user in a hop-by-hop manner.¶
The Reply block TLV is an envelope for the Reply sub-block TLV(s) (explained in the next section).¶
16 bits¶
Format of the Value field. For the Reply block TLV, the type value MUST be T_DISC_REPLY shown in Table 6 in the current specification.¶
16 bits¶
Length of the Value field in octets. This length is the total length of the Reply sub-block(s).¶
32 bits¶
Same definition as given in Section 3.1.1.¶
variable length¶
Same definition as given in Section 3.1.1.¶
The router on the traced path will add one or multiple Reply sub-blocks followed by the Reply block TLV before sending the Reply to its neighbor router. This section describes the Reply sub-block TLV for informing various cache states and conditions as shown in Figure 10. (Other Reply sub-block TLVs will be discussed in separate document(s).)¶
Note that some routers may not be capable of reporting the following values: Object Size, Object Count, # Received Interest, First Seqnum, Last Seqnum, Elapsed Cache Time, and Remain Cache Lifetime (shown in Figure 10). Or, some routers do not report these values due to their policy. In that case, the routers MUST set these fields to a value of all ones (i.e., 0xFFFFFFFF). The value of each field MUST be also all-one when the value is equal to or bigger than the maximum size expressed by the 32-bit field. The CCNinfo user program MUST inform that these values are not valid if the fields received are set to the value of all ones.¶
If the cache is refreshed after reboot, the value in each field MUST be refreshed (i.e., MUST be set to 0). If the cache remains after reboot, the value MUST NOT be refreshed (i.e., MUST be reflected as it is).¶
Type | Name |
---|---|
0x0000 | T_DISC_CONTENT |
0x0001 | T_DISC_CONTENT_PUBLISHER |
0x0FFF | T_ORG |
0x1000-0x1FFF | Reserved for Experimental Use |
16 bits¶
Format of the Value field. For the Reply sub-block TLV, the type value MUST be either T_DISC_CONTENT or T_DISC_CONTENT_PUBLISHER defined in the CCNx Reply Types (Table 7). T_DISC_CONTENT is specified when a content forwarder replies with the cache information. T_DISC_CONTENT_PUBLISHER is specified when a FHR attached to a publisher replies with the original content information.¶
16 bits¶
Length of the Value field in octets.¶
32 bits¶
The total size (KB) of the unexpired Content Objects. Values less than 1 KB are truncated. Note that the maximum size expressed by the 32-bit field is approximately 4.29 TB.¶
32 bits¶
The number of the unexpired Content Objects. Note that the maximum count expressed by the 32-bit field is approximately 4.29 billion.¶
32 bits¶
The total number of the received Interest messages to retrieve the cached Content Objects.¶
32 bits¶
The first sequential number of the unexpired Content Objects.¶
32 bits¶
The last sequential number of the unexpired Content Objects. The First Seqnum and Last Seqnum do not guarantee the consecutiveness of the cached Content Objects; however, knowing these values may help in the analysis of consecutive or discontinuous chunks such as [CONSEC-CACHING].¶
32 bits¶
The elapsed time (seconds) after the oldest Content Object of the content is cached.¶
32 bits¶
The lifetime (seconds) of a Content Object, which is lastly cached.¶
A CCNinfo user invokes a CCNinfo user program (e.g., ccninfo command) that initiates a CCNinfo Request message and sends it to the user's adjacent neighbor router(s) of interest. The user later obtains both the routing path information and in-network cache information in the single Reply.¶
When the CCNinfo user program initiates a Request message, it MUST insert the necessary values, i.e., the "Request ID" and the "Node Identifier", in the Request block. The Request ID MUST be unique for the CCNinfo user until they receive the corresponding Reply message(s) or the Request is timed out.¶
Owing to some policies, a router may want to validate the CCNinfo Requests using the CCNx ValidationPayload TLV (whether it accepts the Request or not) especially when the router receives the "full discovery request" (see Section 5.3.2). Accordingly, the CCNinfo user program MAY require validating the Request message and appending the user's signature into the CCNx ValidationPayload TLV. The router then forwards the Request message. If the router does not approve the Request, it rejects the Request message as described in Section 6.11.¶
After the CCNinfo user program sends the Request message, until the Reply is timed out or the expected numbers of Replies or a Reply message with a non-zero ReturnCode in the fixed header is received, the CCNinfo user program MUST keep the following information: HopLimit (specified in the fixed header), Request ID and Flags (specified in the Request header block), and Node Identifier and Request Arrival Time (specified in the Request block).¶
A CCNinfo user can send a CCNinfo Request for investigating the routing path information for the specified named content. Using the Request, a legitimate user can obtain 1) the node identifiers of the intermediate routers, 2) the node identifier of the content forwarder, 3) the number of hops between the content forwarder and the consumer, and 4) the RTT between the content forwarder and the consumer, per name prefix. This CCNinfo Request is terminated when it reaches the content forwarder.¶
A CCNinfo user can send a CCNinfo Request for investigating in-network cache information. Using the Request, a legitimate user can obtain 1) the size of cached Content Objects, 2) the number of cached Content Objects, 3) the number of accesses (i.e., received Interests) per content, and 4) the lifetime and expiration time of the cached Content Objects, for Content Store (CS) in the content forwarder, unless the content forwarder is capable of reporting them (see Section 3.2.1.1). This CCNinfo Request is terminated when it reaches the content forwarder.¶
A CCNinfo user program will receive one or multiple CCNinfo Reply messages from the adjacent neighbor router(s). When the program receives the Reply, it MUST compare the kept Request ID and Node Identifier values to identify the Request and Reply pair. If they do not match, the Reply message MUST be silently discarded.¶
If the number of Report blocks in the received Reply is more than the initial HopLimit value (which was inserted in the original Request), the Reply MUST be silently ignored.¶
After the CCNinfo user has determined that they have traced the whole path or the maximum path that they can be expected to, they might collect statistics by waiting for a timeout. Useful statistics provided by CCNinfo are stated in Section 8.¶
Upon receiving a CCNinfo Request message, a router MAY examine whether a valid CCNinfo user has sent the message. If the router recognizes that the Request sender's signature specified in the Request is invalid, it SHOULD terminate the Request, as defined in Section 6.4.¶
Upon receiving a CCNinfo Request or Reply message, a router MAY examine whether the message comes from a valid adjacent neighbor node. If the router recognizes that the Request or Reply sender is invalid, it SHOULD silently ignore the message, as specified in Section 10.9.¶
After a router accepts the CCNinfo Request message, it performs the following steps.¶
When a router decides to forward a Request message with its Report block to its upstream router(s), it specifies the Request Arrival Time and Node Identifier values in the Report block of the Request message. The router then forwards the Request message upstream toward the publisher or caching router based on the FIB entry like the ordinary Interest-Data exchanges in CCN.¶
When the router forwards the Request message, it MUST record the F flag and Request ID in the Request block of the Request message and exploiting path labels (specified in Section 1) at the corresponding PIT entry. The router can later check the PIT entry to correctly forward the Reply message(s) back.¶
CCNinfo supports multipath forwarding. The Request messages can be forwarded to multiple neighbor routers. Some routers may have a strategy for multipath forwarding; when a router sends Interest messages to multiple neighbor routers, it may delay or prioritize to send the message to the upstream routers. The CCNinfo Request, as the default, complies with such strategies; a CCNinfo user could trace the actual forwarding path based on the forwarding strategy and will receive a single Reply message such as a Content Object.¶
There may be a case wherein a CCNinfo user wants to discover all possible forwarding paths and content forwarders based on the routers' FIBs. The "full discovery request" enables this functionality. If a CCNinfo user sets the F flag in the Request block of the Request message (as seen in Table 5) to request the full discovery, the upstream routers simultaneously forward the Requests to all multiple upstream routers based on the FIBs. Then, the CCNinfo user can trace all possible forwarding paths. As seen in Figure 11, each router forwards the Reply message along its PIT entry, and finally, the CCNinfo user receives two Reply messages: one from the FHR (Router C) and the other from the Caching router.¶
To receive different Reply messages forwarded from different routers, the PIT entries initiated by CCNinfo remain until the configured CCNinfo Reply Timeout (Section 7.1) is expired. In other words, unlike the ordinary Interest-Data exchanges in CCN, if routers that accept the full discovery request receive the full discovery request, the routers SHOULD NOT remove the PIT entry created by the full discovery request until the CCNinfo Reply Timeout value expires.¶
Note that the full discovery request is an OPTIONAL implementation of CCNinfo; it may not be implemented on routers. Even if it is implemented on a router, it may not accept the full discovery request from non-validated CCNinfo users or routers or because of its policy. If a router does not accept the full discovery request, it rejects the full discovery request as described in Section 6.11. Routers that enable the full discovery request MAY rate-limit Replies, as described in Section 10.8 as well.¶
If there is a caching router or FHR for the specified content within the specified hop count along the path, the caching router or FHR sends back the Reply message toward the CCNinfo user and terminates the Request.¶
When a router decides to send a Reply message to its downstream neighbor router or the CCNinfo user with a NO_ERROR return code, it inserts a Report block with the Request Arrival Time and Node Identifier values to the Request message. Then, the router inserts the corresponding Reply sub-block(s) (Figure 10) to the payload. The router finally changes the Type field in the fixed header from PT_CCNINFO_REQUEST to PT_CCNINFO_REPLY and forwards the message back as the Reply toward the CCNinfo user in a hop-by-hop manner.¶
If a router cannot continue the Request, the router MUST put an appropriate ReturnCode in the Request message, change the Type field value in the fixed header from PT_CCNINFO_REQUEST to PT_CCNINFO_REPLY, and forward the Reply message back toward the CCNinfo user to terminate the Request (see Section 6).¶
When a router receives a CCNinfo Reply whose Request ID and Node Identifier values match those in the PIT entry, which is sent from a valid adjacent neighbor router, it forwards the CCNinfo Reply back toward the CCNinfo user. If the router does not receive the corresponding Reply within the [CCNinfo Reply Timeout] period, then it removes the corresponding PIT entry and terminates the trace.¶
The Flags field in the Request block TLV is used to indicate whether the router keeps the PIT entry during the CCNinfo Reply Timeout even after one or more corresponding Reply messages are forwarded. When the CCNinfo user does not set the F flag (i.e., "0"), the intermediate routers immediately remove the PIT entry whenever they forward the corresponding Reply message. When the CCNinfo user sets the F flag (i.e., "1"), which means the CCNinfo user chooses the "full discovery request" (see Section 5.3.2), the intermediate routers keep the PIT entry within the [CCNinfo Reply Timeout] period. After this timeout, the PIT entry is removed.¶
CCNinfo Replies MUST NOT be cached in routers upon the transmission of Reply messages.¶
Within a network with a multipath condition, there is a case (Figure 12) wherein a single CCNinfo Request is split into multiple Requests (e.g., at Router A), which are injected into a single router (Router D). In this case, multiple Replies with the same Request ID and Node Identifier values, including different Report blocks, are received by the router (Router D).¶
To recognize different CCNinfo Reply messages, the routers MUST distinguish the PIT entries by the Request ID and exploiting path labels, which could be a hash value of the concatenation information of the cumulate node identifiers in the hop-by-hop header and the specified content name. For example, when Router D in Figure 12 receives a CCNinfo Request from Router B, its PIT includes the Request ID and value such as H((Router_A|Router_B)|content_name), where "H" indicates some hash function and "|" indicates concatenation. When Router D receives a CCNinfo Request from Router C, its PIT includes the same Request ID and value of H((Router_A|Router_C)|content_name). Two different Replies are later received on Router D, and each Reply is appropriately forwarded to Router B and Router C, respectively. Note that two Reply messages coming from Router B and Router C are reached at Router A, but the CCNinfo user can only receive the first Reply message either from Router B or Router C as Router A removes the corresponding PIT entry after it forwards the first Reply.¶
To avoid routing loops, when a router seeks the cumulate node identifiers of the Report blocks in the hop-by-hop header, it MUST examine whether its own node identifier is not previously inserted. If a router detects its own node identifier in the hop-by-hop header, the router inserts its Report block and terminates the Request as will be described in Section 6.8.¶
When performing a hop-by-hop trace, it is necessary to determine when to stop the trace. There are several cases when an intermediate router might return a Reply before a Request reaches the caching router or the FHR.¶
A CCNinfo Request can be determined to have arrived at the FHR. To ensure that a router recognizes that it is the FHR for the specified content, it needs to have a FIB entry (or to attach) to the corresponding publisher or the content.¶
A CCNinfo Request can be determined to have arrived at the router having the specified content cache within the specified HopLimit.¶
A CCNinfo Request can be determined to have arrived at the last router of the specified HopLimit. If the last router does not have the corresponding cache, it MUST insert its Report block and send the Reply message with a NO_INFO return code without appending any Reply block or sub-block TLVs.¶
If the router does not validate the Request or the Reply even it is required, the router MUST note a ReturnCode of INVALID_REQUEST in the fixed header of the message, insert its Report block, and forward the message as the Reply back to the CCNinfo user. The router MAY, however, randomly ignore the received invalid messages. (See Section 10.7.)¶
If the router cannot determine the routing paths or neighbor routers for the specified name prefix within the specified HopLimit, it MUST note a ReturnCode of NO_ROUTE in the fixed header of the message, insert its Report block, and forward the message as the Reply back to the CCNinfo user.¶
If the router does not have any information about the specified name prefix within the specified HopLimit, it MUST note a ReturnCode of NO_INFO in the fixed header of the message, insert its Report block, and forward the message as the Reply back to the CCNinfo user.¶
If appending the Report block, the Reply block, or Reply sub-block would make the hop-by-hop header longer than 247 bytes or the Request packet longer than the MTU of the Incoming face, the router MUST note a ReturnCode of NO_SPACE in the fixed header of the message and forward the message as the Reply back to the CCNinfo user.¶
If a CCNinfo Request has encountered a fatal error, the router MUST note a ReturnCode of FATAL_ERROR in the fixed header of the message and forward the message as the Reply back to the CCNinfo user. This may happen, for example, when the router detects some routing loop in the Request blocks (see Section 1). The fatal error can be encoded with another error: if a router detects routing loop but cannot insert its Report block, it MUST note NO_SPACE and FATAL_ERROR ReturnCodes (i.e., 0x85) in the fixed header and forward the message back to the CCNinfo user.¶
If a router receives the Request or Reply message that expires its own [CCNinfo Reply Timeout] value (Section 7.1), the router will silently discard the Request or Reply message.¶
Cases will arise in which a router or a FHR along the path does not support CCNinfo. In such cases, a CCNinfo user and routers that forward the CCNinfo Request will time out the CCNinfo request.¶
If CCNinfo is administratively prohibited, the router rejects the Request message and MUST send the CCNinfo Reply with the ReturnCode of ADMIN_PROHIB. The router MAY, however, randomly ignore the Request messages to be rejected (see Section 10.7).¶
The [CCNinfo Reply Timeout] value is used to time out a CCNinfo Reply. The value for a router can be statically configured by the router's administrators and/or operators. The default value is 3 (seconds). The [CCNinfo Reply Timeout] value SHOULD NOT be larger than 4 (seconds) and SHOULD NOT be lower than 2 (seconds).¶
If a CCNinfo user does not specify the HopLimit value in the fixed header for a Request message as the HopLimit, the HopLimit is set to 32. Note that 0 HopLimit is an invalid Request; hence, the router in this case follows the way defined in Section 6.4.¶
A router MAY configure the valid or invalid networks to enable an access control. The access control MAY be defined per name prefix, such as "who can retrieve which name prefix" (see Section 10.2).¶
A CCNinfo Request message is forwarded in a hop-by-hop manner and each forwarding router appends its own Report block. We can then verify the number of hops to reach the content forwarder or publisher and the RTT between the content forwarder or publisher.¶
While some routers may hide their node identifiers with all-zeros in the Report blocks (as seen in Section 10.1), the routers in the path from the CCNinfo user to the content forwarder can be identified.¶
By taking the HopLimit from the content forwarder and forwarding the TTL threshold over all hops, it is possible to discover the TTL or hop limit required for the content forwarder to reach the CCNinfo user.¶
If the routers have synchronized clocks, it is possible to estimate the propagation and queuing delays from the differences between the timestamps at the successive hops. However, this delay includes the control processing overhead; therefore, it is not necessarily indicative of the delay that would be experienced by the data traffic.¶
By obtaining the path stretch "d / P", where "d" is the hop count of the data and "P" is the hop count from the consumer to the publisher, we can measure the improvements in path stretch in various cases, such as in different caching and routing algorithms. We can then facilitate the investigation of the performance of the protocol.¶
CCNinfo can show the number of received Interests per cache or chunk on a router. Accordingly, CCNinfo measures the content popularity (i.e., the number of accesses for each content and/or cache), thereby enabling the investigation of the routing/caching strategy in networks.¶
This section details each kind of CCNx protocol value that has been registered. As per [RFC8126], four assignments have been made in existing registries, and a new Reply Type registry has been created in the "Content-Centric Networking (CCNx)" registry group.¶
As shown in Table 1, CCNinfo defines two packet types, PT_CCNINFO_REQUEST and PT_CCNINFO_REPLY, whose values are 0x03 and 0x04, respectively.¶
As shown in Table 2, CCNinfo defines one top-level type, T_DISCOVERY, whose value is 0x0005.¶
As shown in Table 4, CCNinfo defines two hop-by-hop types, T_DISC_REQHDR and T_DISC_REPORT, whose values are 0x0008 and 0x0009, respectively.¶
As shown in Table 6, CCNinfo defines two message types, T_DISC_REQ and T_DISC_REPLY, whose values are 0x000D and 0x000E, respectively.¶
IANA has created the "CCNx Reply Types" registry and allocated the reply types. The registration procedure is "RFC Required" [RFC8126]. The Type value is 2 octets. The range is 0x0000-0xFFFF. As shown in Table 7, CCNinfo defines three reply types, T_DISC_CONTENT, T_DISC_CONTENT_PUBLISHER, and T_ORG, whose values are 0x0000, 0x0001, and 0x0FFF, respectively.¶
This section addresses some of the security considerations.¶
Although CCNinfo gives excellent troubleshooting cues, some network administrators or operators may not want to disclose everything about their network to the public or may wish to securely transmit private information to specific members of their networks. CCNinfo provides policy-based information provisioning, thereby allowing network administrators to specify their response policy for each router.¶
The access policy regarding "who is allowed to retrieve" and/or "what kind of cache information" can be defined for each router. For the former type of access policy, routers with the specified content MAY examine the signature enclosed in the Request message and decide whether they should notify the content information in the Reply. If the routers decide to not notify the content information, they MUST send the CCNinfo Reply with the ReturnCode of ADMIN_PROHIB without appending any Reply block or sub-block TLVs. For the latter type of policy, the permission, whether (1) All (all cache information is disclosed), (2) Partial (cache information with a particular name prefix can (or cannot) be disclosed), or (3) Deny (no cache information is disclosed), is defined at the routers.¶
In contrast, we entail that each router does not disrupt the forwarding of CCNinfo Request and Reply messages. When a Request message is received, the router SHOULD insert the Report block if the ReturnCode is NO_ERROR. Here, according to the policy configuration, the Node Identifier field in the Report block MAY be null (i.e., all-zeros), but the Request Arrival Time field SHOULD NOT be null. Finally, the router SHOULD forward the Request message to the upstream router toward the content forwarder if the ReturnCode is kept with NO_ERROR.¶
A router MAY support an access control mechanism to filter out Requests from invalid CCNinfo users. To accomplish this, invalid networks (or domains) could, for example, be configured via a list of allowed or disallowed networks (as observed in Section 7.3). If a Request is received from a disallowed network (according to the node identifier in the Request block), the Request MUST NOT be processed and the Reply with the ReturnCode of INFO_HIDDEN SHOULD be used to note that. The router MAY, however, perform rate-limited logging of such events.¶
CCNinfo can be used to discover actively used topologies. If a network topology is not disclosed, CCNinfo Requests SHOULD be restricted at the border of the domain using the ADMIN_PROHIB return code.¶
CCNinfo can be used to discover the type of content being sent by publishers. If this information is a secret, CCNinfo Requests SHOULD be restricted at the border of the domain, using the ADMIN_PROHIB return code.¶
CCNinfo may impose heavy tasks at content forwarders because it makes content forwarders seek their internal cache states reported in the Reply messages whenever they form the Reply messages. The current CCNinfo specification allows to return null values for several fields, such as First/Last Seqnum or Elapsed Cache Time fields in the Reply sub-block. As mentioned in Section 3.2.1.1, these values MAY be null. This means that the content forwarder cannot only hide these values owing to privacy and security policies but also skip the implementations of the complex functions to report these values.¶
Routers can configure CCNinfo Reply Timeout (Section 7.1), which is the allowable timeout value to keep the PIT entry. If routers configure a longer timeout value, there may be an attractive attack vector against the PIT memory. Moreover, especially when the full discovery request option (Section 5.3) is specified for the CCNinfo Request, several Reply messages may be returned and cause a response storm. (See Section 10.8 for rate-limiting to avoid the storm). To avoid DoS attacks, routers MAY configure the timeout value, which is shorter than the user-configured CCNinfo timeout value. However, if it is too short, the Request may be timed out and the CCNinfo user does not receive all Replies; they only retrieve the partial path information (i.e., information about a part of the tree).¶
There may be a way to enable incremental exploration (i.e., to explore the part of the tree that was not explored by the previous operation); however, discussing such mechanisms is out of scope of this document.¶
A router MAY rate-limit CCNinfo Requests by ignoring some of the consecutive messages. The router MAY randomly ignore the received messages to minimize the processing overhead, i.e., to keep fairness in processing requests or to prevent traffic amplification. In such a case, no error message is returned. The rate limit function is left to the router's implementation.¶
CCNinfo supporting multipath forwarding may result in one Request returning multiple Reply messages. To prevent abuse, the routers in the traced path MAY need to rate-limit the Replies. In such a case, no error message is returned. The rate limit function is left to the router's implementation.¶
It is assumed that the CCNinfo Request and Reply messages are forwarded by adjacent neighbor nodes or routers. The CCNx message format or semantics do not define a secure way to verify the node and/or router adjacency, while a hop-by-hop authentication such as [DCAuth] provides a possible method for an adjacency verification and defines the corresponding message format for adjacency verification as well as the router behaviors. CCNinfo MAY use a similar method for node adjacency verification.¶
CCNinfo is implemented in Cefore [Cefore] [Cefore-site]. The command invoked by the CCNinfo user (e.g., consumer) is named "ccninfo". The ccninfo command sends the Request message and receives the Reply message(s). There are several options that can be specified with ccninfo, while the content name prefix (e.g., ccnx:/news/today) is the mandatory parameter.¶
The usage of the ccninfo command is as follows:¶
ccninfo [-c] [-f] [-o] [-V] [-r hop_count] [-s hop_count] [-v algorithm] name_prefix¶
The authors would like to thank Jérôme François, Erik Kline, Spyridon Mastorakis, Paulo Mendes, Ilya Moiseenko, David Oran, and Thierry Turletti for their valuable comments and suggestions on this document.¶