RFC 6353
Transport Layer Security (TLS) Transport Model for the Simple Network Management Protocol (SNMP)
Pages: 65
Internet Standard: 78
STD 78 is also: 5343 5590 5591
Obsoletes: 5953
Updated by: 8996 9456
Internet Standard: 78
STD 78 is also: 5343 5590 5591
Obsoletes: 5953
Updated by: 8996 9456
Part 2 of 4 – Pages 8 to 31
3. How the TLSTM Fits into the Transport Subsystem
A transport model is a component of the Transport Subsystem. The TLS Transport Model thus fits between the underlying (D)TLS transport layer and the Message Dispatcher [RFC3411] component of the SNMP engine.
The TLS Transport Model will establish a session between itself and the TLS Transport Model of another SNMP engine. The sending transport model passes unencrypted and unauthenticated messages from the Dispatcher to (D)TLS to be encrypted and authenticated, and the receiving transport model accepts decrypted and authenticated/ integrity-checked incoming messages from (D)TLS and passes them to the Dispatcher. After a TLS Transport Model session is established, SNMP messages can conceptually be sent through the session from one SNMP message Dispatcher to another SNMP Message Dispatcher. If multiple SNMP messages are needed to be passed between two SNMP applications they MAY be passed through the same session. A TLSTM implementation engine MAY choose to close the session to conserve resources. The TLS Transport Model of an SNMP engine will perform the translation between (D)TLS-specific security parameters and SNMP- specific, model-independent parameters.
The diagram below depicts where the TLS Transport Model (shown as "(D)TLS TM") fits into the architecture described in RFC 3411 and the Transport Subsystem: +------------------------------+ | Network | +------------------------------+ ^ ^ ^ | | | v v v +-------------------------------------------------------------------+ | +--------------------------------------------------+ | | | Transport Subsystem | +--------+ | | | +-----+ +-----+ +-------+ +-------+ | | | | | | | UDP | | SSH | |(D)TLS | . . . | other |<--->| Cache | | | | | | | TM | | TM | | | | | | | | | +-----+ +-----+ +-------+ +-------+ | +--------+ | | +--------------------------------------------------+ ^ | | ^ | | | | | | | Dispatcher v | | | +--------------+ +---------------------+ +----------------+ | | | | Transport | | Message Processing | | Security | | | | | Dispatch | | Subsystem | | Subsystem | | | | | | | +------------+ | | +------------+ | | | | | | | +->| v1MP |<--->| | USM | | | | | | | | | +------------+ | | +------------+ | | | | | | | | +------------+ | | +------------+ | | | | | | | +->| v2cMP |<--->| | Transport | | | | | | Message | | | +------------+ | | | Security |<--+ | | | Dispatch <---->| +------------+ | | | Model | | | | | | | +->| v3MP |<--->| +------------+ | | | | | | | +------------+ | | +------------+ | | | | PDU Dispatch | | | +------------+ | | | Other | | | | +--------------+ | +->| otherMP |<--->| | Model(s) | | | | ^ | +------------+ | | +------------+ | | | | +---------------------+ +----------------+ | | v | | +-------+-------------------------+---------------+ | | ^ ^ ^ | | | | | | | v v v |
| +-------------+ +---------+ +--------------+ +-------------+ | | | COMMAND | | ACCESS | | NOTIFICATION | | PROXY | | | | RESPONDER |<->| CONTROL |<->| ORIGINATOR | | FORWARDER | | | | application | | | | applications | | application | | | +-------------+ +---------+ +--------------+ +-------------+ | | ^ ^ | | | | | | v v | | +----------------------------------------------+ | | | MIB instrumentation | SNMP entity | +-------------------------------------------------------------------+3.1. Security Capabilities of This Model
3.1.1. Threats
The TLS Transport Model provides protection against the threats identified by the RFC 3411 architecture [RFC3411]: 1. Modification of Information - The modification threat is the danger that an unauthorized entity may alter in-transit SNMP messages generated on behalf of an authorized principal in such a way as to effect unauthorized management operations, including falsifying the value of an object. (D)TLS provides verification that the content of each received message has not been modified during its transmission through the network, data has not been altered or destroyed in an unauthorized manner, and data sequences have not been altered to an extent greater than can occur non-maliciously. 2. Masquerade - The masquerade threat is the danger that management operations unauthorized for a given principal may be attempted by assuming the identity of another principal that has the appropriate authorizations. The TLSTM verifies the identity of the (D)TLS server through the use of the (D)TLS protocol and X.509 certificates. A TLS Transport Model implementation MUST support the authentication of both the server and the client. 3. Message stream modification - The re-ordering, delay, or replay of messages can and does occur through the natural operation of many connectionless transport services. The message stream modification threat is the danger that messages may be maliciously re-ordered, delayed, or replayed to an extent that is greater than can occur through the natural operation of
connectionless transport services, in order to effect
unauthorized management operations.
(D)TLS provides replay protection with a Message Authentication
Code (MAC) that includes a sequence number. Since UDP provides
no sequencing ability, DTLS uses a sliding window protocol with
the sequence number used for replay protection (see [RFC4347]).
4. Disclosure - The disclosure threat is the danger of eavesdropping
on the exchanges between SNMP engines.
(D)TLS provides protection against the disclosure of information
to unauthorized recipients or eavesdroppers by allowing for
encryption of all traffic between SNMP engines. A TLS Transport
Model implementation MUST support message encryption to protect
sensitive data from eavesdropping attacks.
5. Denial of Service - The RFC 3411 architecture [RFC3411] states
that denial-of-service (DoS) attacks need not be addressed by an
SNMP security protocol. However, connectionless transports (like
DTLS over UDP) are susceptible to a variety of DoS attacks
because they are more vulnerable to spoofed IP addresses. See
Section 4.2 for details on how the cookie mechanism is used.
Note, however, that this mechanism does not provide any defense
against DoS attacks mounted from valid IP addresses.
See Section 9 for more detail on the security considerations
associated with the TLSTM and these security threats.
3.1.2. Message Protection
The RFC 3411 architecture recognizes three levels of security:
o without authentication and without privacy (noAuthNoPriv)
o with authentication but without privacy (authNoPriv)
o with authentication and with privacy (authPriv)
The TLS Transport Model determines from (D)TLS the identity of the
authenticated principal, the transport type, and the transport
address associated with an incoming message. The TLS Transport Model
provides the identity and destination type and address to (D)TLS for
outgoing messages.
When an application requests a session for a message, it also
requests a security level for that session. The TLS Transport Model
MUST ensure that the (D)TLS connection provides security at least as
high as the requested level of security. How the security level is translated into the algorithms used to provide data integrity and privacy is implementation dependent. However, the NULL integrity and encryption algorithms MUST NOT be used to fulfill security level requests for authentication or privacy. Implementations MAY choose to force (D)TLS to only allow cipher_suites that provide both authentication and privacy to guarantee this assertion. If a suitable interface between the TLS Transport Model and the (D)TLS Handshake Protocol is implemented to allow the selection of security-level-dependent algorithms (for example, a security level to cipher_suites mapping table), then different security levels may be utilized by the application. The authentication, integrity, and privacy algorithms used by the (D)TLS Protocols may vary over time as the science of cryptography continues to evolve and the development of (D)TLS continues over time. Implementers are encouraged to plan for changes in operator trust of particular algorithms. Implementations SHOULD offer configuration settings for mapping algorithms to SNMPv3 security levels.3.1.3. (D)TLS Connections
(D)TLS connections are opened by the TLS Transport Model during the elements of procedure for an outgoing SNMP message. Since the sender of a message initiates the creation of a (D)TLS connection if needed, the (D)TLS connection will already exist for an incoming message. Implementations MAY choose to instantiate (D)TLS connections in anticipation of outgoing messages. This approach might be useful to ensure that a (D)TLS connection to a given target can be established before it becomes important to send a message over the (D)TLS connection. Of course, there is no guarantee that a pre-established session will still be valid when needed. DTLS connections, when used over UDP, are uniquely identified within the TLS Transport Model by the combination of transportDomain, transportAddress, tmSecurityName, and requestedSecurityLevel associated with each session. Each unique combination of these parameters MUST have a locally chosen unique tlstmSessionID for each active session. For further information, see Section 5. TLS over TCP sessions, on the other hand, do not require a unique pairing of address and port attributes since their lower-layer protocols (TCP) already provide adequate session framing. But they must still provide a unique tlstmSessionID for referencing the session.
The tlstmSessionID MUST NOT change during the entire duration of the session from the TLSTM's perspective, and MUST uniquely identify a single session. As an implementation hint: note that the (D)TLS internal SessionID does not meet these requirements, since it can change over the life of the connection as seen by the TLSTM (for example, during renegotiation), and does not necessarily uniquely identify a TLSTM session (there can be multiple TLSTM sessions sharing the same D(TLS) internal SessionID).3.2. Security Parameter Passing
For the (D)TLS server-side, (D)TLS-specific security parameters (i.e., cipher_suites, X.509 certificate fields, IP addresses, and ports) are translated by the TLS Transport Model into security parameters for the TLS Transport Model and security model (e.g., tmSecurityLevel, tmSecurityName, transportDomain, transportAddress). The transport-related and (D)TLS-security-related information, including the authenticated identity, are stored in a cache referenced by tmStateReference. For the (D)TLS client side, the TLS Transport Model takes input provided by the Dispatcher in the sendMessage() Abstract Service Interface (ASI) and input from the tmStateReference cache. The (D)TLS Transport Model converts that information into suitable security parameters for (D)TLS and establishes sessions as needed. The elements of procedure in Section 5 discuss these concepts in much greater detail.3.3. Notifications and Proxy
(D)TLS connections may be initiated by (D)TLS clients on behalf of SNMP applications that initiate communications, such as command generators, notification originators, proxy forwarders. Command generators are frequently operated by a human, but notification originators and proxy forwarders are usually unmanned automated processes. The targets to whom notifications and proxied requests should be sent are typically determined and configured by a network administrator. The SNMP-TARGET-MIB module [RFC3413] contains objects for defining management targets, including transportDomain, transportAddress, securityName, securityModel, and securityLevel parameters, for notification originator, proxy forwarder, and SNMP-controllable command generator applications. Transport domains and transport addresses are configured in the snmpTargetAddrTable, and the securityModel, securityName, and securityLevel parameters are configured in the snmpTargetParamsTable. This document defines a MIB
module that extends the SNMP-TARGET-MIB's snmpTargetParamsTable to specify a (D)TLS client-side certificate to use for the connection. When configuring a (D)TLS target, the snmpTargetAddrTDomain and snmpTargetAddrTAddress parameters in snmpTargetAddrTable SHOULD be set to the snmpTLSTCPDomain or snmpDTLSUDPDomain object and an appropriate snmpTLSAddress value. When used with the SNMPv3 message processing model, the snmpTargetParamsMPModel column of the snmpTargetParamsTable SHOULD be set to a value of 3. The snmpTargetParamsSecurityName SHOULD be set to an appropriate securityName value, and the snmpTlstmParamsClientFingerprint parameter of the snmpTlstmParamsTable SHOULD be set to a value that refers to a locally held certificate (and the corresponding private key) to be used. Other parameters, for example, cryptographic configuration such as which cipher_suites to use, must come from configuration mechanisms not defined in this document. The securityName defined in the snmpTargetParamsSecurityName column will be used by the access control model to authorize any notifications that need to be sent.4. Elements of the Model
This section contains definitions required to realize the (D)TLS Transport Model defined by this document.4.1. X.509 Certificates
(D)TLS can make use of X.509 certificates for authentication of both sides of the transport. This section discusses the use of X.509 certificates in the TLSTM. While (D)TLS supports multiple authentication mechanisms, this document only discusses X.509-certificate-based authentication; other forms of authentication are outside the scope of this specification. TLSTM implementations are REQUIRED to support X.509 certificates.4.1.1. Provisioning for the Certificate
Authentication using (D)TLS will require that SNMP entities have certificates, either signed by trusted Certification Authorities (CAs), or self signed. Furthermore, SNMP entities will most commonly need to be provisioned with root certificates that represent the list of trusted CAs that an SNMP entity can use for certificate verification. SNMP entities SHOULD also be provisioned with an X.509 certificate revocation mechanism which can be used to verify that a certificate has not been revoked. Trusted public keys from either CA certificates and/or self-signed certificates MUST be installed into
the server through a trusted out-of-band mechanism and their
authenticity MUST be verified before access is granted.
Having received a certificate from a connecting TLSTM client, the
authenticated tmSecurityName of the principal is derived using the
snmpTlstmCertToTSNTable. This table allows mapping of incoming
connections to tmSecurityNames through defined transformations. The
transformations defined in the SNMP-TLS-TM-MIB include:
o Mapping a certificate's subjectAltName or CommonName components to
a tmSecurityName, or
o Mapping a certificate's fingerprint value to a directly specified
tmSecurityName
As an implementation hint: implementations may choose to discard any
connections for which no potential snmpTlstmCertToTSNTable mapping
exists before performing certificate verification to avoid expending
computational resources associated with certificate verification.
Deployments SHOULD map the "subjectAltName" component of X.509
certificates to the TLSTM specific tmSecurityNames. The
authenticated identity can be obtained by the TLS Transport Model by
extracting the subjectAltName(s) from the peer's certificate. The
receiving application will then have an appropriate tmSecurityName
for use by other SNMPv3 components like an access control model.
An example of this type of mapping setup can be found in Appendix A.
This tmSecurityName may be later translated from a TLSTM specific
tmSecurityName to an SNMP engine securityName by the security model.
A security model, like the TSM security model [RFC5591], may perform
an identity mapping or a more complex mapping to derive the
securityName from the tmSecurityName offered by the TLS Transport
Model.
The standard View-Based Access Control Model (VACM) access control
model constrains securityNames to be 32 octets or less in length. A
TLSTM generated tmSecurityName, possibly in combination with a
messaging or security model that increases the length of the
securityName, might cause the securityName length to exceed 32
octets. For example, a 32-octet tmSecurityName derived from an IPv6
address, paired with a TSM prefix, will generate a 36-octet
securityName. Such a securityName will not be able to be used with
standard VACM or TARGET MIB modules. Operators should be careful to
select algorithms and subjectAltNames to avoid this situation.
A pictorial view of the complete transformation process (using the
TSM security model for the example) is shown below:
+-------------+ +-------+ +-----+
| Certificate | | | | |
| Path | | TLSTM | tmSecurityName | TSM |
| Validation | --> | | ----------------->| |
+-------------+ +-------+ +-----+
|
| securityName
V
+-------------+
| application |
+-------------+
4.2. (D)TLS Usage
(D)TLS MUST negotiate a cipher_suite that uses X.509 certificates for
authentication, and MUST authenticate both the client and the server.
The mandatory-to-implement cipher_suite is specified in the TLS
specification [RFC5246].
TLSTM verifies the certificates when the connection is opened (see
Section 5.3). For this reason, TLS renegotiation with different
certificates MUST NOT be done. That is, implementations MUST either
disable renegotiation completely (RECOMMENDED), or they MUST present
the same certificate during renegotiation (and MUST verify that the
other end presented the same certificate).
For DTLS over UDP, each SNMP message MUST be placed in a single UDP
datagram; it MAY be split to multiple DTLS records. In other words,
if a single datagram contains multiple DTLS application_data records,
they are concatenated when received. The TLSTM implementation SHOULD
return an error if the SNMP message does not fit in the UDP datagram,
and thus cannot be sent.
For DTLS over UDP, the DTLS server implementation MUST support DTLS
cookies ([RFC4347] already requires that clients support DTLS
cookies). Implementations are not required to perform the cookie
exchange for every DTLS handshake; however, enabling it by default is
RECOMMENDED.
For DTLS, replay protection MUST be used.
4.3. SNMP Services
This section describes the services provided by the TLS Transport Model with their inputs and outputs. The services are between the Transport Model and the Dispatcher. The services are described as primitives of an abstract service interface (ASI) and the inputs and outputs are described as abstract data elements as they are passed in these abstract service primitives.4.3.1. SNMP Services for an Outgoing Message
The Dispatcher passes the information to the TLS Transport Model using the ASI defined in the Transport Subsystem: statusInformation = sendMessage( IN destTransportDomain -- transport domain to be used IN destTransportAddress -- transport address to be used IN outgoingMessage -- the message to send IN outgoingMessageLength -- its length IN tmStateReference -- reference to transport state ) The abstract data elements returned from or passed as parameters into the abstract service primitives are as follows: statusInformation: An indication of whether the sending of the message was successful. If not, it is an indication of the problem. destTransportDomain: The transport domain for the associated destTransportAddress. The Transport Model uses this parameter to determine the transport type of the associated destTransportAddress. This document specifies the snmpTLSTCPDomain and the snmpDTLSUDPDomain transport domains. destTransportAddress: The transport address of the destination TLS Transport Model in a format specified by the SnmpTLSAddress TEXTUAL-CONVENTION. outgoingMessage: The outgoing message to send to (D)TLS for encapsulation and transmission. outgoingMessageLength: The length of the outgoingMessage.
tmStateReference: A reference used to pass model-specific and
mechanism-specific parameters between the Transport Subsystem and
transport-aware Security Models.
4.3.2. SNMP Services for an Incoming Message
The TLS Transport Model processes the received message from the
network using the (D)TLS service and then passes it to the Dispatcher
using the following ASI:
statusInformation =
receiveMessage(
IN transportDomain -- origin transport domain
IN transportAddress -- origin transport address
IN incomingMessage -- the message received
IN incomingMessageLength -- its length
IN tmStateReference -- reference to transport state
)
The abstract data elements returned from or passed as parameters into
the abstract service primitives are as follows:
statusInformation: An indication of whether the passing of the
message was successful. If not, it is an indication of the
problem.
transportDomain: The transport domain for the associated
transportAddress. This document specifies the snmpTLSTCPDomain
and the snmpDTLSUDPDomain transport domains.
transportAddress: The transport address of the source of the
received message in a format specified by the SnmpTLSAddress
TEXTUAL-CONVENTION.
incomingMessage: The whole SNMP message after being processed by
(D)TLS.
incomingMessageLength: The length of the incomingMessage.
tmStateReference: A reference used to pass model-specific and
mechanism-specific parameters between the Transport Subsystem and
transport-aware Security Models.
4.4. Cached Information and References
When performing SNMP processing, there are two levels of state information that may need to be retained: the immediate state linking a request-response pair, and potentially longer-term state relating to transport and security. "Transport Subsystem for the Simple Network Management Protocol (SNMP)" [RFC5590] defines general requirements for caches and references.4.4.1. TLS Transport Model Cached Information
The TLS Transport Model has specific responsibilities regarding the cached information. See the Elements of Procedure in Section 5 for detailed processing instructions on the use of the tmStateReference fields by the TLS Transport Model.4.4.1.1. tmSecurityName
The tmSecurityName MUST be a human-readable name (in snmpAdminString format) representing the identity that has been set according to the procedures in Section 5. The tmSecurityName MUST be constant for all traffic passing through a single TLSTM session. Messages MUST NOT be sent through an existing (D)TLS connection that was established using a different tmSecurityName. On the (D)TLS server side of a connection, the tmSecurityName is derived using the procedures described in Section 5.3.2 and the SNMP- TLS-TM-MIB's snmpTlstmCertToTSNTable DESCRIPTION clause. On the (D)TLS client side of a connection, the tmSecurityName is presented to the TLS Transport Model by the security model through the tmStateReference. This tmSecurityName is typically a copy of or is derived from the securityName that was passed by application (possibly because of configuration specified in the SNMP-TARGET-MIB). The Security Model likely derived the tmSecurityName from the securityName presented to the Security Model by the application (possibly because of configuration specified in the SNMP-TARGET-MIB). Transport-Model-aware security models derive tmSecurityName from a securityName, possibly configured in MIB modules for notifications and access controls. Transport Models SHOULD use predictable tmSecurityNames so operators will know what to use when configuring MIB modules that use securityNames derived from tmSecurityNames. The TLSTM generates predictable tmSecurityNames based on the configuration found in the SNMP-TLS-TM-MIB's snmpTlstmCertToTSNTable and relies on the network operators to have configured this table appropriately.
4.4.1.2. tmSessionID
The tmSessionID MUST be recorded per message at the time of receipt. When tmSameSecurity is set, the recorded tmSessionID can be used to determine whether the (D)TLS connection available for sending a corresponding outgoing message is the same (D)TLS connection as was used when receiving the incoming message (e.g., a response to a request).4.4.1.3. Session State
The per-session state that is referenced by tmStateReference may be saved across multiple messages in a Local Configuration Datastore. Additional session/connection state information might also be stored in a Local Configuration Datastore.5. Elements of Procedure
Abstract service interfaces have been defined by [RFC3411] and further augmented by [RFC5590] to describe the conceptual data flows between the various subsystems within an SNMP entity. The TLSTM uses some of these conceptual data flows when communicating between subsystems. To simplify the elements of procedure, the release of state information is not always explicitly specified. As a general rule, if state information is available when a message gets discarded, the message-state information should also be released. If state information is available when a session is closed, the session state information should also be released. Sensitive information, like cryptographic keys, should be overwritten appropriately prior to being released. An error indication in statusInformation will typically include the Object Identifier (OID) and value for an incremented error counter. This may be accompanied by the requested securityLevel and the tmStateReference. Per-message context information is not accessible to Transport Models, so for the returned counter OID and value, contextEngine would be set to the local value of snmpEngineID and contextName to the default context for error counters.5.1. Procedures for an Incoming Message
This section describes the procedures followed by the (D)TLS Transport Model when it receives a (D)TLS protected packet. The required functionality is broken into two different sections.
Section 5.1.1 describes the processing required for de-multiplexing multiple DTLS connections, which is specifically needed for DTLS over UDP sessions. It is assumed that TLS protocol implementations already provide appropriate message demultiplexing. Section 5.1.2 describes the transport processing required once the (D)TLS processing has been completed. This will be needed for all (D)TLS-based connections.5.1.1. DTLS over UDP Processing for Incoming Messages
Demultiplexing of incoming packets into separate DTLS sessions MUST be implemented. For connection-oriented transport protocols, such as TCP, the transport protocol takes care of demultiplexing incoming packets to the right connection. For DTLS over UDP, this demultiplexing will either need to be done within the DTLS implementation, if supported, or by the TLSTM implementation. Like TCP, DTLS over UDP uses the four-tuple <source IP, destination IP, source port, destination port> for identifying the connection (and relevant DTLS connection state). This means that when establishing a new session, implementations MUST use a different UDP source port number for each active connection to a remote destination IP-address/port-number combination to ensure the remote entity can disambiguate between multiple connections. If demultiplexing received UDP datagrams to DTLS connection state is done by the TLSTM implementation (instead of the DTLS implementation), the steps below describe one possible method to accomplish this. The important output results from the steps in this process are the remote transport address, incomingMessage, incomingMessageLength, and the tlstmSessionID. 1) The TLS Transport Model examines the raw UDP message, in an implementation-dependent manner. 2) The TLS Transport Model queries the Local Configuration Datastore (LCD) (see [RFC3411], Section 3.4.2) using the transport parameters (source and destination IP addresses and ports) to determine if a session already exists. 2a) If a matching entry in the LCD does not exist, then the UDP packet is passed to the DTLS implementation for processing. If the DTLS implementation decides to continue with the connection and allocate state for it, it returns a new DTLS connection handle (an implementation dependent detail). In
this case, TLSTM selects a new tlstmSessionId, and caches
this and the DTLS connection handle as a new entry in the
LCD (indexed by the transport parameters). If the DTLS
implementation returns an error or does not allocate
connection state (which can happen with the stateless cookie
exchange), processing stops.
2b) If a session does exist in the LCD, then its DTLS connection
handle (an implementation dependent detail) and its
tlstmSessionId is extracted from the LCD. The UDP packet
and the connection handle are passed to the DTLS
implementation. If the DTLS implementation returns success
but does not return an incomingMessage and an
incomingMessageLength, then processing stops (this is the
case when the UDP datagram contained DTLS handshake
messages, for example). If the DTLS implementation returns
an error, then processing stops.
3) Retrieve the incomingMessage and an incomingMessageLength from
DTLS. These results and the tlstmSessionID are used below in
Section 5.1.2 to complete the processing of the incoming message.
5.1.2. Transport Processing for Incoming SNMP Messages
The procedures in this section describe how the TLS Transport Model
should process messages that have already been properly extracted
from the (D)TLS stream. Note that care must be taken when processing
messages originating from either TLS or DTLS to ensure they're
complete and single. For example, multiple SNMP messages can be
passed through a single DTLS message and partial SNMP messages may be
received from a TLS stream. These steps describe the processing of a
singular SNMP message after it has been delivered from the (D)TLS
stream.
1) Determine the tlstmSessionID for the incoming message. The
tlstmSessionID MUST be a unique session identifier for this
(D)TLS connection. The contents and format of this identifier
are implementation dependent as long as it is unique to the
session. A session identifier MUST NOT be reused until all
references to it are no longer in use. The tmSessionID is equal
to the tlstmSessionID discussed in Section 5.1.1. tmSessionID
refers to the session identifier when stored in the
tmStateReference and tlstmSessionID refers to the session
identifier when stored in the LCD. They MUST always be equal
when processing a given session's traffic.
If this is the first message received through this session, and
the session does not have an assigned tlstmSessionID yet, then
the snmpTlstmSessionAccepts counter is incremented and a
tlstmSessionID for the session is created. This will only happen
on the server side of a connection because a client would have
already assigned a tlstmSessionID during the openSession()
invocation. Implementations may have performed the procedures
described in Section 5.3.2 prior to this point or they may
perform them now, but the procedures described in Section 5.3.2
MUST be performed before continuing beyond this point.
2) Create a tmStateReference cache for the subsequent reference and
assign the following values within it:
tmTransportDomain = snmpTLSTCPDomain or snmpDTLSUDPDomain as
appropriate.
tmTransportAddress = The address from which the message
originated.
tmSecurityLevel = The derived tmSecurityLevel for the session,
as discussed in Sections 3.1.2 and 5.3.
tmSecurityName = The derived tmSecurityName for the session as
discussed in Section 5.3. This value MUST be constant during
the lifetime of the session.
tmSessionID = The tlstmSessionID described in step 1 above.
3) The incomingMessage and incomingMessageLength are assigned values
from the (D)TLS processing.
4) The TLS Transport Model passes the transportDomain,
transportAddress, incomingMessage, and incomingMessageLength to
the Dispatcher using the receiveMessage ASI:
statusInformation =
receiveMessage(
IN transportDomain -- snmpTLSTCPDomain or snmpDTLSUDPDomain,
IN transportAddress -- address for the received message
IN incomingMessage -- the whole SNMP message from (D)TLS
IN incomingMessageLength -- the length of the SNMP message
IN tmStateReference -- transport info
)
5.2. Procedures for an Outgoing SNMP Message
The Dispatcher sends a message to the TLS Transport Model using the following ASI: statusInformation = sendMessage( IN destTransportDomain -- transport domain to be used IN destTransportAddress -- transport address to be used IN outgoingMessage -- the message to send IN outgoingMessageLength -- its length IN tmStateReference -- transport info ) This section describes the procedure followed by the TLS Transport Model whenever it is requested through this ASI to send a message. 1) If tmStateReference does not refer to a cache containing values for tmTransportDomain, tmTransportAddress, tmSecurityName, tmRequestedSecurityLevel, and tmSameSecurity, then increment the snmpTlstmSessionInvalidCaches counter, discard the message, and return the error indication in the statusInformation. Processing of this message stops. 2) Extract the tmSessionID, tmTransportDomain, tmTransportAddress, tmSecurityName, tmRequestedSecurityLevel, and tmSameSecurity values from the tmStateReference. Note: the tmSessionID value may be undefined if no session exists yet over which the message can be sent. 3) If tmSameSecurity is true and tmSessionID is either undefined or refers to a session that is no longer open, then increment the snmpTlstmSessionNoSessions counter, discard the message, and return the error indication in the statusInformation. Processing of this message stops. 4) If tmSameSecurity is false and tmSessionID refers to a session that is no longer available, then an implementation SHOULD open a new session, using the openSession() ASI (described in greater detail in step 5b). Instead of opening a new session an implementation MAY return an snmpTlstmSessionNoSessions error to the calling module and stop the processing of the message. 5) If tmSessionID is undefined, then use tmTransportDomain, tmTransportAddress, tmSecurityName, and tmRequestedSecurityLevel to see if there is a corresponding entry in the LCD suitable to send the message over.
5a) If there is a corresponding LCD entry, then this session
will be used to send the message.
5b) If there is no corresponding LCD entry, then open a session
using the openSession() ASI (discussed further in
Section 5.3.1). Implementations MAY wish to offer message
buffering to prevent redundant openSession() calls for the
same cache entry. If an error is returned from
openSession(), then discard the message, discard the
tmStateReference, increment the snmpTlstmSessionOpenErrors,
return an error indication to the calling module, and stop
the processing of the message.
6) Using either the session indicated by the tmSessionID (if there
was one) or the session resulting from a previous step (4 or 5),
pass the outgoingMessage to (D)TLS for encapsulation and
transmission.
5.3. Establishing or Accepting a Session
Establishing a (D)TLS connection as either a client or a server
requires slightly different processing. The following two sections
describe the necessary processing steps.
5.3.1. Establishing a Session as a Client
The TLS Transport Model provides the following primitive for use by a
client to establish a new (D)TLS connection:
statusInformation = -- errorIndication or success
openSession(
IN tmStateReference -- transport information to be used
OUT tmStateReference -- transport information to be used
IN maxMessageSize -- of the sending SNMP entity
)
The following describes the procedure to follow when establishing an
SNMP over a (D)TLS connection between SNMP engines for exchanging
SNMP messages. This process is followed by any SNMP client's engine
when establishing a session for subsequent use.
This procedure MAY be done automatically for an SNMP application that
initiates a transaction, such as a command generator, a notification
originator, or a proxy forwarder.
1) The snmpTlstmSessionOpens counter is incremented.
2) The client selects the appropriate certificate and cipher_suites
for the key agreement based on the tmSecurityName and the
tmRequestedSecurityLevel for the session. For sessions being
established as a result of an SNMP-TARGET-MIB based operation,
the certificate will potentially have been identified via the
snmpTlstmParamsTable mapping and the cipher_suites will have to
be taken from a system-wide or implementation-specific
configuration. If no row in the snmpTlstmParamsTable exists,
then implementations MAY choose to establish the connection using
a default client certificate available to the application.
Otherwise, the certificate and appropriate cipher_suites will
need to be passed to the openSession() ASI as supplemental
information or configured through an implementation-dependent
mechanism. It is also implementation-dependent and possibly
policy-dependent how tmRequestedSecurityLevel will be used to
influence the security capabilities provided by the (D)TLS
connection. However this is done, the security capabilities
provided by (D)TLS MUST be at least as high as the level of
security indicated by the tmRequestedSecurityLevel parameter.
The actual security level of the session is reported in the
tmStateReference cache as tmSecurityLevel. For (D)TLS to provide
strong authentication, each principal acting as a command
generator SHOULD have its own certificate.
3) Using the destTransportDomain and destTransportAddress values,
the client will initiate the (D)TLS handshake protocol to
establish session keys for message integrity and encryption.
If the attempt to establish a session is unsuccessful, then
snmpTlstmSessionOpenErrors is incremented, an error indication is
returned, and processing stops. If the session failed to open
because the presented server certificate was unknown or invalid,
then the snmpTlstmSessionUnknownServerCertificate or
snmpTlstmSessionInvalidServerCertificates MUST be incremented and
an snmpTlstmServerCertificateUnknown or
snmpTlstmServerInvalidCertificate notification SHOULD be sent as
appropriate. Reasons for server certificate invalidation
include, but are not limited to, cryptographic validation
failures and an unexpected presented certificate identity.
4) The (D)TLS client MUST then verify that the (D)TLS server's
presented certificate is the expected certificate. The (D)TLS
client MUST NOT transmit SNMP messages until the server
certificate has been authenticated, the client certificate has
been transmitted, and the TLS connection has been fully
established.
If the connection is being established from a configuration based
on SNMP-TARGET-MIB configuration, then the snmpTlstmAddrTable
DESCRIPTION clause describes how the verification is done (using
either a certificate fingerprint, or an identity authenticated
via certification path validation).
If the connection is being established for reasons other than
configuration found in the SNMP-TARGET-MIB, then configuration
and procedures outside the scope of this document should be
followed. Configuration mechanisms SHOULD be similar in nature
to those defined in the snmpTlstmAddrTable to ensure consistency
across management configuration systems. For example, a command-
line tool for generating SNMP GETs might support specifying
either the server's certificate fingerprint or the expected host
name as a command-line argument.
5) (D)TLS provides assurance that the authenticated identity has
been signed by a trusted configured Certification Authority. If
verification of the server's certificate fails in any way (for
example, because of failures in cryptographic verification or the
presented identity did not match the expected named entity), then
the session establishment MUST fail, and the
snmpTlstmSessionInvalidServerCertificates object is incremented.
If the session cannot be opened for any reason at all, including
cryptographic verification failures and snmpTlstmCertToTSNTable
lookup failures, then the snmpTlstmSessionOpenErrors counter is
incremented and processing stops.
6) The TLSTM-specific session identifier (tlstmSessionID) is set in
the tmSessionID of the tmStateReference passed to the TLS
Transport Model to indicate that the session has been established
successfully and to point to a specific (D)TLS connection for
future use. The tlstmSessionID is also stored in the LCD for
later lookup during processing of incoming messages
(Section 5.1.2).
5.3.2. Accepting a Session as a Server
A (D)TLS server should accept new session connections from any client
for which it is able to verify the client's credentials. This is
done by authenticating the client's presented certificate through a
certificate path validation process (e.g., [RFC5280]) or through
certificate fingerprint verification using fingerprints configured in
the snmpTlstmCertToTSNTable. Afterward, the server will determine
the identity of the remote entity using the following procedures.
The (D)TLS server identifies the authenticated identity from the (D)TLS client's principal certificate using configuration information from the snmpTlstmCertToTSNTable mapping table. The (D)TLS server MUST request and expect a certificate from the client and MUST NOT accept SNMP messages over the (D)TLS connection until the client has sent a certificate and it has been authenticated. The resulting derived tmSecurityName is recorded in the tmStateReference cache as tmSecurityName. The details of the lookup process are fully described in the DESCRIPTION clause of the snmpTlstmCertToTSNTable MIB object. If any verification fails in any way (for example, because of failures in cryptographic verification or because of the lack of an appropriate row in the snmpTlstmCertToTSNTable), then the session establishment MUST fail, and the snmpTlstmSessionInvalidClientCertificates object is incremented. If the session cannot be opened for any reason at all, including cryptographic verification failures, then the snmpTlstmSessionOpenErrors counter is incremented and processing stops. Servers that wish to support multiple principals at a particular port SHOULD make use of a (D)TLS extension that allows server-side principal selection like the Server Name Indication extension defined in Section 3.1 of [RFC4366]. Supporting this will allow, for example, sending notifications to a specific principal at a given TCP or UDP port.5.4. Closing a Session
The TLS Transport Model provides the following primitive to close a session: statusInformation = closeSession( IN tmSessionID -- session ID of the session to be closed ) The following describes the procedure to follow to close a session between a client and server. This process is followed by any SNMP engine closing the corresponding SNMP session. 1) Increment either the snmpTlstmSessionClientCloses or the snmpTlstmSessionServerCloses counter as appropriate. 2) Look up the session using the tmSessionID. 3) If there is no open session associated with the tmSessionID, then closeSession processing is completed.
4) Have (D)TLS close the specified connection. This MUST include
sending a close_notify TLS Alert to inform the other side that
session cleanup may be performed.
6. MIB Module Overview
This MIB module provides management of the TLS Transport Model. It
defines needed textual conventions, statistical counters,
notifications, and configuration infrastructure necessary for session
establishment. Example usage of the configuration tables can be
found in Appendix A.
6.1. Structure of the MIB Module
Objects in this MIB module are arranged into subtrees. Each subtree
is organized as a set of related objects. The overall structure and
assignment of objects to their subtrees, and the intended purpose of
each subtree, is shown below.
6.2. Textual Conventions
Generic and Common Textual Conventions used in this module can be
found summarized at http://www.ops.ietf.org/mib-common-tcs.html.
This module defines the following new Textual Conventions:
o A new TransportAddress format for describing (D)TLS connection
addressing requirements.
o A certificate fingerprint allowing MIB module objects to
generically refer to a stored X.509 certificate using a
cryptographic hash as a reference pointer.
6.3. Statistical Counters
The SNMP-TLS-TM-MIB defines counters that provide network management
stations with information about session usage and potential errors
that a device may be experiencing.
6.4. Configuration Tables
The SNMP-TLS-TM-MIB defines configuration tables that an
administrator can use for configuring a device for sending and
receiving SNMP messages over (D)TLS. In particular, there are MIB
tables that extend the SNMP-TARGET-MIB for configuring (D)TLS
certificate usage and a MIB table for mapping incoming (D)TLS client
certificates to SNMPv3 tmSecurityNames.
6.4.1. Notifications
The SNMP-TLS-TM-MIB defines notifications to alert management stations when a (D)TLS connection fails because a server's presented certificate did not meet an expected value (snmpTlstmServerCertificateUnknown) or because cryptographic validation failed (snmpTlstmServerInvalidCertificate).6.5. Relationship to Other MIB Modules
Some management objects defined in other MIB modules are applicable to an entity implementing the TLS Transport Model. In particular, it is assumed that an entity implementing the SNMP-TLS-TM-MIB will implement the SNMPv2-MIB [RFC3418], the SNMP-FRAMEWORK-MIB [RFC3411], the SNMP-TARGET-MIB [RFC3413], the SNMP-NOTIFICATION-MIB [RFC3413], and the SNMP-VIEW-BASED-ACM-MIB [RFC3415]. The SNMP-TLS-TM-MIB module contained in this document is for managing TLS Transport Model information.6.5.1. MIB Modules Required for IMPORTS
The SNMP-TLS-TM-MIB module imports items from SNMPv2-SMI [RFC2578], SNMPv2-TC [RFC2579], SNMP-FRAMEWORK-MIB [RFC3411], SNMP-TARGET-MIB [RFC3413], and SNMPv2-CONF [RFC2580].
(page 31 continued on part 3)