| NETCONF Working Group | K. Watsen |
| Internet-Draft | Juniper Networks |
| Intended status: Standards Track | J. Schoenwaelder |
| Expires: March 26, 2015 | Jacobs University Bremen |
| September 22, 2014 |
NETCONF Server Configuration Model
draft-ietf-netconf-server-model-03
This draft defines a NETCONF server configuration data model. This data model enables configuration of the NETCONF service itself, including which transports it supports, what ports they listen on, whether they support device-initiated connections, and associated parameters.
This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.
Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet-Drafts is at http://datatracker.ietf.org/drafts/current/.
Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress."
This Internet-Draft will expire on March 26, 2015.
Copyright (c) 2014 IETF Trust and the persons identified as the document authors. All rights reserved.
This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (http://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License.
This draft defines a NETCONF [RFC6241] server configuration data model. This data model enables configuration of the NETCONF service itself, including which transports are supported, what ports does the server listen on, whether call-home is supported, and associated parameters.
The keywords "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in RFC 2119 [RFC2119].
A simplified graphical representation of data models is used in this document. The meaning of the symbols in these diagrams is as follows:
The primary purpose of the YANG module defined herein is to enable the configuration of the NETCONF service on the device. This scope includes the following objectives:
The YANG module should support all current NETCONF transports, namely NETCONF over SSH [RFC6242] and NETCONF over TLS [rfc5539bis], and be extensible to support future transports as necessary.
Since implementations may not support all transports, the module should use YANG "feature" statements so that implementations can accurately advertise which transports are supported.
While each transport is unique in its protocol and may have some distinct configurations, there remains a significant overlap between them. Thus the YANG module should use "grouping" statements so that the common aspects can be configured similarly.
NETCONF has always supported the server opening a port to listen for client connections. More recently the NETCONF working group defined support for call-home ([draft-ietf-netconf-call-home]). The module should configure both listening for connections and call-home.
Since implementations may not support both listening for connections and call home, YANG "feature" statements should be used so that implementation can accurately advertise the connection types it supports.
The following objectives only pertain to call home connections.
A device may be managed by more than one northbound application. For instance, a deployment may have one application for provisioning and another for fault monitoring. Therefore, when it is desired for a device to initiate call home connections, it should be able to do so for more than one application.
An application managing a device may implement a high-availability strategy employing a multiplicity of active and/or passive servers. Therefore, when it is desired for a device to initiate call home connections, it should be able to connect to any of the applications servers.
Assuming an application has more than one server, then it becomes necessary to configure how a device should reconnect to the application should it lose its connection to the application's servers. Of primary interest is if the device should start with first server defined in a user-ordered list of servers or with the last server it was connected to. Secondary settings might specify the frequency of attempts and number of attempts per server. Therefore, a reconnection strategy should be configurable.
Applications may vary greatly on how frequently they need to interact with a device, how responsive interactions with devices need to be, and how many simultaneous connections they can support. Some applications may need a persistent connection to devices to optimize real-time interactions, while others are satisfied with periodic interactions and reduced resources required. Therefore, when it is necessary for devices to initiate connections, the type of connection desired should be configured.
The reconnection strategy should apply to both persistent and periodic connections. How it applies to periodic connections becomes clear when considering that a periodic "connection" is a logical connection to a single server. That is, the periods of unconnectedness are intentional as opposed to due to external reasons. A periodic "connection" should always reconnect to the same server until it is no longer able to, at which time the reconnection strategy guides how to connect to another server.
If a persistent connection is desired, it is the responsibility of the connection-initiator to actively test the aliveness of the connection. The connection initiator must immediately work to reestablish a persistent connection as soon as the connection is lost. How often the connection should be tested is driven by applications requirements, and therefore keep-alive settings should be configurable on a per-application basis.
If a periodic connection is desired, it is necessary for the device to know how often it should connect. This delay essentially determines how long the application might have to wait to send data to the device. This setting does not constrain how often the device must wait to send data to the application, as the device should immediately connect to the application whenever it has data to send to it.
A common communication pattern is that one data transmission is many times closely followed by another. For instance, if the device needs to send a notification message, there's a high probability that it will send another shortly thereafter. Likewise, the application may have a sequence of pending messages to send. Thus, it should be possible for a device to hold a connection open until some amount of time of no data being transmitted as transpired.
The following subtree illustrates how this YANG module enables configuration for listening for remote connections, as described in [RFC6242] and [rfc5539bis]. Feature statements are used to limit both if listening is supported at all as well as for which transports. If listening for connections is supported, then the model enables configuring a list of listening endpoints, each configured with a user-specified name (the key field), the transport to use (i.e. SSH, TLS), and the IP address and port to listen on. The port field is optional, defaulting to the transport-specific port when not configured.
INSERT_TEXT_FROM_FILE(refs/listen-tree.txt)
The following subtree illustrates how this YANG module enables configuration for call home, as described in [draft-ietf-netconf-call-home]. Feature statements are used to limit both if call-home is supported at all as well as for which transports, if it is. If call-home is supported, then the model supports configuring a list of applications to connect to. Each application is configured with a user-specified name (the key field), the transport to be used (i.e. SSH, TLS), and a list of remote endpoints, each having a name, an IP address, and an optional port. Additionally, the configuration for each remote application indicates the connection-type (persistent vs. periodic) and associated parameters, as well as the reconnection strategy to use.
INSERT_TEXT_FROM_FILE(refs/call-home-tree.txt)
The following subtree illustrates how this YANG module enables authentication of TLS client certificates and mapping TLS clients to NETCONF user names. More specifically, the "trusted-ca-certs" and "trusted-client-certs" containers are used to authenticate TLS client certificates, while "cert-maps" and "psk-maps" are used to map TLS clients to NETCONF user names.
INSERT_TEXT_FROM_FILE(refs/tls-client-auth-tree.txt)
This YANG module imports YANG types from [RFC6991], [RFC6536], and [draft-ietf-netmod-snmp-cfg].
RFC Ed.: update the date below with the date of RFC publication and remove this note. <CODE BEGINS> file "ietf-netconf-server@YYYY-MM-DD.yang" INSERT_TEXT_FROM_FILE(ietf-netconf-server.yang) <CODE ENDS>
One the objectives listed above, Keep-Alives for Persistent Connections [keepalives] indicates a need for a "keep-alive" mechanism. This section specifies how the NETCONF keep-alive mechanism is to be implemented.
Both SSH and TLS have the ability to support keep-alives. Using these mechanisms, the keep-alive messages are sent inside the encrypted tunnel, thus thwarting spoof attacks.
The SSH keep-alive solution that is expected to be used when configured using the data model defined in this document is ubiquitous in practice, though never being explicitly defined in an RFC. The strategy used is to purposely send a malformed request message with a flag set to ensure a response. More specifically, per section 4 of [RFC4253], either SSH peer can send a SSH_MSG_GLOBAL_REQUEST message with "want reply" set to '1' and that, if there is an error, will get back a SSH_MSG_REQUEST_FAILURE response. Similarly, section 5 of [RFC4253] says that either SSH peer can send a SSH_MSG_CHANNEL_REQUEST message with "want reply" set to '1' and that, if there is an error, will get back a SSH_MSG_CHANNEL_FAILURE response.
To ensure that the request will fail, current implementations send an invalid "request name" or "request type", respectively. Abiding to the extensibility guidelines specified in Section 6 of [RFC4251], these implementations use the "name@domain". For instance, when configured to send keep-alives, OpenSSH sends the string "keepalive@openssh.com". In order to remain compatible with existing implementations, this draft does not require a specific "request name" or "request type" string be used.
The TLS keep-alive solution is defined in [RFC6520]. This solution allows both peers to advertise if they can receive heartbeat request messages from its peer. For standard NETCONF over TLS connections, devices SHOULD advertise "peer_allowed_to_send", as per [RFC6520]. This advertisement is not a "MUST" in order to grandfather existing NETCONF over TLS implementations. For NETCONF over TLS Call Home, the network management system MUST advertise "peer_allowed_to_send" per [RFC6520]. This is a "MUST" so as to ensure devices can depend in it always being there for call home connections, which is conveniently when keep-alives are needed the most.
The YANG modules defined in this memo are designed to be accessed via the NETCONF protocol [RFC6241]. Authorization for access to specific portions of conceptual data and operations within this module is provided by the NETCONF access control model (NACM) [RFC6536].
There are a number of data nodes defined in the "ietf-netconf-server" and "ietf-system-tls-auth" YANG modules which are writable/creatable/deletable (i.e., config true, which is the default). These data nodes may be considered sensitive or vulnerable in some network environments. Write and read operations to these data nodes can have a negative effect on network operations. It is thus important to control write and read access to these data nodes. Below are the data nodes and their sensitivity/vulnerability.
ietf-netconf-server:
ietf-system-tls-auth:
This document registers two URIs in the IETF XML registry [RFC2119]. Following the format in [RFC3688], the following registrations are requested:
URI: urn:ietf:params:xml:ns:yang:ietf-netconf-server
Registrant Contact: The NETCONF WG of the IETF.
XML: N/A, the requested URI is an XML namespace.
URI: urn:ietf:params:xml:ns:yang:ietf-system-tle-auth
Registrant Contact: The NETCONF WG of the IETF.
XML: N/A, the requested URI is an XML namespace.
This document registers two YANG modules in the YANG Module Names registry [RFC6020].
name: ietf-netconf-server namespace: urn:ietf:params:xml:ns:yang:ietf-netconf-server prefix: ncserver reference: RFC XXXX name: ietf-system-tls-auth namespace: urn:ietf:params:xml:ns:yang:ietf-system-tls-auth prefix: sys-tls-auth reference: RFC XXXX
The YANG module define herein does not itself support virtual routing and forwarding (VRF). It is expected that external modules will augment in VRF designations when needed.
The authors would like to thank for following for lively discussions on list and in the halls (ordered by last name): Andy Bierman, Martin Bjorklund, Benoit Claise, David Lamparter, Alan Luchuk, Ladislav Lhotka, Radek Krejci, Tom Petch, and Phil Shafer.
Juergen Schoenwaelder and was partly funded by Flamingo, a Network of Excellence project (ICT-318488) supported by the European Commission under its Seventh Framework Programme.
| [RFC3688] | Mealling, M., "The IETF XML Registry", BCP 81, RFC 3688, January 2004. |
INSERT_TEXT_FROM_FILE(refs/ex-ssh.xml)
INSERT_TEXT_FROM_FILE(refs/ex-tls.xml)
Please see: https://github.com/netconf-wg/server-model/issues.