| Network Working Group | S. Bhandari |
| Internet-Draft | B. Fajalia |
| Intended status: Informational | R. Schmieder |
| Expires: January 16, 2014 | S. Orr |
| A. Dutta | |
| Cisco | |
| July 15, 2013 |
Extending multicast DNS across local links in Campus and Enterprise networks
draft-bhandari-dnssdext-mdns-gateway-00
This document describes the requirements for extending multicast DNS in enterprise networks. It provides an overview of a solution to extend multicast DNS services across links that has been implemented in routers, switches and wireless LAN controllers.
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in RFC 2119 [RFC2119].
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 January 16, 2014.
Copyright (c) 2013 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.
Service discovery using multicast DNS (mDNS) as defined in [RFC6762] is limited in scope to L3 boundaries due to the use of link-local scoped multicast addresses. Networks are partitioned into multiple segments by means of virtual local area networks (VLANs) or subnet creation for various reasons. The need for network wide, seamless service discovery demands the extension of the discovery protocol beyond the L3 boundary. There are also challenges in Wi-Fi networks when a large number of multicast messages are exchanged due to existing 802.11 multicast implementations.
Enabling Service Discovery across L3 boundaries can be accomplished in one of the following ways using existing, unmodified protocols:
As a first step, this draft lists out the approach to use a mDNS gateway on a network element (2) to extend the services across all the network segments attached to the element. While this approach does not preclude (1) or (3), it allows the extension of service discovery in a limited number of segments with minimal provisioning. Approach (4) is not explored further as it would add to the flood of service discovery messages in the scope defined by the multicast address and it would also require changes on mDNS clients, which is undesirable.
This section describes requirements for extending multicast DNS in an enterprise environment:
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in "Key words for use in RFCs to Indicate Requirement Levels" [RFC2119].
This document uses the multicast DNS and DNS terminology and conventions from [RFC6762] and [RFC6763]. It uses the same convention for services on the same link as defined in [I-D.cheshire-mdnsext-hybrid], repeated here for quick reference:
Multicast Multicast DNS works between a hosts on the same link. A set of hosts is considered to be "on the same link", if:
when any host A from that set sends a packet to any other host B in that set, using unicast, multicast, or broadcast, the entire link-layer packet payload arrives unmodified, and
a broadcast sent over that link by any host from that set of hosts can be received by every other host in that set
The link-layer *header* may be modified, such as in Token Ring Source Routing [802.5], but not the link-layer *payload*. In particular, if any device forwarding a packet modifies any part of the IP header or IP payload then the packet is no longer considered to be on the same link. This means that the packet may pass through devices such as repeaters, bridges, hubs or switches and still be considered to be on the same link for the purpose of this document, but not through a device such as an IP router that decrements the TTL or otherwise modifies the IP header.
The solution introduces a mDNS gateway function which is co-located on the network element that connects to multiple links, typically an IP router. The mDNS gateway function will be responsible for:
The mDNS gateway maintains a database of DNS Resource Records (RR) required to advertise and resolve services. At a minimum, the cache will contain PTR, SRV,TXT and A/AAAA RRs for each service with NSEC RR support for optimization. In addition, the link on which the service and host are seen is also remembered. This is required to refresh the records on TTL expiry.
A service filtering policy is configured with an action to permit or deny services into the cache or to filter services included in the response / advertisement messages based on matching criteria. The matching criteria can be defined based on:
A Service filtering policy is applied for incoming and outgoing messages. Each link can be applied with a unique filtering policy.
When a mDNS message is received by the mDNS gateway that matches the action set for the link, policy matching is executed. The incoming advertisement is processed against the mDNS gateway filtering policy applied in the incoming direction on the link where the advertisement is received. If the action is 'permit' the service is added to the cache. For a response or advertisement to be sent out, a filtering policy applied in the outgoing direction is processed and if the resulting action is 'deny' then the service and its corresponding RRs are not included in the message sent out.
The mDNS gateway listens for all service announcements. When a service announcement is received, the announcement and all the additional RRs learnt are added to the cache or ignored based on the result of the configured filter policy.
The RRs containing link-local information e.g. A or AAAA RRs that contain link-local scoped addresses are not stored in the cache.
When the mDNS gateway learns a service it can also forward the advertisement on other attached links.
The mDNS gateway processes all the queries against the configured filtering policy. If the response to the query is permitted then it constructs the answers and additional records required to resolve the service from its cache for the services that are permitted. Services that reside on the same link where the query is received are not included as the owner of the service will also see the query and would send the response directly. Only services learnt from different links are considered in the response.
Any query received for additional RRs to resolve the service e.g. query for SRV, A, AAAA etc are responded to in the same way. If the records do not exist in the cache due to expiry or purging of cache for any other reason, mDNS gateway sends out an explicit query to fetch the records on the link where the service resides.
According to [RFC6762] before registering a RR probing is performed to ensure unique names. As the mDNS gateway maintains cache of all the RRs that are extended across the links it responds to probe records like any other query. This will help in detecting and resolving conflicts across links for names that are extended across links.
When the mDNS gateway receives a service update or withdrawal it updates or removes the service and all corresponding records from its cache. It forwards the withdraw messages to other attached links.
The RRs describing the service and resolving it have a TTL that defines the validity of the RR. The mDNS gateway can continuously refresh each of the RRs in the cache as per the TTL rules. For the purpose of optimization, the mDNS gateway can rely on the host interested in the RRs to trigger a refresh by setting the TTLs in the response to the time remaining since the record was learnt by the mDNS gateway. If a client is interested in the RR then it would trigger a refresh when a fraction of the TTL is reached. While responding to queries from hosts, the mDNS gateway inturn sends out queries to refresh the records that are about to expire on the link where the records were learnt.
Deploying the mDNS gateway in wireless networks has a few additional requirements w.r.t to multicast radio optimization and mobility aspects. This section describes some additional capabilities added to the mDNS gateway to satisfy these requirements.
In order to conserve wireless bandwidth due to limitations of the multicast implementation in 802.11, the mDNS gateway sends service advertisements to wireless devices as L2 unicast messages.
In a wireless network, the mDNS gateway co-located on the network element that bridges the link-local multicast packets can act as a passive device and respond only if wireless clients sends a mDNS query. When bridging is turned off sending layer 2 unicast message is needed to on even when the provider is on the same link as the requestor. Bridging of link-local mDNS messages can be turned off based on configuration. This is useful in the following scenario:
mDNS functionality is not impacted by acting as a passive gateway because the client would always send the mDNS query when inquiring for a service.
Wireless clients are mobile in nature. The mDNS gateway should learn the service instance only from the authenticated wireless client. The mDNS gateway should tag each service instance from a wireless client with the client's MAC address. This MAC address should be used for device tracking. If the wireless client leaves the network, the mDNS gateway should not wait until the TTL expires but it should actively clean up the service instance provided by that wireless client. This is done to protect the mDNS gateway cache resources.
Wireless deployments supports seamless mobility. In such a scenario, the mDNS gateway needs to be aware of the client location. If the location changes, the mDNS gateway needs to update its mDNS cache. The mDNS gateway should tag each service instance with the device location. The device location can be derived based on the Access Point (AP) to which the wireless client is attached. If the client, which is providing any service, changes its location, this change needs to be reflected in the mDNS gateway. If the client roams from one mDNS gateway to another mDNS gateway, then the old gateway should provide the service instance information pertaining to the roamed client to the new gateway and then it must clear the mDNS cache for that particular client. If the mDNS gateway is not acting as a passive gateway, it may choose to update the network about the new service instance it has learnt.
All mDNS packets are sent to the multicast link-local IP address. When the mDNS gateway starts forwarding the mDNS advertisements across L3 boundaries then the number of such advertisement on any network would increase. 802.11 wireless networks operate in half duplex mode so any packet on air would need to contend for the wireless medium to transmit the packet. If there are many mDNS packets going on air then it would impact other data traffic. Hence mDNS traffic optimization is required.
When services are advertised, each record has an associated TTL value. When the TTL expires, the gateway needs to send a query (at 85%, 90% and 95% of the TTL) for that record to confirm its validity. If the TTL value of each record is different, then mDNS gateway needs to send a query for individual records. To minimize the mDNS traffic, queries for multiple RRs for that service record set can be initiated towards the source of the service. Such a query can be sent with the QU bit set as described in [RFC6762] to solicit a unicast response.
The mDNS gateway for wireless networks should act as a passive gateway as explained in Section 3.8.1. When it is acting as a passive gateway and bridging of mDNS packets is turned off it has to respond to queries on the link even when the provider of the service resides on the same link.
This section lists out limitations and challenges faced as part of the the solution described in this draft.
The solution documented here is limited to extending services in a across links attached to a single network element or mDNS gateways. For a broader application, the service discovery solution described in [I-D.cheshire-mdnsext-hybrid] should be realized with any provisioning as needed.
Similar to auto provisioning and realization of the hybrid proxy approach for homenet as described in [I-D.stenberg-homenet-dnssdext-hybrid-proxy-ospf] a solution needs to be built for enterprise and campus networks extending what has been described in this draft.
There are other considerations such as including the location information so that services can be ordered based on proximity of the service.
This document makes no request of IANA.
Note to RFC Editor: this section may be removed on publication as an RFC.
N/A
| [RFC2119] | Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, March 1997. |
| [RFC1035] | Mockapetris, P., "Domain names - implementation and specification", STD 13, RFC 1035, November 1987. |
| [RFC6762] | Cheshire, S. and M. Krochmal, "Multicast DNS", RFC 6762, February 2013. |
| [RFC6763] | Cheshire, S. and M. Krochmal, "DNS-Based Service Discovery", RFC 6763, February 2013. |
| [I-D.cheshire-mdnsext-hybrid] | Cheshire, S., "Hybrid Unicast/Multicast DNS-Based Service Discovery", Internet-Draft draft-cheshire-mdnsext-hybrid-02, July 2013. |
| [I-D.stenberg-homenet-dnssdext-hybrid-proxy-ospf] | Stenberg, M., "Hybrid Unicast/Multicast DNS-Based Service Discovery Auto-Configuration Using OSPFv3", Internet-Draft draft-stenberg-homenet-dnssdext-hybrid-proxy-ospf-00, June 2013. |
| [I-D.sekar-dns-llq] | Sekar, K., "DNS Long-Lived Queries", Internet-Draft draft-sekar-dns-llq-01, August 2006. |