Re: [amqp] Global addressing model - brain dump

[Alan Conway <aconway@redhat.com>]

This looks pretty sane to me. At first I found it odd to see container in the address but Qpid Dispatch router has a similar concept. It uses "magic" address prefixes to included router-ids in an address so e.g. amqp://host:port/$management is the management interface of the directly connected router, but amqp:/_topo/0/foo/$management lets you manage the router with id=foo. Using the standard container-id concept makes sense here - so in your scheme it would be

amqp://host:port/(foo)/$management. Equating router-id and container-id seems like a sensible notion.

On Fri, Sep 14, 2018 at 6:04 AM Clemens Vasters <clemensv@microsoft.com> wrote:

For the Management spec I also need the Addressing spec to make progress, so Iâm now taking a fresh look at this.

Â

From the F2Fs I recall that our biggest holdup was that weâve got tension between several distinct goals for addressing expressions:

1. An address _expression_ should enable a âclientâ to find a âserverâ on an IP network to establish an AMQP TLS-first (5671) or TLS-upgrade (5672) connection, establish link(s) and transfer messages. The address _expression_ therefore needs to hold DNS/IP/TCP addressing information, a protocol selector, and a link target.
2. An address _expression_ should enable a âsenderâ to have messages routed to a âreceiverâ inside a (yet poorly defined) AMQP network and fully abstracting the underlying IP layer, whereby the âreceiverâ address is a logical construct that doesnât contain network-related information. With such an address, it should be possible to find the receiving container (and node) by ways of:
   1. resolving the logical address to container name/location OR
   2. evaluating routing rules based on filter expressions that indirectly yield a container location
3. An address _expression_ should enable a âsenderâ to route messages to a âreceiverâ inside a singular AMQP container, where address expressions correspond directly to node names inside of said container.

Â

We also have amqp/amqps URIs being used in the wild based on the current draft and I donât want to break those.

Â

Hereâs a proposal for an extended URI format that aims to hit all the goals at once:

Â

{ amqp | amqps } : [ //user:key@ip-endpoint/ ] [ (container-name)/ ] [ [$] node-path ]

Â

Iâll start the explanation with examples, because theyâll probably tell a whole lot of the story already:

• amqps://broker.example.com/queue1 â DNS qualified address of a TCP/IP endpoint of an AMQP container with path for the âqueue1â AMQP node
• amqp:(london-branch-34)/atm â Logical AMQP network address of âlondon-branch-34â with path for the âatmâ AMQP node
• amqps://megabank.example.com/(london-branch-34)/atm â DNS qualified address of a âgatewayâ into an AMQP network, with logical AMQP network address as suffix
• amqp:()/control â Anonymous AMQP network address to be resolved into a container location by evaluating message metadata, with path for the âcontrolâ node on the resulting container
• somewhere â Local container address of the âsomethingâ AMQP node
• $somewhere â AMQP reserved local container address

Â

To start, Iâm splitting up the AMQP notion of container into two distinct concepts: âgatewayâ and âcontainerâ.

• A gateway is a TCP/IP listener that speaks the AMQP wire protocol. Clients establish connections and sessions with gateways. Gateways are onramps into AMQP networks.
• A container is a host for AMQP nodes. Multiple containers may be interrelated (and opportunistically interconnected) to form a logical network, with lookup mechanisms that allow resolving logical names to specific containers.
• Each container MUST have a gateway. Each configured relationship between containers holds information about which gateway ought to be used to reach the respective container. The configured destination gateway for any container MAY, however, be an intermediary, meaning it might take one or more further network resolution/routing steps to reach the destination container.
• Not all gateways have to be containers. A gateway can just do connection/session handling and never host nodes and/or take ownership of links; it might be a pure relay bridging sessions to another gateway.

Â

With the URI format, I am proposing a three-layer addressing model:

• With DNS/IP addressing in the ip-endpoint section, you pick an AMQP endpoint as gateway into an AMQP network.
• With container-name, you pick the AMQP container you want to communicate with via the chosen gateway. If omitted, the container is thought to be directly associated with âthisâ gateway (like today). If the container-name _expression_ is present but empty, itâs up to the gateway rules/configuration to determine the destination container based on the message/link properties or other criteria.
• With node-path, you Âpick the AMQP node you want to route/link to inside the chosen/resolved container.

Â

Â

With a message addressed to amqps://megabank.example.com/(london-branch-34)/atm, a client first resolves the DNS name and IP address to some AMQP gateway instance via DNS and/or a load balancer, connects, and establishes a session â or reuses an existing session to that gateway.

Â

Then, the client will either attempt to attach a link to (london-branch-34)/atm on the gateway-associated container, or identify an already existing link. Alternatively, it might attempt to route to said address via the containerâs anonymous terminus by using it in the âproperties:toâ property.

Â

How the (container-name) destination gets resolved is up to the container. In a simple case, a container might maintain a routing table with {container-name, gateway ip-endpoint} pairs, and will create/reuse connections/sessions to the listed gateways for routing/linking to the respective containers. The routing table may point to intermediary gateways. For link routing, the â(container-name)/node-pathâ _expression_ serves as the link target name in all containers acting as intermediaries. It resolves to node-path on the container that self-identifies as the given container-name.

Â

In effect, you can attach a link to â(container-name)/node-pathâ on any container on a network supporting global addressing, and the container will then patch that link through to the next hop network destination configured for that container name, and until it gets to the target container. For anonymous terminus routing, messages get routed along to the next containerâs anonymous terminus. The ârouting cookiesâ spec then applies for how to establish reply routes.

Â

Â

Thoughts?