Help us improve your experience.

Let us know what you think.

Do you have time for a two-minute survey?

 
 

Service Chain Route Reorigination

Overview: Service Chaining in Contrail

In Contrail, the service chaining feature allows the operator to insert dynamic services to control the traffic between two virtual networks. The service chaining works on a basic rule of next-hop stitching.

In Figure 1, the service chain is inserted between the Left VN and the Right VN. The service chain contains one or more service instances to achieve a required network policy.

In the example, the route for the VM in the Right VN is added to the routing table for the Left VN, with the next hop modified to ensure that the traffic is sent by means of the left interface of the service chain. This is an example of route reorigination.

Figure 1: Route ReoriginationRoute Reorigination

Using reorigination of routes for service chaining (for example, putting the route for the right network in the left routing table) requires the following features:

  • Route aggregation

    For scaling purposes, it is useful to publish an aggregated route as the service chain route, rather than publishing every route of each VM (/32). This reduces the memory footprint for the route table in the gateway router and also reduces route exchanges between control nodes and the gateway router. The route can be aggregated to the default route (0/0), to the VN subnet prefix, or to any arbitrary route prefix.

  • Path attribute modification for reoriginated routes

    There are cases where the BgpPath attribute for the service chain route needs to be modified. An example is the case of service chain failover, in which there are two service chains with identical services that are connected between the same two VNs. The operator needs to control which service chain is used for traffic between two networks, in addition to ensuring redundancy and high availability by providing failover support. Path attribute modification for reoriginated routes is implemented by means of routing policy, by providing an option to alter the MED (multi-exit discriminator) or local-pref of the reoriginated service chain route.

  • Control to enable and disable reorigination of the route

    In some scenarios, the operator needs a control to stop reorigination of the route as the service chain route, for example, when static routes are configured on service VM interfaces. Control to enable or disable reorigination of the route is implemented by tagging the routes with the no-reoriginate community. Routes with the no-reoriginate community tag are skipped for route reorigination.

Route Aggregation

The route aggregation configuration object contains a list of prefixes to aggregate. The next-hop field in the route aggregate object contains the address of the route whose next hop is stitched as a next hop of the aggregate route.

Route aggregation is configured on the service instance. The operator can attach multiple route aggregation objects to a service instance. For example, if routes from the Right VN need to be aggregated and reoriginated in the route table of the Left VN, the route aggregate object is created with a prefix of the Right VN’s subnet prefix and attached to the left interface of the service instance.

If the service chain has multiple service instances, the route aggregate object is attached to the left interface of the left-most service instance and to the right interface of the right-most service instance.

The relationships are shown in Figure 2.

Figure 2: Route Aggregate RelationshipsRoute Aggregate Relationships

The schema transformer sets the next-hop field of the route aggregate object to the service chain interface address. The schema transformer also links the route aggregate object to the internal routing instance created for the service instance.

Using the configuration as described, the Contrail control service reads the route aggregation object on the routing instance. When the first, more specific route or contributing route is launched (when the first VM is launched on the right VN), the aggregate route is published. Similarly, the aggregated route is deleted when the last, more specific route or contributing route is deleted (when the last VM is deleted in the right VN). The aggregated route is published when the next hop for the aggregated route gets resolved.

By default, in BGP or XMPP route exchanges, the control node will not publish contributing routes of an aggregate route.

Schema for Route Aggregation

Route Aggregate Object

The following is the schema for route aggregate objects. Multiple prefixes can be specified in a single route aggregate object.

Service Instance Link to Route Aggregate Object

The following is the schema for the service instance link to route aggregation objects. The operator can link multiple route aggregate objects to a single service interface.

Routing Instance Link to Route Aggregate Object

The following is the schema for the routing instance link to the route aggregation object. A routing instance can be linked to multiple route aggregate objects to perform route aggregation for multiple route prefixes.

Configuring and Troubleshooting Route Aggregation

Configure Route Aggregate Object

You can use the Contrail UI, Configure > Networking > Routing > Create >Route Aggregate screen to name the route aggregate object and identify the routes to aggregate. See Figure 3.

Figure 3: Create Route AggregateCreate Route Aggregate
Example VNC Script to Create a Route Aggregate Object

You can use a VNC script to create a route aggregate object, as in the following example:

Configuring a Service Instance

Create a service instance with the route aggregate object linked to the aggregate left network subnet prefix in the right virtual network. See the example in Figure 4.

Figure 4: Create Service InstanceCreate Service Instance

Create a Virtual Network and Network Policy

Create a left and right virtual network with the subnets 1.1.1/24 and 2.2.2/24, respectively. Create a network policy to apply a service chain between the left VN and the right VN. See the following example.

Attach the network policy to create the service chain between the left and right VNs. See the following example.

Validate the Route Aggregate Object in the API Server

Validate the route aggregate object in the API server configuration database. Verify the routing instance reference and the service instance reference for the aggregate object. The aggregate_route_nexthop field in the route aggregate object is initialized by the schema transformer to the service chain address. See the following example.

Validate the Route Aggregate Object in the Control Node

Validate the instance configurations of the route aggregate by checking the control node introspect for the service instance internal routing instance. For example:

http://<control-node>:8083/Snh_ShowBgpInstanceConfigReq?search_string=default- domain:admin:right:service-ace7ae00-56e3-42d1-96ec-7fe77088d97f-default- domain_admin_si-aggregate

See the following example.

To check the state of the route aggregate object on the control node, point your browser to:

http://<control-node>:8083/Snh_ShowRouteAggregateReq

See the following example.

You can also check the route table for the aggregate route in the right VN BGP able. For example:

http://<control-node>:8083/Snh_ShowRouteReq?x=default-domain:admin:right:right.inet.0

See the following example.

Routing Policy

Contrail uses routing policy infrastructure to manipulate the route and path attribute dynamically. Contrail also supports attaching the import routing policy on the service instances.

The routing policy contains list terms. A term can be a terminal rule, meaning that upon a match on the specified term, no further terms are evaluated and the route is dropped or accepted, based on the action in that term.

If the term is not a terminal rule, subsequent terms are evaluated for the given route.

The list terms are structured as in the following example.

The matches and actions of the policy term lists operate similarly to the Junos language match and actions operations. A visual representation is the following.

Each term is represented as in the following:

The term should not contain an any match condition, for example, an empty from should not be present.

If an any match condition is present, all routes are considered as matching the term.

However, the then condition can be empty or the action can be unspecified.

Applying Routing Policy

The routing policy evaluation has the following key points:

  • If the term of a routing policy consists of multiple match conditions, a route must satisfy all match conditions to apply the action specified in the term.

  • If a term in the policy does not specify a match condition, all routes are evaluated against the match.

  • If a match occurs but the policy does not specify an accept, reject, or next term action, one of the following occurs:

    • The next term, if present, is evaluated.

    • If no other terms are present, the next policy is evaluated.

    • If no other policies are present, the route is accepted. The default routing policy action is “accept”.

  • If a match does not occur with a term in a policy, and subsequent terms in the same policy exist, the next term is evaluated.

  • If a match does not occur with any terms in a policy, and subsequent policies exist, the next policy is evaluated.

  • If a match does not occur by the end of a policy or all policies, the route is accepted.

A routing policy can consist of multiple terms. Each term consists of match conditions and actions to apply to matching routes.

Each route is evaluated against the policy as follows:

  1. The route is evaluated against the first term. If it matches, the specified action is taken. If the action is to accept or reject the route, that action is taken and the evaluation of the route ends. If the next term action is specified or if no action is specified, or if the route does not match, the evaluation continues as described above to subsequent terms.

  2. Upon hitting the last non-terminal term of the given routing policy, the route is evaluated against the next policy, if present, in the same manner as described in step 1.

Match Condition: From

The match condition from contains a list of match conditions to be satisfied for applying the action specified in the term. It is possible that the term doesn’t have any match condition. This indicates that all routes match this term and action is applied according to the action specified in the term.

The following table describes the match conditions supported by Contrail.

Match Condition

User Input

Description

Prefix

List of prefixes to match

Each prefix in the list is represented as prefix and match type, where the prefix match type can be:

  • exact

  • orlonger

  • longer

Example: 1.1.0.0/16 orlonger

A route matches this condition if its prefix matches any of the prefixes in the list.

Community

Community string to match

Represented as either a well-known community string with no export or no reoriginate, or a string representation of a community (64512:11).

Protocol

Array of path source or path protocol to match

BGP | XMPP | StaticRoute | ServiceChain | Aggregate. A path is considered as matching this condition if the path protocol is one of protocols in the list.

Routing Policy Action and Update Action

The policy action contains two parts, action and update action.

The following table describes action as supported by Contrail.

Action

Terminal?

Description

Reject

Yes

Reject the route that matches this term. No more terms are evaluated after hitting this term.

Accept

Yes

Accept the route that matches this term. No more terms are evaluated after hitting this term. The route is updated using the update specified in the policy action.

Next Term

No

This is the default action taken upon matching the policy term. The route is updated according to the update specified in the policy action. Next terms present in the routing policy are processed on the route. If there are no more terms in the policy, the next routing policy is processed, if present.

The update action section specifies the route modification to be performed on the matching route.

The following table describes update action as supported by Contrail.

Update Action

User Input

Description

community

List of community

As part of the policy update, the following actions can be taken for community:

  • Add a list of community to the existing community.

  • Set a list of community.

  • Remove a list of community (if present) from the existing community.

MED

Update the MED of the BgpPath

Unsigned integer representing the MED

local-pref

Update the local-pref of the BgpPath

Unsigned integer representing local-pref

Routing Policy Configuration

Routing policy is configured on the service instance. Multiple routing policies can be attached to a single service instance interface.

When the policy is applied on the left interface, the policy is evaluated for all the routes that are reoriginated in the left VN for routes belonging to the right VN. Similarly, the routing policy attached to the right interface influences the route reorigination in the right VN, for routes belonging to the left VN.

The following figure illustrates a routing policy configuration.

The policy sequence number specified in the routing policy link data determines the order in which the routing policy is evaluated. The routing policy link data on the service instance also specifies whether the policy needs to be applied to the left service interface, to the right service interface, or to both interfaces.

It is possible to attach the same routing policy to both the left and right interfaces for a service instance, in a different order of policy evaluation. Consequently, the routing policy link data contains the sequence number for policy evaluation separately for the left and right interfaces.

The schema transformer links the routing policy object to the internal routing instance created for the service instance. The transformer also copies the routing policy link data to ensure the same policy order.

Configuring and Troubleshooting Routing Policy

This section shows how to create a routing policy for service chains and how to validate the policy.

Create Routing Policy

First, create the routing policy, Configure > Networking > Routing > Create >Routing Policy. See the following example.

Configure Service Instance

Create a service instance and attach the routing policy to both the left and right interfaces. The order of the policy is calculated by the UI, based on the order of the policy specified in the list.

Configure the Network Policy for the Service Chain

At Edit Policy, create a policy for the service chain, see the following example.

Using a VNC Script to Create Routing Policy

The following example shows use of a VNC API script to create a routing policy.

Verify Routing Policy in API Server

You can verify the service instance references and the routing instance references for the routing policy by looking in the API server configuration database. See the following example.

Verify Routing Policy in the Control Node

You can verify the routing policy in the control node.

Point your browser to:

http://<control-node>:8083/Snh_ShowRoutingPolicyReq?search_string=failover

See the following example.

Verify Routing Policy Configuration in the Control Node

You can verify the routing policy configuration in the control node.

Point your browser to:

http://<control-node>:8083/Snh_ShowBgpRoutingPolicyConfigReq?search_string=failover

See the following example.

Verify Routing Policy Configuration on the Routing Instance

You can verify the routing policy configuration on the internal routing instance.

Point your browser to:

http://<control-node>:8083/Snh_ShowBgpInstanceConfigReq?search_string=<name-of-internal-vrf>

See the following example.

You can also verify the routing policy on the routing instance operational object.

Point your browser to:

http://<control-node>:8083/Snh_ShowRoutingInstanceReq?x=<name-of-internal-vrf>

See the following example.

Control for Route Reorigination

The ability to prevent reorigination of interface static routes is typically required when routes are configured on an interface that belongs to a service VM.

As an example, the following image shows a service chain that has multiple service instances, with an in-net-nat service instance as the last service VM, also with the right VN as the public VN.

The last service instance performs NAT by using a NAT pool. The right interface of the service VM must be configured with an interface static route for the NAT pool so that the destination in the right VN knows how to reach addresses in the NAT pool. However, the NAT pool prefix should not be reoriginated into the left VN.

To prevent route reorigination, the interface static route is tagged with a well-known BGP community called no-reoriginate.

When the control node is reoriginating the route, it skips the routes that are tagged with the BGP community.

Configuring and Troubleshooting Reorigination Control

The community attribute on the static routes for the interface static route of the service instance is specified during creation of the service instance. See the following example.

Use the following example to verify that the service instance configuration object in the API server has the correct community set for the static route. See the following example.