This project creates virtual infrastructure with Terraform on public cloud, which will be used for Cisco SD-WAN Cloud onRamp for Multi-cloud lab. Some modules can be used separate from each other. For example Centralized Firewall Design or Multi-Region Fabric use cases can be deployed totally separated.
If you are new to Cisco Cloud onRamp, please visit Cisco Online Documentation and do the Sandbox lab first. The Sandbox offers a possibility to learn the basics and configuration workflow by following step-by-step configuration guide. You do not need any cloud accounts or prerequisites, the whole setup will be brought up for you automatically and you have full access for 2 hours free of charge.
If you want to create your own setup and test specific use cases, this project helps you and automates creation of two virtual branches and two cloud-based apps. You can use this infrastructure to test site-to-site and site-to-cloud uses cases, which Cisco Cloud onRamp offers. The following diagram shows topology, which will be created.
As cloud- and branch-based apps we use a simple web server running on different ports on Linux VM. By creating an SD-WAN policy, you can route traffic via public internet or CSP backbone. An example for such traffic steering policy will be given in the SD-WAN part based on TCP ports 8001-8004.
Both virtual branches will be implemented with Cisco CSR 1000v virtual routers. Branch 1 has also a WAN emulator, which can introduce configurable loss, latency and delay on public internet path. This is implemented using Linux Traffic Control (tc) capabilities.
Amazon Web Services (AWS) is used as Cloud Service Provider (CSP) to host both virtual branches. This can easily be changed to Azure or GCP by changing Terraform Providers and adjusting the code. CSP choice for branch hosting is not relevant for Cloud onRamp functionality and tests.
Centralized Firewall Inspection with SD-WAN is a very common design topic. It is implemented in the chapter 05. This section can be used standalone if you are interested only in this topic. The implemented solution allows scalable north-south, east-west traffic inspection using Cisco FTDv virtual firewalls and AWS Gateway Load Balancer (GWLB) and shown below.
Cisco SD-WAN Multi-Region Fabric (MRF) provides the ability to segment the SD-WAN overlay into multiple regional networks that operate distinctly from one another, and a central core-region network for managing inter-regional traffic. Many customers have at least two Cloud Service Providers (CSP) and would like to use CSP backbone for site-to-site communication between different regions. In a flat (non MRF) design without any segmentation a control policy for traffic steering between different sites can be complex and not easy to create. It does not scale well with the number of sites. MRFsolves this problem by splitting the network into different regions. Second problem beside complexity is the ask to isolate specific subregions within one particular region. For example, in emergency case all branches within Northern California must be isolated from the rest of the network with one click of a button.
Table of Content
Before you begin, please make sure, that the following mandatory criteria are met:
- a computer with installed Terraform environment including CSP provider and configuration.
- you have an account and can use at least one of the following CSP: AWS, Azure or GCP
- you have an existing SD-WAN network with vManage, vBond and vSmart (SD-WAN Controllers) running the following software versions:
CSP Controller AWS 20.9 Azure 20.9 GCP 20.9 - 4 SD-WAN licenses for Cisco Catalyst 8000v (c8kv) virtual routers acting as cloud gateways. In each CSP region, Cloud onRamp Automation will spin up two c8kv instances.
- 2 SD-WAN licenses for Cisco CSR1000v virtual routers acting as branch routers.
For example if you want to test Cloud onRamp with Azure vWAN, you will need at least 20.4 software running on all SD-WAN controllers.
You can still have virtual branches hosted on a different CSP, but if you want to test Cloud onRamp automation workflow for Azure, you need at least 20.4 controller software.
If you do not have SD-WAN setup at all, you will need to create one. Please work with your local Cisco contacts and ask for Proof of Concept (PoC) setup.
Please make sure, that your CSP account has enough ressources to spin up needed infrastructure. In AWS case, please double check the following:
- you can create enough basic elements like VPCs, Subnets, IGWs, TGW, etc.
- you have enough elastic IPs: calculate with two Elastic IPs per VM (one for out-of-band management and one for internet access) to be conservative
- you have ssh key registered in appropriate regions
- Terraform v1.3.7, AWS Provider v4.49.0, Google Provider v4.47.0
- SD-WAN Controllers: 20.10.0
- CSR1000v on virtual branches: 17.03.02prd9
- C8kv as cloud gateways: 17.09.01, 17.10 for MRF with Subregions Use Case
- Linux for apps and WAN-Emulator: Amazon Linux 2 AMI, kernel version 4.14.219-164.354.amzn2.x86_64
Before you can start any tests with Cisco Cloud onRamp for Multicloud, you will need to have running SD-WAN controllers: vManage, vBond and vSmart. Cisco supports two different operations options:
- on-prem: controllers running in a Data Center for example on ESXi
- cloud-based: Cisco or MSP cloudops team will create a set of controllers for you.
This lab repo provides Terraform scripts, which will spin up a basic controller environment only if you will have private controllers AMIs shared with you by Cisco. Please note, that SD-WAN controller images are currently not available on marketplace. Please work with your Cisco Account team if your project requires private AMIs. This can be done for testing only special cases and currently cannot be used for production. The main reason is performance and stability considerations - Cisco performs intensive tests with ESXi and for Cloudops-operated controllers, defines strict requirements for Hypervisors and underlying hardware.
The included scripts will create new basic controller installation. Once vManage, vSmart and vBond are up and running, you will need to provide basic configuration: org-id, system IPs, etc. Also if own PKI infrastructure is used, you will need to install own certificates. Please note, that the current version of scripts is not using cloud-init files like others sections. Please refer to the branch scripts if you need an example for cloud-init Day 0 configuration, which is used during the first VM boot.
This section describes creation of a virtual branch consisting of one SD-WAN virtual router (CSR1000v) and Linux App. Branch 1 also has virtual WAN emulator, which can introduce configurable loss and latency to the public internet connection.
Step 1: ssh pem file
Make sure, that your have ssh pem file in the same folder with terraform files for the virtual branch.
In the example below you see "aws-key-20-3-setup.pem" file, which will be used for ssh connectivity to the virtual routers.
Please note, that you need to copy your pem file to the Terraform folder or provide a path to it. Otherwise, you will be not able to create VMs.
aws-key-20-3-setup.pem <-- pem ssh key for the appropriate AWS region
branch1.tf <-- Main Terraform script
cloud-init-branch1-r1.user_data <-- Day-0 SD-WAN CSR1000v configuration generated by vManage
provider.tf <-- Provider file with AWS region definition
vars.tf <-- File with all variables, which are used in the main Terraform script
Step 2: Day-0 config for SD-WAN CSR1000v
In Cisco vManage under configuration - devices, select free CSR100v and generate cloud-init file.
Please refer to the following Cisco Online Documentation for details about cloud-init file.
In the example for cloud-init file shown below we use the following NIC allocation:
- 1st NIC (GigabitEthernet1) for out-of-band management aka VPN512 or Mgmt-intf vrf
- 2nd NIC (GigabitEthernet2) for SD-WAN over public-internet aka VPN0
- 3rd NIC for "LAN" connectivity to the Branch App (Linux VM) aka service vpn or vrv 10.
You can create the router configuration using vManage Feature and Device Templates, CLI template or simply generate the cloud-init file without vManage-generated config and add the lines shown below to the cloud-init file. In this case the virtual router will come up, join the SD-WAN fabric and remain in CLI mode. Please refer to this Cisco Online Documentation for more details on Device Templates.
Content-Type: multipart/mixed; boundary="==BOUNDARY=="
MIME-Version: 1.0
--==BOUNDARY==
Content-Type: text/cloud-config; charset="us-ascii"
#cloud-config
vinitparam:
- uuid : CSR-BBD516D8-XXXX-XXXX-XXXX-4FA9CB62259D
- org : Demo-npitaev
- vbond : 44.XXX.XXX.68
- otp : xxxx50b93df741dbb3719489e98axxxx
--==BOUNDARY==
Content-Type: text/cloud-boothook; charset="us-ascii"
#cloud-boothook
hostname Branch1-R1
!
system
system-ip 10.111.1.11
site-id 111
organization-name Demo-npitaev
vbond 44.XXX.XXX.68
!
!
vrf definition 10
rd 1:10
address-family ipv4
route-target export 64550:1
route-target import 64550:1
exit-address-family
!
address-family ipv6
exit-address-family
!
!
vrf definition Mgmt-intf
description Management
rd 1:512
address-family ipv4
route-target export 1:512
route-target import 1:512
exit-address-family
!
address-family ipv6
exit-address-family
!
!
interface GigabitEthernet1
no shutdown
vrf forwarding Mgmt-intf
ip address dhcp client-id GigabitEthernet1
ip dhcp client default-router distance 1
ip mtu 1500
mtu 1500
negotiation auto
exit
!
interface GigabitEthernet2
no shut
ip address dhcp client-id GigabitEthernet2
ip dhcp client default-router distance 1
ip mtu 1500
mtu 1500
negotiation auto
!
!
interface GigabitEthernet3
no shut
!
!
interface Tunnel2
no shutdown
ip unnumbered GigabitEthernet2
no ip redirects
ipv6 unnumbered GigabitEthernet2
no ipv6 redirects
tunnel source GigabitEthernet2
tunnel mode sdwan
exit
!
!
sdwan
interface GigabitEthernet2
tunnel-interface
encapsulation ipsec weight 1
no border
color default
no last-resort-circuit
no low-bandwidth-link
no vbond-as-stun-server
vmanage-connection-preference 5
port-hop
carrier default
nat-refresh-interval 5
hello-interval 1000
hello-tolerance 12
allow-service all
no allow-service bgp
allow-service dhcp
allow-service dns
allow-service icmp
allow-service sshd
allow-service netconf
allow-service ntp
no allow-service ospf
no allow-service stun
allow-service https
no allow-service snmp
exit
exit
appqoe
no tcpopt enable
!
omp
no shutdown
send-path-limit 4
ecmp-limit 4
graceful-restart
no as-dot-notation
timers
holdtime 60
advertisement-interval 1
graceful-restart-timer 43200
eor-timer 300
exit
address-family ipv4
advertise bgp
advertise connected
advertise static
!
address-family ipv6
advertise bgp
advertise connected
advertise static
!
!
!
!
--==BOUNDARY==
Step 3: configure Terraform variables in the vars.tf file
Please review and change ALL variables defined in vars.tf.
The default AWS region fro Branch1 is us-west-2. If you want to change it, please make sure to change the AMIs for CSR1000v and Linux.
You can follow this HowTo if you need to find out the appropriate AMI ID from AWS Marketplace for CSR1000v and AMI Linux.
Also make sure to use the right ssh pem file, which is available in the appropriate region.
Step 4: test your Terraform files
Once you configured all variables in the vars.tf file and double checked the region in the provider.tf file, please initiate and test all before deployment.
~/terraform-cisco-sdwan-cor-lab/01-Branch1# terraform init
Initializing the backend...
Initializing provider plugins...
- Finding latest version of hashicorp/aws...
- Installing hashicorp/aws v3.34.0...
- Installed hashicorp/aws v3.34.0 (signed by HashiCorp)
Terraform has created a lock file .terraform.lock.hcl to record the provider
selections it made above. Include this file in your version control repository
so that Terraform can guarantee to make the same selections by default when
you run "terraform init" in the future.
Terraform has been successfully initialized!
You may now begin working with Terraform. Try running "terraform plan" to see
any changes that are required for your infrastructure. All Terraform commands
should now work.
If you ever set or change modules or backend configuration for Terraform,
rerun this command to reinitialize your working directory. If you forget, other
commands will detect it and remind you to do so if necessary.
~/terraform-cisco-sdwan-cor-lab/01-Branch1#
~/terraform-cisco-sdwan-cor-lab/01-Branch1#
~/terraform-cisco-sdwan-cor-lab/01-Branch1#
~/terraform-cisco-sdwan-cor-lab/01-Branch1#
~/terraform-cisco-sdwan-cor-lab/01-Branch1# terraform plan
provider.aws.region
The region where AWS operations will take place. Examples
are us-east-1, us-west-2, etc.
Enter a value: us-west-2 <-- enter your region
... output omitted for brevity ...
Plan: 32 to add, 0 to change, 0 to destroy. <-- check the output, scroll up and validate ressources to be created
~/terraform-cisco-sdwan-cor-lab/01-Branch1#
Step 5: deploy your Terraform files
Once you verified the output from "terraform plan", you can now deploy the virtual branch by executing "terraform apply"
~/Downloads/terraform-cisco-sdwan-cor-lab/01-Branch1# terraform apply
provider.aws.region
The region where AWS operations will take place. Examples
are us-east-1, us-west-2, etc.
Enter a value: us-west-2 <-- enter your region
... output omitted for brevity ...
Plan: 32 to add, 0 to change, 0 to destroy.
Do you want to perform these actions?
Terraform will perform the actions described above.
Only 'yes' will be accepted to approve.
Enter a value: yes <-- confirm by typing yes
... output omitted for brevity ...
After few minutes your virtual branch will be ready. In case of any problems, for example if you will run into some AWS limits like not enough elastic IPs, you can increase the AWS limits and repeat "terraform apply". Terraform will add needed ressources and not touch already existing items, which do not need modifications.
The WAN Emulator Linux VM will be able to imair public internet link and introduce loss and latency using Network Address Translation (NAT) and Linux Traffic Control (TC) capabilities. This section describes NAT, Firewall and TC cionfiguration.
WAN Emulator has 3 NICs:
- eth0 for out-of-band management
- eth1 pointing to CSR1000v
- eth2 pointing to public internet.
Preparation:
- make sure, that the WAN Emulator VM was successfully deployed and can be accessed via ssh.
- AWS source and destination check should be disabled on all NICs. Please double check it in AWS EC2 console.
- Check the AWS route tables and make sure, that the routing is correct.
Configuring Firewall and NAT:
- Install firewall: sudo yum install firewalld
- Start firewall: sudo systemctl start firewalld
- Check and enable IP forwarding:
sudo sysctl net.ipv4.ip_forward
sudo sysctl -w net.ipv4.ip_forward=1
sudo sysctl net.ipv4.ip_forward - Configure NAT and define internal and external zones:
sudo firewall-cmd --permanent --direct --passthrough ipv4 -t nat -I POSTROUTING -o eth2 -j MASQUERADE -s 10.201.2.0/24
sudo firewall-cmd --zone=external --add-masquerade --permanent
sudo firewall-cmd --zone=external --add-interface=eth2 --permanent
sudo firewall-cmd --zone=internal --add-interface=eth1 --permanent
sudo firewall-cmd --complete-reload
sudo firewall-cmd --list-all-zones - Reboot: sudo reboot
Please check additional documentation on NAT and Firewall described in this article.
Check FW status:
[ec2-user@linux-tc-for-gcp ~]$ sudo systemctl status firewalld
firewalld.service - firewalld - dynamic firewall daemon
Loaded: loaded (/usr/lib/systemd/system/firewalld.service; enabled; vendor preset: enabled)
Active: active (running) since Thu 2020-03-19 17:58:32 UTC; 7min ago
Docs: man:firewalld(1)
Main PID: 2020 (firewalld)
CGroup: /system.slice/firewalld.service
2020 /usr/bin/python -Es /usr/sbin/firewalld --nofork --nopid
Mar 19 17:58:32 linux-tc-for-gcp systemd[1]: Starting firewalld - dynamic firewall daemon...
Mar 19 17:58:32 linux-tc-for-gcp systemd[1]: Started firewalld - dynamic firewall daemon.
[ec2-user@linux-tc-for-gcp ~]$
Check IP forwarding:
[ec2-user@ip-10-13-1-10 ~]$ sudo sysctl net.ipv4.ip_forward
net.ipv4.ip_forward = 1
[ec2-user@ip-10-13-1-10 ~]$
You may want to adjust route table on WAN Emulator and make sure, that is routes packets via needed NIC. Check the route table with ip ro and change it as needed. The following example installs a route for the out-of-band management via 1st NIC (eth0( and then leaves only one default route via NIC3 (eth2).
sudo route add -net 128.107.0.0/16 gw 10.111.1.1 <-- Allowed CIDR for the out of band management interface
sudo route del default gw 10.111.1.1
sudo route del default gw 10.111.2.1
sudo route del default gw 10.111.4.1
sudo route add default gw 10.111.4.1 <-- Public internet access only via eth2
The following example shows the full interface configuration and route table for the working WAN Emulator in Branch1
[ec2-user@ip-10-111-1-10 ~]$ ifconfig
eth0: flags=4163<UP,BROADCAST,RUNNING,MULTICAST> mtu 9001 <-- Out of band management interface
inet 10.111.1.10 netmask 255.255.255.0 broadcast 10.111.1.255
...
eth1: flags=4163<UP,BROADCAST,RUNNING,MULTICAST> mtu 9001
inet 10.111.2.10 netmask 255.255.255.0 broadcast 10.111.2.255 <-- NIC pointing to CSR1000v
...
eth2: flags=4163<UP,BROADCAST,RUNNING,MULTICAST> mtu 9001 <-- NIC pointing to public internet via IGW
inet 10.111.4.10 netmask 255.255.255.0 broadcast 10.111.4.255
...
[ec2-user@ip-10-111-1-10 ~]$
[ec2-user@ip-10-111-1-10 ~]$ ip ro
default via 10.111.4.1 dev eth2
10.111.1.0/24 dev eth0 proto kernel scope link src 10.111.1.10
10.111.2.0/24 dev eth1 proto kernel scope link src 10.111.2.10
10.111.4.0/24 dev eth2 proto kernel scope link src 10.111.4.10
128.107.0.0/16 via 10.111.1.1 dev eth0
169.254.169.254 dev eth0
[ec2-user@ip-10-111-1-10 ~]$
Once you have verified NAT and reachability, you can configure delay and/or loss using Linux Traffic Control capabilities. This article describes basic functionality and provides additional details. You can install tc using the following commands sudo yum install tc and sudo yum install iproute-tc. Then you can enable delay of 200 ms like shown in this example:
[ec2-user@linux-tc-for-gcp ~]$ sudo tc qdisc add dev eth1 root netem delay 200ms
[ec2-user@linux-tc-for-gcp ~]$
Check tc and delays on the appropriate interface:
[ec2-user@linux-tc-for-gcp ~]$ sudo tc -s qdisc ls dev eth1
qdisc netem 8001: root refcnt 3 limit 1000 delay 200.0ms
Sent 1240 bytes 18 pkt (dropped 0, overlimits 0 requeues 0)
backlog 0b 0p requeues 0
[ec2-user@linux-tc-for-gcp ~]$
You can also match on a specific IP and introduce delays only for it as described in this article. The following example introduces latency of 400ms for all packets sent to 54.112.112.125 (public IP of the branch2):
sudo tc qdisc add dev eth2 root handle 1: prio
sudo tc filter add dev eth2 protocol all parent 1: prio 1 u32 match ip dst 54.253.240.125/32 flowid 1:1
sudo tc filter add dev eth2 protocol all parent 1: prio 2 u32 match ip dst 0.0.0.0/0 flowid 1:2
sudo tc qdisc add dev eth2 parent 1:1 handle 10: netem delay 400ms
sudo tc qdisc add dev eth2 parent 1:2 handle 20: sfq
In order to enable the direct communication between two branches over public internet, you need to find out public IP addresses and adjust security groups, which were created during the initial deployment. You can locate the public IP address in AWS EC2 Management Console and then adjust the following security group lines in the branch1.tf and branch2.tf files. The example below shows just one side.
//Allow all from Branch1 Public IP WANEM - IP address must be changed after initial deployment
ingress {
from_port = 0
to_port = 0
protocol = "-1"
cidr_blocks = ["54.111.111.122/32"] <-- change to your public IP
}
//Allow all from Branch1 Public IP Linux Host with TE Mgmt - IP address must be changed after initial deployment
ingress {
from_port = 0
to_port = 0
protocol = "-1"
cidr_blocks = ["52.111.111.14/32"] <-- change to your public IP
}
The folder 02-Branch2 will be used for Branch2 creation. It has the same structure as Branch1 folder. Follow the same steps as for Branch1. Branch2 does not need WAN Emulator, which simplifies the process. Please make sure, that you are using the right pem ssh file for this right region and will not run into cloud limits described above in the branch1 section.
Terraform scripts in the folders 03-CSP-Region1-Cloud-App and 04-CSP-Region2-Cloud-App create cloud apps in a configured AWS region. A simple nginx webserver running on port 8003 and 8004 will be used to simulate critical traffic. The following post-install CLIs will install and configure the web server.
sudo amazon-linux-extras install nginx1
sudo nano /etc/nginx/nginx.conf <-- adjust TCP port to 8003 / 8004
sudo service nginx start
sudo chkconfig nginx on
ss -tlpn | grep :8003 <-- verify that the web server is running on the right port, in this case 8003
You can install any traffic generation / monitoring tool. We use ThousandEyes agent on Branch1 for monitoring and page load tests. Assuming you have access to ThousandEyes Dashboard, you can copy the token and install the agent with the following 3 lines.
curl -Os https://downloads.thousandeyes.com/agent/install_thousandeyes.sh
chmod +x install_thousandeyes.sh
sudo ./install_thousandeyes.sh -b XXX-Token-XXX
The first two scripts will create two cloud Apps with Web Server running in two different Availability Zones (AZ). The third script will create Shared Services VPC with two Cisco FTDv virtual firewall in two different AZs and AWS GWLB load-balancing traffic across two AZs. The last script creates SD-WAN VPC with two Catalyst 8000v virtual SD-WAN routers running in two different AZs.
Please note, that Terraform currently (Nov. 2021) does NOT support TGW Connect (GRE) attachments - see details here You can connect SD-WAN VPC manually as Connect Attachment to TGW, use VPN Attachment instead or use other tools.
Simplified Packet From Host VPC to SD-WAN: Host VPC -> AWS TGW -> GWLB -> FTDv -> TGW -> SD-WAN Returning traffic: SD-WAN -> AWS TGW -> GWLB -> FTDv -> TGW -> Host VPC Please see detailed steps for the packet flow in this AWS blog.
GENEVE protocol is used for load balancing between GWLB and FTDv. FTDv software version 7.1 or later supports GENEVE protocol.
Appliance mode is required for symmetric routing, it will be enabled for the Shared Services VPC attachment to AWS TGW.
The MRF Multicloud solution described below has the fiollowing benefits:
- Simplifies network design by introducing regions like US West or Europe.
- Improves scalability
- Improves reliability by using two different CSPs
- Simplifies control policies for traffic steering
- Subregions provide another granularity level for control and isolation
Summary of the deployment steps
- Use Terraform scripts to deploy the topology shown above.
- Make sure, that your Controllers and SD-WAN routers are running 20.10 / 17.10 Software
- Configure MRF - refer to this CCO Config Guide. In a nutshell, you will need to configure the following:
- Correct Region and Subregion on the Edge routers
- Regional vSmart serving both regions
- Enable MRF on vManage and create regions 1 and 2 in the Network Hierarchy Manager.
Use Case 1: Redundancy / Load Balancing The following requirements will be met:
- Subregion 1 (SJ) uses by default San Francisco Border Router
- Subregion 2 (San Diego) uses Los Angeles Border Router
- In case of a single backbone failure -> auto failover
Border Router for San Francisco will be configured with Region 1 and Subregion 2 as shown below:
system
system-ip 101.1.1.1
site-id 101
region 1
subregion 1
!
role border-router
organization-name mrf-multicloud-demo
vbond 44.227.177.103
!
Border Router in Los Angeles will be placed in the Subregion 2:
system
system-ip 103.1.1.1
site-id 103
region 1
subregion 2
!
role border-router
organization-name mrf-multicloud-demo
vbond 44.227.177.103
!
In normal case an edge router in San Jose (Subregion 1) will have the following entry for an IP address of the Region 2:
Reg1-Sub1-ER1#sh ip ro vrf 10
...
m 10.211.1.11 [251/0] via 101.1.1.1, 06:58:01, Sdwan-system-intf
...
Reg1-Sub1-ER1#
In case of San Francisco Border Router outage, the route will be replaced with the next hop of the Los Angeles BR:
Reg1-Sub1-ER1#sh ip ro vrf 10
...
m 10.211.1.11 [251/0] via 103.1.1.1, 06:58:01, Sdwan-system-intf
...
Reg1-Sub1-ER1#
Use Case 2: Isolate a subregion with a simple control policy A centralized control policy will be used for on-demand isolation of one Subregion from the rest of the network:
policy
control-policy block-reg1-sub1
sequence 1
match route
region-enhanced region 1
region-enhanced subregion 1
!
action reject
!
!
sequence 2
match tloc
region-enhanced region 1
region-enhanced subregion 1
!
action reject
!
!
default-action accept
!
!
apply-policy
region 1
role border-router
control-policy block-reg1-sub1 out
!
!
Once you have successfully brought up two virtual branches and cloud-based app infrastructure from the previous section, you can proceed with Cisco SD-WAN Cloud onRamp configuration. All configuration steps for site-to-site and site-to-cloud use cases will be done in Cisco vManage. Please refer to this Cisco Online Documentation article for details.
In a nutshell, the workflow has the following steps:
- setup cloud account
- discover cloud infrastructure
- create cloud gateways
- connect cloud infrastructure with SD-WAN networks.
After successful creation of two cloud gateways in appropriate regions, you should create an SD-WAN control policy, which will route critical traffic via CSP backbone and send best effort traffic via public internet. In our example we use ports 8001-8004 to simulate critical traffic. Such policy can be created in vManage using graphical user interface. The following control policy is a CLI preview/export from vManage. It matches traffic sent to TCP port 8001 and sets next hop to the closest CSP entry location. It has 4 different blocks because it follows the traffic flow: Branch1 -> CSP Region 1 -> CSP Region 2 -> Branch2. And back.
viptela-policy:policy
data-policy _vpn10-list_USwest-Branch
vpn-list vpn10-list
sequence 1
match
source-ip 0.0.0.0/0
destination-port 8001 <-- matching the critical traffic
!
action accept
set
vpn 10
tloc 41.41.41.1 color public-internet encap ipsec <-- setting next hop
!
!
!
default-action accept
!
data-policy _vpn10-list_Megaport-GW-Sydney
vpn-list vpn10-list
sequence 1
match
source-ip 0.0.0.0/0
source-port 8001
!
action accept
set
vpn 10
tloc 41.41.41.1 color private2 encap ipsec
!
!
!
default-action accept
!
data-policy _vpn10-list_Sydney-Branch
vpn-list vpn10-list
sequence 1
match
source-ip 0.0.0.0/0
source-port 8001
!
action accept
set
vpn 10
tloc 42.42.42.1 color public-internet encap ipsec
!
!
!
default-action accept
!
data-policy _vpn10-list_Megaport-GW-USwest
vpn-list vpn10-list
sequence 1
match
source-ip 0.0.0.0/0
destination-port 8001
!
action accept
set
vpn 10
tloc 42.42.42.1 color private2 encap ipsec
!
!
!
default-action accept
!
lists
site-list megaport-sydney-list
site-id 42
!
site-list megaport-uswest-list
site-id 41
!
site-list sydney-branch-list
site-id 112
!
site-list uswest-branch-list
site-id 111
!
vpn-list vpn10-list
vpn 10
!
!
!
apply-policy
site-list megaport-sydney-list
data-policy _vpn10-list_Megaport-GW-Sydney from-tunnel
!
site-list uswest-branch-list
data-policy _vpn10-list_USwest-Branch from-service
!
site-list sydney-branch-list
data-policy _vpn10-list_Sydney-Branch from-service
!
site-list megaport-uswest-list
data-policy _vpn10-list_Megaport-GW-USwest from-tunnel
!
!
- Nikolai Pitaev - Initial work - Cisco