aws-sipp

Use Ansible to build an AWS-based SIP test network

Set up AWS instances

Configure your AWS access key and secret in environment varibles prior to running Ansible. For example, you may add the following stanza to your .bash_profile:

export AWS_ACCESS_KEY_ID=accesskey
export AWS_SECRET_ACCESS_KEY=secret

cd to one of the project directories, either ansible-bare-AWS or ansible-docker-AWS

To run the playbook to provision the VPC and 2 EC2 instances, run ansible-playbook -i inventory provision.yml Note that your SSH key from ~/.ssh/id_rsa.pub will automatically be loaded for ec2-user.

Once the instances are provisioned, to install sipp, or docker, respectively, run ansible-playbook -i ec2.py deploy.yml

To terminate all instances, run ansible-playbook -v -i inventory terminate.yml

Testing bare AWS instances

Run the ansible deploy above, and note the external hostnames displayed. Make an note of one instance as client and one instance as server. Software on both is identical, so it doesn't matter which is which. A future enhancement may be able to designate and list the servers for you and avoid having the enter the IP address.

1. Open at least 3 terminal windows; in one of them, ssh ec2-user@client, and in two others, ssh ec2-user@server

2. In a terminal window on the server, run sudo tcpdump -s 768 -w udp.pcap udp to capture all UDP traffic.

3. In the second terminal window on the server, run ifconfig and note the network 10 address displayed, for example

   >   eth0      Link encap:Ethernet  HWaddr 0E:BE:9C:62:17:56  
      inet addr:10.2.2.16  Bcast:10.2.2.63  Mask:255.255.255.192
   

4. On that same server, cd sipp and run ./sipp-server.sh. This will run the server until stopped with a test scenario using a media file.

5. In a terminal window on the client, cd sipp. There are at least two scripted test scenarios:

./sipp-client-one-call.sh 10.2.2.57 using the IP address of the server you determined in step 3. This will run a single call test.

./sipp-client.sh 10.2.2.57 Again using the IP address of the server, will run a test at 400 simultaneous calls.

Testing Docker on AWS instances

Run the ansible deploy in the ansible-docker-AWS directory:

cd ansible-docker-AWS
ansible-playbook -i inventory provision.yml
ansible-playbook -i ec2.py deploy.yml

Note the external hostnames or IP addresses; again designate one as client and one as server.

1. Open at least 3 terminal windows; in one of them, ssh ec2-user@client, and in two others, ssh ec2-user@server

2. In a terminal window on the server, run sudo tcpdump -s 768 -w udp.pcap udp to capture all UDP traffic.

3. In the second terminal window on the server, run ifconfig and note the network 10 address displayed, for example

   >   eth0      Link encap:Ethernet  HWaddr 0E:BE:9C:62:17:56
             inet addr:10.2.2.16  Bcast:10.2.2.63 Mask:255.255.255.192
   

4. On the server, cd sipp/docker/sipp and run ./docker-server.sh. This will download the image from Docker Hub and run sipp in UAS mode.

5. On the client, cd sipp/docker/sipp and run the two scripted test scenarios, substituting the internal IP address of the server:

./docker-client-one-call.sh 10.2.2.16 will run a single test call

./docker-client-400calls.sh 10.2.2.16 will run 400 simultaneous calls`

Analyze results

Command-line Wireshark (tshark) will do jitter analysis of each stream: tshark -nr udp.pcap -q -z rtp,streams