aws

Using AWS to get the data

As described in the paper, our results were produced using Amazon EC2 instances. We used dedicated instances. c5.2xlarge was used for PIKOS with k from 1 to 4 and IKOS, c5.4xlarge for PIKOS with k from 5 to 8, and c5.9xlarge for PIKOS with k larger than equal to 9,

We used 200 instances running together to get the numbers in a reasonable time. We created an image (AMI) for this purpose. EC2 Instances were spawned from this image, and user-data were passed to run the assigned batches. The instances access two S3 buckets to obtain the batch and save the results. Thus, they should have access permissions to those buckets through IAM role.

Prerequisite

The scripts uses awscli (>=1.16.142) and boto3 (>=1.9.132).

$ pip install --user awscli boto3

BenchExec mentioned in the dependencies is also required to generate tables.

AWS set up

AMI

One first needs to create an AMI image. To create this image, follow the steps in Install Dependencies. The scripts assume that pikos is in PATH. ../tools/ in this repository should be in /home/ubuntu. A dummy benchmark file (dummy.bc) in /home/ubuntu/ is used to warm up the EC2 instances. SVC and OSS benchmarks should be in /home/ubuntu/.

S3 Buckets

Create two S3 buckets. BUCKET1 stores the batch configurations, and BUCKET2 stores the experiment results.

Creating S3 bucket

IAM role

Create an IAM role that gives access rights on the buckets to instances. The instances uses this role to read from BUCKET1 and write to BUCKET2.

Creating IAM role

Go to IAM, policies. Create a policy using JSON. Edit the policy boilerplate below, substituting for POLICYNAMEHERE and BUCKETNAMEHERE. This policy will allow the role to read/write files in the S3 bucket. Do this two times for BUCKET1 and BUCKET2 created above.

{
  "Version": "2012-10-17",
    "Statement": [
    {
      "Sid": "POLICYNAMEHERE",
      "Effect": "Allow",
      "Action": "s3:*",
      "Resource": "arn:aws:s3:::BUCKETNAMEHERE/*"
    },
    {
      "Sid": "VisualEditor1",
      "Effect": "Allow",
      "Action": "s3:ListAllMyBuckets",
      "Resource": "*"
    },
    {
      "Sid": "AllowRootListing",
      "Action": [
        "s3:ListBucket"
      ],
      "Effect": "Allow",
      "Resource": [
        "arn:aws:s3:::BUCKETNAMEHERE"
      ]
    }
  ]
}

The trust policy for your job role needs to extend to ec2 and ECS. Go to edit trust relationships and add,

{
  "Version": "2012-10-17",
  "Statement": [
  {
    "Effect": "Allow",
    "Principal": {
      "Service": "ec2.amazonaws.com"
    },
    "Action": "sts:AssumeRole"
  }
  ]
}

Security group

Create a security group.

Creating a security group

Key and Security group

Create a key pair.

Creating key pair.

Scripts to run experiments on AWS

.
├── upload_to_s3.sh
├── run_instances.py
├── user-data-template.txt
├── download_from_s3.sh
├── merge_csv.py
├── partition.py
└── batches/
    ├── batch-b24-0.xml
    ├── batch-b24-1.xml
    └── ...

• upload_to_s3.sh uploads batch definitions in batches/ to BUCKET1.
• run_instances.py runs the instances. It takes as arguments an instance type, an ami Id, a security group, a key name, and an iam role name.
• user-data-template.txt is a template of script that will be run in instances on boot.
• download_from_s3.sh downloads the results in BUCKET2 to current directory.
• merge_csv.py merges the downloaded results into unprocessed.csv.
• partition.py partitions the list of benchmarks into batches in batches/.
• batches/ contains the batch configurations.
  • batches/batch-{config}-{batch number}.xml is fed into benchexec in an instance.

Reproducing data on AWS instances

Steps to reproduce data ../results-paper/all.csv and table 3.

0. Follow the steps in prerequisite.

1. Run upload_to_s3.sh to upload the batches/ to BUCKET1.

$ ./upload_to_s3.sh your-BUCKET1-name

2. Run run_instances.py to run the instances. The script assumes that you are in region us-west-2 (Oregon).

First, run baseline IKOS, Pikos<2>, and Pikos<4> in c5.2xlarge. You need to be able to run 200 c5.2xlarge instances in your account.

$ ./run_instances.py b24 ami-01f8031a36fcb210d your-security-group-id your-key-name your-iam-role-name your-BUCKET1-name your-BUCKET2-name

After all instances finish, run Pikos<6> and Pikos<8> in c5.4xlarge. You need to be able to run 200 c5.4xlarge instances in your account.

$ ./run_instances.py 68 ami-01f8031a36fcb210d your-security-group-id your-key-name your-iam-role-name your-BUCKET1-name your-BUCKET2-name

After all instances finish, run Pikos<12> and Pikos<16> in c5.9xlarge. You need to be able to run 5 c5.9xlarge instances in your account.

$ ./run_instances.py 1216 ami-01f8031a36fcb210d your-security-group-id your-key-name your-iam-role-name your-BUCKET1-name your-BUCKET2-name

3. Run download_from_s3.sh to download the results from BUCKET2 to ./results-ikos.

$ ./download_from_s3.sh your-BUCKET2-name

4. Run merge_csv.py to generate the merged table (b24.csv, 68.csv, 1216.csv).

$ ./merge_csv.py

5. Manually process the b24.csv, 68.csv, and 1216.csv to obtain the csv file as the one in ../results-paper/all.csv.