VPC Flow Logs Indexer with Databases for Elasticsearch

The vpc-flowlogs-elasticsearch command line tool ("the tool"), lets you read your VPC Flow Logs from a Cloud Object Storage (COS) Bucket and index them to an Elasticsearch cluster running in the IBM Cloud. The tool also includes a basic search/results capability.

This repository assumes that you have an existing VPC that is configured with a Flow Logs collector(1) to write to a COS bucket(2). The tool requires:

• a second bucket(4) to be created within the same COS instance to store all flow logs that have previously been indexed,
• a Database for Elasticsearch instance(3) to exist (can be any IBM Cloud region).

When executed the tool will:

• read(5) all flow logs objects from the flow logs bucket(2),
• index to Elasticsearch(7.x),
• delete the indexed objects from the flow logs bucket(2) and write them to the indexed flow logs bucket(7),
• once indexed in Elasticsearch the tool can be used to query Elasticsearch or another tool such as Postman can also be used(8).

Costs and Requirements

You must have a Pay-as-You-Go account in IBM Cloud™ to follow the steps in this repository to create resources. Since the costs for these resources will vary, use the Cost Estimator to generate a cost estimate based on your projected usage.

In addition to having an existing VPC in IBM Cloud and configured VPC Flow Logs, this project requires the provisioning of an instance of Databases for Elasticsearch. Make sure to delete services when they are no longer required in order to not incur charges in your account.

• Account with IBM Cloud
• Existing VPC
• VPC Flow Logs already configured to write to a COS Bucket
• IBM Cloud Databases for Elasticsearch version 7.9
• The Elasticsearch client is specific to version 7.x

Note: Previous version supporting Elasticsearch 6.x is available under the es_6 branch of this repository.

• Install the following tools as needed:
  • IBM Cloud CLI (If you prefer the command line over using the web console)
  • Go version 1.15.x (If you choose to build the tool from source)
  • Docker (If you choose to build the tool from source but don't want to install GO)

Install from releases binary

You can install the binary from source here

Build and install from source

1. Clone this repository to your local computer.

git clone git@github.com:dprosper/vpc-flowlogs-elasticsearch.git

2. Build from your local machine

go build

Build using Docker for target OS:

Mac OS

docker run --rm -v "$PWD":/usr/src/vpc-flowlogs-elasticsearch -w /usr/src/vpc-flowlogs-elasticsearch -e GOOS=darwin -e GOARCH=amd64 golang:latest go build -v -o build/vpc-flowlogs-elasticsearch

Linux

docker run --rm -v "$PWD":/usr/src/vpc-flowlogs-elasticsearch -w /usr/src/vpc-flowlogs-elasticsearch -e GOOS=linux -e GOARCH=amd64 golang:latest go build -v -o build/vpc-flowlogs-elasticsearch

Windows

docker run --rm -v "$PWD":/usr/src/vpc-flowlogs-elasticsearch -w /usr/src/vpc-flowlogs-elasticsearch -e GOOS=windows -e GOARCH=amd64 golang:latest go build -v -o build/vpc-flowlogs-elasticsearch

Provision a Databases for Elasticsearch instance

Create the Elasticsearch cluster.

IBM Cloud Web Console

1. Use the IBM Cloud Console to create a new instance of Databases for Elasticsearch instance.

2. Create Service Credentials which will be required to update the flowlogs.json file.

IBM Cloud CLI

1. Create an instance of Databases for Elasticsearch instance using a standard plan. Replace <region_name> accordingly.

   ibmcloud resource service-instance-create flowlogs-es databases-for-elasticsearch databases-for-elasticsearch-standard <region_name>

   The service will take a few minutes to provision. You can get updates on the status by issuing the following command periodically

   ibmcloud resource service-instance flowlogs-es

2. Once you have verified the service status changed to "create succeeded", you may proceed to create a service key:

   ibmcloud resource service-key-create flowlogs-es-key --instance-name flowlogs-es

3. Obtain the service key details in JSON format and use the information provided to populate the relevant values for elasticsearch in the flowlogs.json.

   ibmcloud resource service-key flowlogs-es-key --output json

Create a bucket for storing indexed logs

The tool will index all logs found in the COS bucket that is used for your VPC Flow Logs, as each log is indexed it is deleted and moved to another bucket, you need to provide a destination bucket for all the Flow Logs that have already been indexed. It needs to exist in the same instance as the bucket used by the Flow Logs you intend on indexing.

How to use:

Configuring

1. Copy the config/flowlogs.json to your the $HOME directory of the current user.
2. Edit the ~./flowlogs.json file and replace all instances of <provide_value> with the relevant values for COS and Elasticsearch captured above.

You can place the flowlogs.json file in any directory of your choice and pass the --config flag with the location of the file to any of the commands below.

It is also possible to set environment variables instead of using the flowlogs.json config file. Environment variables will match the name of the attributes in the json with _ replacing the ., i.e. cos.serviceEndpoint --> COS_SERVICEENDPOINT

Indexing

1. Index your existing flow logs by issuing the following command:

   ./vpc-flowlogs-elasticsearch index

   Add a --trace if you want to see the POST and response body from each index request to Elasticsearch.

2. The tool indexes 25 flow logs at a time. The indexing process may take a while based on the number of flow logs in the COS bucket. You can view the status of the indexing process by viewing the system.log under the logs directory. Errors are in the error.log.

Searching

Using the tool

1. The config/queries.json file contains a few example queries:

• "top_25_target_ips",
• "top_10_initiator_ips",
• "total_direction_by_outbound_inbound",
• "output_ommitted_es_response_body_used",
• "14_days_top_5_rejected_by_target_ip",
• "14_days_top_10_rejected_by_initiator_ip".

The output is a JSON array.

  ./vpc-flowlogs-elasticsearch search --query 14_days_top_5_rejected_by_target_ip

Add a --trace if you want to see the POST and response body from each index request to Elasticsearch.

Using Postman or similar client

1. Review the config/sample_queries.md for example Elasticsearch endpoints and queries.

Issues

Please open issues here: New Issue

Related Content

• Use IBM Log Analysis with LogDNA to Analyze VPC Network Traffic from IBM Cloud Flow Logs for VPC
• vpc-flowlogs-logdna repository

License

See License for license information.