Skip to content

Latest commit

 

History

History
88 lines (74 loc) · 3.84 KB

README.md

File metadata and controls

88 lines (74 loc) · 3.84 KB
Raw

Overview

This repo contains a Terraform module to deploy the Apiary data lake component. The module deploys various stateful components in a typical Hadoop-compatible data lake in AWS.

For more information please refer to the main Apiary project page.

Architecture

Datalake architecture

Key Features

  • Highly Available(HA) metastore service - packaged as Docker container and running on an ECS Fargate Cluster.
  • PrivateLinks - Network load balancers and VPC endpoints to enable federated access to read-only and read/write metastores.
  • Managed schemas - integrated way of managing Hive schemas, S3 buckets and bucket policies.
  • SNS Listener - A Hive metastore event listener to publish all metadata updates to a SNS topic, see ApiarySNSListener for more details.
  • Gluesync - A metastore event listener to replay Hive metadata events in a Glue catalog.
  • Metastore authorization - A metastore pre-event listener to handle authorization using Ranger.
  • Grafana dashboard - If deployed in EKS, a Grafana dashboard will be created that shows S3 bucket sizes for each Apiary bucket.

Variables

Please refer to VARIABLES.md.

Usage

NB: This module currently requires you to use it from a machine with bash, aws, mysql, and jq CLI tools installed.

Example module invocation:

module "apiary" {
  source                   = "git::https://github.com/ExpediaGroup/apiary-data-lake.git"
  aws_region               = "us-west-2"
  instance_name            = "test"
  apiary_tags              = "${var.tags}"
  private_subnets          = ["subnet1", "subnet2", "subnet3"]
  vpc_id                   = "vpc-123456"
  hms_docker_image         = "${aws_account}.dkr.ecr.${aws_region}.amazonaws.com/apiary-metastore"
  hms_docker_version       = "1.0.0"
  hms_ro_cpu               = "2048"
  hms_rw_cpu               = "2048"
  hms_ro_heapsize          = "8192"
  hms_rw_heapsize          = "8192"
  apiary_log_bucket        = "s3-logs-bucket"
  db_instance_class        = "db.t2.medium"
  db_backup_retention      = "7"
  apiary_managed_schemas   = [
    {
        schema_name = "db1",
        s3_lifecycle_policy_transition_period = "30"
    },
    {
        schema_name = "db_2",
        s3_storage_class = "INTELLIGENT_TIERING"
    },
    {
        schema_name = "secure_db",
        encryption   = "aws:kms" //supported values for encryption are AES256,aws:kms
        admin_roles = "role1_arn,role2_arn" //kms key management will be restricted to these roles.
        client_roles = "role3_arn,role4_arn" //s3 bucket read/write and kms key usage will be restricted to these roles.
        customer_accounts = "account_id1,account_id2" //this will override module level apiary_customer_accounts
    }
  ]
  apiary_customer_accounts = ["aws_account_no_1", "aws_account_no_2"]
  ingress_cidr             = ["10.0.0.0/8"]
  apiary_assume_roles      = [
    {
        name = "client_name"
        principals = [ "arn:aws:iam::account_number:role/cross-account-role" ]
        schema_names = [ "dm","lz","test_1" ]
        max_role_session_duration_seconds = "7200",
        allow_cross_region_access = true 
    }
  ]
}

Notes

The Apiary metastore Docker image is not yet published to a public repository, you can build from this repo and then publish it to your own ECR.

Contact

Mailing List

If you would like to ask any questions about or discuss Apiary please join our mailing list at

https://groups.google.com/forum/#!forum/apiary-user

Legal

This project is available under the Apache 2.0 License.

Copyright 2018-2019 Expedia, Inc.