Storage Considerations: Quick Deployment - Teradata Software for AWS

Teradata Vantageā„¢ on AWS (DIY) Quick Deployment Guide

Teradata Vantage on AWS
Release Number
November 2018
English (United States)
Last Update
Product Category
Storage Type When Using to Deploy Teradata Products Considerations and Use Cases
Amazon EC2 Instance Local Storage Local storage is used for the data volume, but the root disk uses 350 GB EBS. A minimal charge occurs at the time you deploy any instance that uses a root disk.

Local storage leads to data loss in the following node events: shutdown (init 0), node hang, machine check (hardware fault such as a double bit memory error), stopping the instance through the Amazon EC2 Console, or terminating an instance. Data loss does not occur with certain node events, such as a reboot (init 6) or node panic (sometimes referred to as an oops event). An oops event causes a memory dump and then the node is rebooted.

To avoid data loss, deploy a Teradata Database instance from an EBS-based AMI, or use an automated Teradata backup solution such as Teradata Data Stream Controller.

Local storage is temporary or ephemeral, but provides better performance compared to EBS. It is ideal for testing and development where you do not need to keep data.
Local (both SSD and HDD)
  • Storage requires a running instance
  • Potential for data loss if instance fails, but data protection is available using Teradata DSC
  • Lower reliability than EBS; HDD is lower reliability than SSD
SSD Only
  • Highest performance
  • Highest cost per vCPU per TB
  • Up to 6.4 TB of storage per EC2 instance
HDD Only
  • Lower performance than SSD
  • Lower cost per vCPU per TB than SSD
  • Up to 48 TB storage per EC2 instance
Amazon Elastic Block Storage (EBS) The root volume for an instance and additional EBS volumes that you attach to the instance persist by default. Unless you set attached volumes to delete on termination, all data remains and storage-related charges continue to accrue. You can enable EBS storage deletion either when configuring an instance or after deploying an instance.

For Teradata Database, only m4 instance types on AWS support EBS storage. Instance types supporting EBS storage may vary across other Teradata Software for AWS products.

Teradata recommends m4 instance types if you are deploying Teradata Database MPP instances. Teradata Database runs at the speed of the slowest node so using any of these EBS optimized instance types is important to get the maximum EBS volume bandwidth for your instance.
Manually attaching EBS volumes to your instance on Linux and updating the /etc/fstab file can result in data loss or corruption. EBS volumes are automatically attached at the time the instance is deployed.

For EBS volume type performance characteristics, see Amazon EBS Product Details.

EBS is ideal if you want persistent data for production systems.
  • Highest reliability
  • Lower performance (network-attached)
  • Lowest cost per vCPU even with extra charges for EBS
  • Up to 72 TB remote storage (pre-provisioned) on SSD volumes attached to the instance
  • Data persists independent of running an instance which provides the ability to shut down an instance between sessions to reduce EC2 costs and lower cost of ownership
  • Total node storage size for EBS instance types can be set using the Data Storage parameter when deploying Teradata Database for Advanced and Enterprise tiers.
  • Expand EBS storage after deploying a Teradata Database instance.
Amazon Simple Storage Service (S3) Supported by Teradata Access Module for AWS, Teradata DSC, and Teradata QueryGrid.

Although you can use either EBS or S3 storage to back up and restore your Teradata Database, Teradata recommends using S3 storage as a backup target if you are using Teradata DSC.

Use cases:
  • Use Teradata Access Module for AWS with TPT to load and export between Amazon S3 and Teradata Database.
  • Use Teradata DSC to back up and restore between Amazon S3 and Teradata Database.
  • Use a Teradata QueryGrid Presto connector to query data stored in Amazon S3.