Amazon continues to be the market champion when it comes to cloud services, and with more than 160 to choose from, it's no real surprise. One of those services, the Elastic Block Store (EBS), is a vital element in many AWS customers' cloud system.

If you are currently using EBS, or are thinking about using the service, check out the features covered here. This article should give you a better idea of the sort of features available to you and help you identify ways to maximize your EBS value.

EBS Overview

EBS is a service that provides scalable, persistent block storage for use with Elastic Cloud Compute (EC2) instances. Depending on your configuration, EBS volumes remain even after an instance has been terminated, ensuring that data is not lost when an instance fails. Each volume can store up to 16 TiB of data and can be used with containers, file systems, applications, and both structured and unstructured databases.

Understanding EBS Pricing

EBS pricing is determined primarily by your storage and performance requirements. Performance costs, unsurprisingly, increase as performance does, regardless of whether it's Input/Output Operations per Second (IOPS) or throughput that improves.

Storage costs depend on the size, type, and region of your volumes. With storage, you pay a GB per month price for the storage you provision, regardless of whether you have that much data stored. Additionally, if you have a volume with provisioned IOPS, you will pay that on top of your provisioned storage fee.

Features to Take Advantage of

While cloud services can present a great opportunity to receive increased performance at a reduced cost, complex pricing or mismanagement can quickly lead to larger bills than you expect. To avoid this, check out the features of EBS covered below and leverage them to ensure you're getting a maximum ROI.

Different Volume Types

EBS offers four different volume types for you to choose from.

• EBS Provisioned IOPS SSD (io1)-designed for Iatency-sensitive transactional operations, these are IOPS provisioned volumes that offer up to 64k IOPS and up to 1k MB/s throughput. They are best used for NoSQL and relational databases.

• EBS General Purpose SSD (gp2)-designed for price/performance balance, these volumes offer up to 16k IOPS and 250 MB/s throughput. They are best used as boot volumes, development and testing, or for low-latency applications.

• Throughput Optimized HDD (st1)-designed for frequently accessed, throughput intensive workloads, these volumes offer up to 500 IOPS and 500 MB/s throughput. They are best used for big data, data warehousing, and log processing.

• Cold HDD (sc1)-designed to provide a low cost for infrequently accessed workloads, these volumes offer up to 250 IOPS and 250 MB/s throughput. They are best used for legacy or archive data.
  

The type you pick will not only determine your pricing but your potential performance so you'll want to choose carefully. Thankfully, if you find that you've selected incorrectly you can modify your volumes whenever you need with little to no downtime.

Snapshots

EBS Snapshots are the primary way of backing up volume information and can be used to restore failed volumes or create new volumes with duplicate data. They are created incrementally, reducing the total storage needed, and cost less to store than active volumes since they are typically kept in AWS S3.

Through AWS Backup, Lifecycle Manager, or custom scripting you can automate the process of creating snapshots as well as specify the frequency with which they are taken, how long they are kept, and when they are created.

To make sure you get the most from your snapshots, be sure to create them during off-business hours so they don't negatively impact performance. When restoring from snapshots, try to access all of your data before putting the new volume into production; this will eliminate initial increases in latency.

Burst Credits

All EBS volumes come initialized with enough burst credits, used to adjust to sudden increases of activity, to cover 3k IOPS for 30 minutes. Once these burst credits are used up, your performance will drop back to baseline, regardless of current activity level, and credits will only replenish once you drop below baseline.

To avoid being negatively impacted by these limits, you should carefully consider the priority of the data being accessed. If the volume being impacted contains high priority data or applications your best option is to simply increase the size of the volume you are using.

Once a volume reaches 1 TiB or larger, it is no longer affected by burst credit limits and will always receive maximum performance. Keep in mind, that if the volume type you're using doesn't offer the performance level you need, you will need to switch volume types first.

Metrics

Metrics from EBS are automatically collected for you and sent to CloudWatch. Monitoring these metrics can provide valuable insight into how your volumes are performing and what you might be able to do to increase their performance.

The metrics available come in three main categories: disk I/O, disk activity, and latency. Through analysis of disk I/O metrics, you can determine if you are using the correct volume sizes and types as well as if you need to add a load balancer or caching layer.

To determine if the limits you have set on your volumes are correct and to identify underused or orphaned volumes, you can use the disk activity metrics. Finally, latency metrics will help you determine if IOPS limits are restricting volume performance.

Raid Configuration

With EBS, you can use any Redundant Array of Independent Disks (RAID) that your OS supports. A RAID architecture is one in which duplicate data is stored in multiple locations to provide parallel access and increase performance, or to eliminate single points of failure.

Although you can choose other configurations, AWS recommends using only RAID 0 or RAID 1 to receive the greatest benefit. RAID 1, also known as disk mirroring, can be useful for duplicating mission-critical data or increasing read performance but will use more storage and does not improve write performance. RAID 0, also known as striping, will not provide data replication but can increase read and write speeds and is useful if single EBS volumes are unable to provide the performance you need.

AMI Sharing

Amazon Machine Images (AMIs), backups of entire EC2 instances as opposed to just their associated data, provide configuration information used to launch an instance. Using AMIs, you can quickly duplicate instances in a single Availability Zone or across zones and regions. This allows you to more easily automate deployment processes, recover failed instances, or share instance configurations with third-parties.

Previously, you could only share unencrypted AMIs with other AWS accounts, which increased data liability risks, but now you can share with any account that has a copy of your CMK.

Conclusion

AWS services and pricing can be difficult to navigate, but with careful configuration and monitoring, you can ensure that you are getting the performance you need without a surprisingly large bill. The features covered here are only a portion of those available to you in EBS but they are a good place to start if you want to optimize your system and improving your ROI.

Once you've maximized these features, make sure you keep an eye on the AWS news blog for information about service improvements or price changes, and conduct periodic audits of your system to ensure that it remains the best configuration for your needs.

##

Image source: Nattanan Kanchanaprat

About the Author

Gilad David Maayan is a technology writer who has worked with over 150 technology companies including SAP, Imperva, Samsung NEXT, NetApp and Ixia, producing technical and thought leadership content that elucidates technical solutions for developers and IT leadership. Today he heads Agile SEO, the leading marketing agency in the technology industry. 

LinkedIn: https://www.linkedin.com/in/giladdavidmaayan/