Understanding the Cost of AWS Storage Solutions
Having the right knowledge going into an AWS storage arrangement will save money and reduce headaches.
In part 1 and part 2 of my blog series on Amazon Web Services storage, I talked about various available storage options in AWS. Since cost is one of the drivers in moving to the cloud, in this blog, I will focus on the cost factors and some typical use cases of the different storage options available in AWS.
To recap, here is the list of storage options available in AWS:
Block Storage:
- Ephemeral Storage
- Elastic Block Store (EBS)
- HDD
- SC1
- ST1
- SSD
- GP2
- IOPS optimized
- HDD
- Elastic File System (EFS)
Object Storage:
- S3 Standard
- S3 Standard-IA
- S3 One Zone-IA
- S3 Glacier
- S3 Glacier Deep Archive
- S3 Intelligent Tiering
Block Storage
The ephemeral storage options are associated with the physical server hosting the virtual machine. Ephemeral storage can be mounted from the virtual machine and it does not cost any money to use ephemeral storage. However, on the downside, the information stored in the ephemeral storage is lost as soon the underlying hardware of the VM changes (for example, by stopping, terminating and restarting the EC2 instance). These are good candidates for storing temporary or intermediate objects from a permanent storage option.
The next category is the externally mounted EBS block devices. The two types of EBS volumes are spinning hard disk and SSD-based disk. There are two types of HDD-based EBS volumes, regular cold HDD SC1 and throughput-optimized ST1. The two categories for SSD based devices are general purpose GP2 and IOPS-optimized IO1. EBS volumes can only be attached to one EC2 instance at any given time. It provides persistence storage and the mount point can be moved across. For normal usage, GP2 storage provides a good price/performance ratio. For storing large sequential data objects, ST1 can potentially provide good value.
Multiple EC2 instances can mount the same EFS volume simultaneously. This provides a common mechanism for sharing objects between multiple users. The cost of EFS is higher than the cost of EBS volumes because it provides a shared storage space from multiple EC2 instances.
EBS cold HDD (SC1) is the most inexpensive option and the price of throughput-optimized EBS volumes (ST1) is roughly two times the price of SC1 volumes. The price of GP2 is roughly two times the price of ST1 and with IO1 volumes incurring additional charges based on the provisioned IOPS. These are summarized in the table below.
Object Storage
There are also multitudes of S3 storage solutions available in AWS. The S3 standard provides a scalable/high throughput object store with high durability, availability and low latency. The regular S3 is less expensive than that provided by block devices, except ephemeral storage options. For example, the price of the first 50TB/month is $0.023 per GB in US-East-1 region. The S3 objects are replicated in multiple availability zones in a region.
S3 Infrequent Access is roughly half the price of S3 Standard Storage. However, there are additional fees associated with API calls and the number of times objects are retrieved from the storage. If S3 objects are retrieved more than two times per month, storage utilizing S3 IA could end up costing more than the standard S3.
S3 One Zone-IA is similar to S3 IA; however, the objects are stored only in one availability zone. It is 20 percent cheaper than S3 IA. However, since the objects are stored only in an AZ, destruction of the AZ would render the object inaccessible.
S3 Glacier provides a low-cost storage option for archiving data. In order to keep costs low, yet providing greater flexibility, S3 Glacier provides three data retrieval options, from a few minutes to a few hours.
S3 Glacier Deep Archive is a super-low-cost storage option for long-term retention and digital preservation for data that may be accessed once or twice in a year. S3 Glacier Deep Archive would be a favorable option for industries such as financial services, healthcare and public sectors, which need to retain data sets for 7-10 years or longer to meet regulatory compliance requirements. Objects stored in S3 Glacier Deep Archive are not readily accessible and can potentially take up to 12 hours before the objects can be accessed. Furthermore, there is some limit on the amount of storage that can be accessed at any given time without any penalties.
S3 Intelligent Tiering optimizes costs by automatically moving data to the most cost-effective access tier, without performance impact or operational overhead. The solution comprises a storage class optimized for frequent access and another lower-cost tier optimized for infrequent access. S3 Intelligent Tiering requires monitoring and automation charge per object and automatically moves objects from one tier to another depending on the access pattern. For example, if an object in the “regular” tier has not been accessed for 30 consecutive days, the AWS automation would move the object to the lower-cost storage class. If an object in the lower-cost tier is accessed, it is automatically moved back to the frequent access tier.
Comparing Storage Options
The cost of AWS storage solutions varies by regions/countries. Pricing information is available on the AWS website, for EBS volume type and for the S3 type. The following table is intended to provide a quick reference summary of storage solutions along with the cost of 100GB of storage in US-East-1.
| Storage Type | Description | Approx. Monthly Cost (US-East-1) for 100 GB | Typical Use Cases |
|---|---|---|---|
| Ephemeral | Storage associated with the physical server hosting the virtual machine; information stored in ephemeral storage is lost as soon as the underlying hosting of the VM changes (such as by stopping and starting the EC2 instance). | Free | Good candidate for storing temporary / intermediate objects, which can be easily recreated from a permanent storage or via some algorithm |
| EBS SC1 | Cold HDD; regular spindle-based hard disk drive. | $2.5 | Sequentially written / accessed files without random updates, such as Hadoop file systems |
| EBS ST1 | Throughput-optimized HDD or HDD with high throughput | $4.5 | High throughput for sequentially written / accessed files, such as Hadoop file systems |
| EBS GP2 | Regular SSD-based hard drive | $10 | Objects requiring random updates |
| SSD IOPS-Optimized Storage | Input / output-optimized disk volumes | $12.5 | Storage for databases being updated at high rate |
| EFS Standard | NFS-based shared file system | $30 | To share files across multiple users from multiple machines |
| S3 Standard | Standard S3 object store | $2.3 | For storing snapshots, AMIs, backups, etc. |
| S3-IA | S3 for infrequent access | $1.25 | Objects that do not need to be accessed frequently, such as security log files after 14 days |
| S3-One Zone IA | S3-One Zone infrequent access | $1 | Intermediate objects that can be recreated in another AZ when there is a problem in accessing S3 in a given region |
| S3 Glacier | Long-term backup | $0.40 | Log files, object archives |
| S3 Glacier Deep Archive | Super low-cost storage | $0.099 | Storage for compliance and regulatory audits |
| S3 Intelligent Tiering – Infrequent / Frequent Access | AWS solution for moving objects across different access tiers depending upon access patterns | $1.25 to $2.3 |
Table 1. Need-to-Know Information on AWS Storage Options
Navigating Cloud Storage Solutions
While I’ve covered the basics in this post, other aspects such as API requests, data egress charges, S3 transfer acceleration and replication pricing also impact the cost of a storage solution. CDW professional services can assist your organization in the cloud journey (including storage) by laying a solid foundation on which additional services can be developed.
A plethora of options are available for creating a storage solution in the cloud. The right choice of storage option depends on the typical use case to be supported. A CDW solution will involve our experts carving out the right solution for your company and team.
