Storage Resources
Upcoming S3 Deployment
S3 bucket support for Team Accounts is scheduled for deployment during the second week of December 2026. Until deployment is complete, some users may not yet have access to the S3 capabilities described in this article.
Storage Resources and Account Types
The storage resources available to you depend on your account type:
- Individual accounts receive a personal home directory hosted on Amazon Elastic File System (EFS).
- Team accounts have access to shared EFS storage and may request a dedicated Amazon Simple Storage Service (S3) bucket.
S3 buckets are not provisioned automatically. Team administrators may request an S3 bucket by submitting a request through Help Desk / Research Nexus. Any user may request a team account through the standard account request process. For information on creating and managing team accounts, see the Team Accounts documentation.
Approximate storage capacities in this document are expressed in decimal terabytes (TB, where 1 TB = 10¹² bytes). Note that the underlying AWS storage services use binary tebibytes (TiB, where 1 TiB = 2⁴⁰ bytes). The values shown in the storage-capacity tables are intended as planning guidelines to help users estimate storage requirements and credit consumption.
Storage consumption depends on both the amount of data stored and how long it is retained, and is expressed in TB-days. For example, storing 1 TB for 10 days is equivalent to storing 10 TB for 1 day (10 TB-days). The storage-capacity tables use this unit to indicate how much storage can be maintained over time for a given number of credits.
EFS Storage
Amazon Elastic File System (EFS) provides a traditional file system that is directly accessible from Nexus sessions. EFS is the primary storage solution for day-to-day work in the Roman Research Nexus, including home directories, team shared storage, Jupyter notebooks, and software development.
Files stored on EFS behave like files on a traditional Linux file system and can be accessed using standard file operations. Users retain access to files and directories based on the permissions granted to them, regardless of whether those files reside in personal or team storage areas.
The EFS Storage Tiers table summarizes the EFS storage tiers used within the Roman Research Nexus, including their approximate storage capacity per credit.
EFS Storage Tiers
| Storage Tier | Description | 10 Credits | 1000 Credits |
|---|---|---|---|
| EFS Standard | Optimized for frequently accessed files and actively used data. | ~1 TB-day | ~105 TB-days |
| EFS Infrequent Access | Lower-cost storage for files that are accessed less frequently. | ~20 TB-days | ~2,000 TB-days |
Storage placement between EFS Standard and EFS Infrequent Access is managed automatically by the underlying AWS infrastructure and is generally transparent to users. See AWS documentation for lifecycle management and Infrequent Access policies.
EFS I/O Costs
In addition to storage capacity, EFS incurs costs associated with reading and writing data. The Roman Research Nexus uses EFS Elastic Throughput, which automatically adjusts throughput based on workload demand. Read and write operations are charged separately, at approximately 0.03 credits/GB for reads and 0.06 credits/GB for writes.
EFS I/O credit consumption can be estimated as:
I/O credits = (GB read × 0.03) + (GB written × 0.06)
Frequent access to files, repeated processing of large datasets, and workflows involving many small file operations can increase I/O consumption.
EFS I/O consumption will soon be included in the Nexus credit monitor, allowing teams to track these costs alongside compute and storage usage.
S3 Storage
Amazon Simple Storage Service (S3) provides storage for large collections of data and long-term retention of project products. Unlike EFS, data stored in S3 do not currently appear as files and directories in a traditional file system. Instead, data are stored in buckets and are typically accessed programmatically using tools such as s3f3 and boto3, or through other S3-compatible interfaces.
S3 buckets are available only to team accounts, but they are not provisioned automatically. Team administrators may request an S3 bucket by submitting a request through Help Desk / Research Nexus. The table below summarizes the S3 storage tiers available, including their approximate storage capacity per credit.
S3 Storage Tiers
| Storage Tier | Description | 10 Credits | 1000 Credits |
|---|---|---|---|
| S3 Standard | Optimized for frequently accessed data. | ~14 TB-days | ~1,400 TB-days |
| S3 Standard-IA | Lower-cost storage for infrequently accessed data. | ~25 TB-days | ~2,500 TB-days |
Storage placement between S3 Standard and S3 Infrequent Access is managed automatically by the underlying AWS infrastructure and is generally transparent to users. See AWS documentation for S3 storage classes. Additional information on working with S3 buckets is available on the Roman Research Nexus.
Choosing Between EFS and S3
Both EFS and S3 provide persistent storage within the Roman Research Nexus, but they are designed for different use cases. In general, EFS is best suited for active work, including notebooks, software development, and data that need to be accessed through a traditional file-system interface. S3 is often the better choice for storing large shared datasets, intermediate and final project products that must be retained for extended periods of time, and data that do not require continuous file-system access.
The choice between EFS and S3 is often influenced by access patterns, I/O requirements, and associated costs. Data that are actively modified or accessed through notebooks and software tools are generally better suited to EFS, while data that are retained, shared, or accessed less frequently are often better suited to S3. EFS incurs additional costs for data reads and writes, which should be considered for workflows involving frequent file operations or large data volumes.
Many teams use both storage systems together: EFS for day-to-day analysis and development, and S3 for larger data collections and long-term storage.
When estimating storage requirements, consider the expected volume of input data, intermediate products, and final results, how frequently data will be accessed or modified, and how long each dataset needs to be retained. These estimates can help determine the appropriate storage solution and anticipated credit consumption.
Monitor Storage Usage
Roman Research Nexus provides tools for monitoring storage usage and associated credit consumption. Regular review of storage utilization can help individuals and teams manage allocations, identify unused data, and make informed decisions about data retention.
The Cost Monitor dashboard on the Roman Research Nexus provides visibility into:
- EFS storage consumption
- S3 storage consumption
- Associated credit usage
- Historical usage trends
For additional information about storage costs and resource utilization, see the Cost Monitor documentation on the Roman Research Nexus.
For additional questions not answered in this article, please contact the Roman Help Desk.
