At IU, what is the Data Capacitor?
On this page:
Overview
The Data Capacitor is a high-speed, high-bandwidth storage system for research computing that serves all Indiana University campuses and TeraGrid users. At peak performance, the Data Capacitor has a 18.21GB per second aggregate transfer rate.
The Data Capacitor provides high read/write speeds for user data and support for very large files. Using a wide area file system, the Data Capacitor lets you access remote data as if the file system were mounted locally, allowing you to share large amounts of data with researchers at multiple remote sites.
The Data Capacitor is not intended for permanent storage of data, and is not backed up. You can archive data stored or created on the Data Capacitor on IU's HPSS; see How do I access IU's MDSS/HPSS from my TeraGrid account? It is your responsibility to arrange for long term storage of any data on the system as needed.
Hardware
The Data Capacitor is configured with:
- 34 Dell servers running Red Hat Enterprise Linux
- 20 Myricom 10-gigabit Ethernet cards
- 20 dual-port 4GB Fibre Channel QLogic QLA2432 HBAs
- 10 DataDirect Networks S2A9550 storage controllers
- 30 DataDirect Networks 48-bay SATA disk chassis
- 678TB usable disk storage using the Lustre filesystem
- 6 water-cooled Rittal racks with magnetic locks
File space
Data Capacitor file space is divided into three categories:
-
Project space: The Data
Capacitor project directories
(
/N/dc/projects/projectname) are dedicated to long-term projects with storage requirements that cannot be met with other existing systems.The entire system contains up to 339TB of shared space. The default size for project requests is 10TB. If you need more than 10TB of space, you must submit a written request for review by the Data Capacitor Allocation Committee. Due to performance issues, the Data Capacitor team discourages storing a large number of small files, but will discuss arrangements if you need to do this.
Files in project space older than 60 days may be purged without notice.
To request project space on the Data Capacitor, fill out and submit the Allocation Request Form. Include:
- Your project proposal and justification
- Project participants
- Amount of space (TB) requested
- Data rate required
- Special requests (e.g., additional mounts)
- Directory name (e.g.,
projectname) for the project space
When you submit the form, it goes to the principal investigators (PIs) on the Data Capacitor grant, who will approve or deny your request. If you have questions about your request, contact the Data Capacitor team.
-
Scratch space: The Data
Capacitor scratch directory is a temporary workspace available to all
Big Red and
Quarry users. Scratch space is not
allocated and its total capacity will fluctuate based on project space
requirements.
Each Big Red and Quarry user has dedicated scratch space located at
/N/dc/scratch/username(replaceusernamewith your IU Network ID username). If you can't find your directory, email the Data Capacitor team and one will be created for you.Scratch space is meant for short-term storage of files. Files in scratch space may be purged without notice 14 days after their creation.
-
WAN space: Data Capacitor WAN
storage is intended for researchers who use computing resources at
institutions other than IU. It is a 339TB Lustre filesystem, separate
from the scratch and project space. The administrators of the Data
Capacitor must coordinate with administrators of remote resources to
mount the Data Capacitor WAN. Once the Data Capacitor WAN has been
mounted on the remote resource, users can access their data as if the
Data Capacitor's filesystem was mounted locally.
The Data Capacitor WAN is mounted on Big Red and Quarry as
/N/dcwan.Data Capacitor WAN space is similar to project space. It is intended for long-term projects, but it is not backed up. Users must make their own arrangements to back up data to HPSS/MDSS. Files may be purged without notice 60 days after their creation.
More information
For more, see the Data Capacitor home page.
This document was developed with support from the National Science Foundation (NSF) under Grant No. 0503697 to the University of Chicago and subcontracted to Indiana University. Additional support was provided by IU through its participation in the TeraGrid, which is supported by the NSF under Grants No. 0833618, SCI451237, SCI535258, and SCI504075. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the NSF.
Last modified on July 17, 2009.







