IBM will do an on-site presentation of their GPFS/HPSS Interface (GHI) product. The topics and questions that were submitted to IBM are listed below.
The following link for current product documentation: http://www.hpss-collaboration.org/online_doc.shtml.
Topics
- General overview of intended use.
- HSM vs. backup.
- Difference between how GHI migrated files are stored in HPSS and how traditional client files are stored.
- Describe any scalability limitations and what we can do to minimize or avoid them. Will we hit a wall?
- Disk cache provisioning for GHI aggregation indexes.
- Day to day administration.
- Troubleshooting.
- Upgrade process.
- Roadmap of future enhancements.
Questions
- What other sites are running GHI in production?
- How large are they?
- How large are their GHI managed file systems?
- What is their tape storage capacity?
- Do you recommend using 3rd party HPSS clients in parallel with GHI?
- If we choose to provide a 3rd party HPSS client such as HSI/HTAR to the users, is there any way to prevent them from archiving files that have already been archived via GHI?
- Is there a GHI command that will allow users to trigger archives immediately? There appears to be a command to stage data prior to needing it.
- Can we prevent users from initiating unintended mass recalls of data, running something like "file"?
- We will have accumulated a significant amount of data before deploying GHI. Describe how we would avoid overloading the system while catching up.
- How many FTE are typically required to support GHI?
- Will GHI development keep pace with GPFS upgrades?