High Performance Storage System |
![]() |
Incremental Scalability
Based on storage needs and deployment schedules, HPSS scales incrementally by adding computer, network and storage resources. A single HPSS namespace can scale from petabytes of data to exabytes of data, from millions of files to billions of files, and from a few file-creates per second to thousands of file-creates per second.
About HPSS :
Best of breed for tape
HPSS is best of breed for tape
-
HPSS is considered best of breed for tape because of our efficient use of tape hardware and media.
Tape hardware and media efficiency translates to a lower total cost of ownership (TCO) of long term data storage.
HPSS increases hardware and media efficiency by maximizing tape transfer performance, reducing wasted movement of tape media and tape robotics, cutting the cost of redundant tape, and protecting against silent data corruption on tape.
HPSS maximizes tape transfer performance
-
Striped tape
Most clustered file systems and high performance computers can perform I/O at many gigabytes per second, while modern tape drives have native transfer rates less than 500 MB/s. HPSS can be configured to transfer a single file in parallel to multiple tape drives. This allows HPSS to transfer large files at speeds far beyond the limits of a single tape drive.
In 2017, Oak Ridge National Laboratory (ORNL) deployed a 4-wide tape stripe into production and are logging single file transfer rates that exceed 1 GB/s using 252 MB/s tape technology.
- Small file tape aggregation
Writing small files to tape is often an arduous task, but NOT for HPSS. The HPSS tar utility (HTAR) can be used to store small files in an efficient manner. Without HTAR, HPSS can be configured to automatically group small files on tape to improve tape transfers. HPSS can often achieve near-native tape transfer rates even when doing tape I/O with the smaller files.
HPSS reduces wasted movement of tape media and tape robotics
- Efficient use of tape media
HPSS groups tape recall requests by tape cartridge to reduces unnecessary tape mounts and dismounts.
HPSS performs offset ordered tape recalls when using LTO tapes. Offset ordered tape recalls improve the efficiency of recalling collocated files. Modern tape drives write data to tape in a serpentine manner, writing data in different directions on alternating wraps. Because of the serpentine nature of data on tape, LTO tape offset ordering does not minimize seek-times for files that are not collocated.
HPSS performs recommended access ordered (RAO) tape recalls when using enterprise tapes. RAO is an enterprise tape drive feature. HPSS sends a request to the enterprise tape drive, and the tape drive efficiently orders the recalls for HPSS. RAO improves the seek-time between files on tape by 40% to 60%. When requirements demand high tape recall rates, RAO and enterprise tape are often the most economical long term storage option at scale.
- Efficient use of tape robotics
HPSS exploits dual active tape robot movement for tape libraries supporting this capability. Lee Jesionowski, IBM's tape automation lead architect says, "This can more than double the tape mount rate efficiency."
HPSS also orders tape mount requests to ensure the tape robot is not moving great distances without a cartridge in the gripper.
HPSS versions 7.5.2 and 7.5.3 have introduced SCSI PVR and Spectra Logic library compatibility improvements that have increased tape mount rate efficiency from 78% to 99% on IBM tape libraries and from 42% to 99% on Spectra Logic tape libraries.
HPSS cuts redundant tape costs
-
Prior to the commercialization of RAIT, redundant tape was accomplished by making multiple tape copies.
HPSS redundant array of independent tape (HPSS RAIT) is tape striping with one or more rotating parity blocks.
HPSS has introduced a feature that cuts the cost of redundant tape by 50%, 66%, 75%, or more.
In addition to reducing the cost of redundant tape, HPSS RAIT users enjoy all the benefits of striped tape - high speed tape reads and tape writes.
To further clarify the Oak Ridge National Laboratory (ORNL) 4-wide tape stripe cited earlier, it is a 4+Parity HPSS RAIT stripe, which also cut their projected dual-copy redundant tape cartridge costs by almost 75%.
HPSS protects against silent data corruption on tape
-
All data can be corrupted - data in motion and data at rest.
Silent data corruption is when data corruption goes undetected.
HPSS validates the integrity of data when data are written to
tape using file checksums and T10 Logical Block Protection (T10 LBP).
With HPSS end-to-end data integrity, corrupted data will not be
written to tape. HPSS also provides a highly efficient data
re-validation tool to validate data on tape. When data
corruption is identified, redundant tape (multi-copy or RAIT)
is used to correct the problem.
Come meet with us!
2023 HUF The 2023 HPSS User Forum (HUF) will be an in-person event scheduled October 30th through November 3rd, 2023, in Herndon, VA. This will be a great opportunity to hear from HPSS users, collaboration developers, testers, support folks and leadership (from IBM and DOE Labs). Would you like to Learn More? Please contact us if you are not a customer but would like to attend. |
HPSS @ SC23 The 2023 international conference for high performance computing, networking, storage and analysis will be in Denver, CO from November 12th through 17th, 2023 - Learn More. As we have each year (pre-pandemic), we are scheduling and meeting with customers via IBM Single Client Briefings. Please contact your local IBM client executive or contact us to schedule a HPSS Single Client Briefing to meet with the IBM business and technical leaders of HPSS. |
HPSS @ STS 2024 The 5th Annual Storage Technology Showcase is in the planning phase, but HPSS expects to support the event. Check out their web site - Learn More. |
HPSS @ MSST 2024 The 38th International Conference on Massive Storage Systems and Technology will be in Santa Clara, California in May of 2024 - Learn More. Please contact us if you would like to meet with the IBM business and technical leaders of HPSS at Santa Clara University. |
HPSS @ ISC 2024 ISC 2024 is the event for high performance computing, machine learning, and data analytics, and will be in Hamburg, Germany at the Congress Center Hamburg, from May 12th through May 16th, 2024 - Learn More. As we have done each year (pre-pandemic), we are scheduling and meeting with folks attending the conference. Please contact us meet with the IBM business and technical leaders of HPSS. |
What's New?
HPSS 10.2 Release - HPSS 10.2 was released on February 16th, 2023 and introduces six new features and numerous minor updates. |
HUF 2022 - The HPSS User Forum was hosted by IBM Houston in October 2021, at their IBM Houston Kurland building. |
Celebrating 30 Years - Fall 2022 marks the 30th anniversary of the High Performance Storage System (HPSS) Collaboration. |
HPSS 10.1 Release - HPSS 10.1 was released on September 30th, 2022 and introduces fourteen new features and numerous minor updates. |
Lots of Data - In March 2022, IBM/HPSS delivered a storage solution to a customer in Canada, and demonstrated a sustained tape ingest rate of 33 GB/sec (2.86 PB/day peak tape ingest x 2 for dual copy), while simultaneously demonstrating a sustained tape recall rate of 24 GB/sec (2.0 PB/day peak tape recall). HPSS pushed six 18-frame IBM TS4500 tape libraries (scheduled to house over 1.6 Exabytes of tape media) to over 3,000 mounts/hour. |
DOE Announces HPSS Milestone - Todd Heer, Deputy Program Lead, Advanced Simulation and Computing (ASC) Facilities, Operations, and User Support (FOUS), announced that DOE High Performance Storage Systems (HPSS) eclipse one exabyte in stored data. |
Atos Press Release - Atos boosts Météo-France’s data storage capacity to over 1 exabyte in 2025 to improve numerical modeling and climate predictions. Want to read more? |
Capacity Leader - ECMWF (European Center for Medium-Range Weather Forecasts) has a single HPSS namespace with over 824 PB spanning over 556 million files. |
File-Count Leader - LLNL (Lawrence Livermore National Laboratory) has a single HPSS namespace with over 78 PB spanning 1.746 billion files. |
Older News - Want to read more? |