System Archiwizacji - HSM System Archiwizacji - HSM

The archiving system provided by PSNC is based on a cluster of two servers sharing disk resources. Shared disk space is managed by the Hierarchical Storage Management (HSM) software and this mechanism is implemented using IBM Spectrum Protect for Space Management software (until recently called IBM Tivoli Storage Manager for Space Management). The first level of the hierarchy is the disk space located on the array and managed using the IBM Spectrum Scale file system (before renaming it was GPFS - General Parallel File System). The second level of data storage in the hierarchy is to stop the tape, where the files are being transferred to, thus freeing up space in the disk space. The tape space is a separate partition of the IBM TS 3500 library, which is equipped with LTO5 and Jaguar gen8 tape drives. Due to the fact that it takes time to transfer files between tape-disk spaces, downloading files from the archive involves waiting for the availability of the file and the waiting time linearly depends on the size of the file (so it can be from a few seconds to several minutes). Waiting time can also be increased if several clients want to access files at the same time and file recovery tasks are waiting for access to drives that are limited in number.

The archiving service is available at the address: backup.man.poznan.pl
It is also possible to access directly to the servers providing the service, i.e:

  • corylus.man.poznan.pl (150.254.161.115)
  • cerasus.man.poznan.pl (150.254.161.117)

    They can use the archiving system in PSNC:
  • users of HPC systems after filling in the grant application in PSNC with the service https://hpc.man.poznan.pl/
  • PSNC users (as indicated on the PSNC Wiki)
  • institutional users - according to individual arrangements with System Administrators - storage[at]man.poznan.pl

    Use your user ID and password to log in to your account
  • dla for PSNC accounts: <user-name> @ < corylus | cerasys>
  • dla for external accounts: <user-name> @ domena @ < corylus | cerasys>

    Creating the first password or later changing the password to the account is done using the service: https://drawer.man.poznan.pl/pwd/?action=sendtoken ATTENTION! According to the security policy, if no one has logged in for 3 months, the password must be generated again.

    The data is accessible via an encrypted protocol using an application using ssh (encrypted transmission):
  • sftp
  • winscp -- For this program, we recommend changing the "time limits" parameter on 1200 s. (this will allow you to wait calmly for the file to be recalled from the tape to disk)
  • sshfs
  • ftp (NOTE! unencrypted connection, first login after scp is required to automatically generate correct home directories)

    Two directories are available in the home directory for use via encrypted protocols
  • .ssh - where the keys to the certificates should be uploaded
  • ARCHIVE - where you can create directories and place data archives in them
    Using the ftp protocol, the user goes straight to the ARCHIVE directory

    Please make sure that the files you upload to the archive system are adapted to the requirements of the HSM system architecture, it is forbidden to upload thousands of small files. Small files cause a significant load on the archive service, so it is preferable to place large files (e.g. large multimedia files, system images, database backups), or many files packed in the archive (zip,tar,gzip etc.).

    Due to long transfer times of 500GB files and more, it happens that our system tries to handle a given file several times while it is still being transferred to PSNC. Therefore, we also ask for an upper limit on file size (preferably a single file should not take more than 2 hours to upload), as this can significantly affect the speed of server support..

    Please do not move files placed on the server between directories, e.g. by renaming the directory e.g. server_X to server_X.old - such an operation causes additional, long-lasting data traffic related to the operations of erasing data from the archiving and backup server (original location - server_x directory) and performing backup and migration for a new data location (server_x.old directory).

    We recommend additional data encryption before placing it in the archive, e.g. with a powerful AES Crypt application.

    System administrators are available at storage[at]man.poznan.pl