Data storage services > Archive server
Sisältö
 
Tehdyt toimenpiteet
Tämä sivu suomeksi

Archive server

Storagetek SL8500 Storagetek SL8500
The archive server stores data on tape. This makes storing massive amounts of data more economical. Compared with disk memory, the drawback is the slower retrieval of data.
The archive server (bender.csc.fi) is a computer that utilizes an automatic tape robot and a large amount of disk space. The archive server has a saving capacity of 560 terabytes. The management program on the archive server is Oracle SAM-FS, offering an automatic hierarchical file service (Hierarchical Storage Management, HSM).

The archive server is used through the network to save files and read files back into the computation servers. The user can move into the archive area with the command:

cd $ARCHIVE

The CSC archive server is the Sun SPARC Enterprise T5440 server machine. The robot mechanism is the StorageTek SL8500 robot.

Features of the Sun SPARC Enterprise T5440 server

  • 2 x 1,4 GHz UltraSPARC-T2+ CPUs
  • 32 gigabytes of central memory
  • 14.6 terabytes of efficient buffer disk space (mirrored system)
  • 4 x Gigabit Ethernet network connections
  • 1 x 10 Gigabit Ethernet network connections
  • 2 x FC HBA controllers to handle communicatrion between robot, taping equipment and server
  • 1 x SAS controller

Features of the robot and the taping equipment

  • Total capacity of 1448 cassettes
  • 4 x IBM LTO3 FC tape drives
  • 2 x HP LTO4 FC tape drives
  • 4 x HP LTO5 FC tape drives
LTO3 recorder's capacity is 400 gigabytes/cassette as non-compressed. Recorders will compress the data, which leads in fact to 1.2 - 1.4 packing density.

An archive server consists of a server machine, a disk-based temporary memory, and a tape robot. When a file is taken back to use from tape, the system must first retrieve it to temporary memory on a disk, and from there it is transferred to the user. The file is retained on the disk until disk space is needed for handling another file. The disk serves as a temporary memory both during search and save operations. The temporary memory is purged so that the copy of older files on the disk is removed to make more recent files quickly available from the disk.

When a file has been saved on the archive server, it is first saved on the archive server disk and from there it is later copied on tape. By default, two copies are always taken of files.

If a file has been archived and it cannot be found on the disk, the transfer time is usually a great deal longer. This is because it takes time to copy the file from the tape archive to the archive server disk. This time can be estimated by considering that inserting and removing a cassette to and from the disk position usually takes less than a minute. After that, you have to play the tape to the right position. This takes between a few seconds to several tens of seconds. Next, you copy the file over the network.

Archiving files to the archive server

You don’t need any separate procedures to use the directory path to access an archive server on CSC’s computing servers. The path of your own archive area is saved in the environment variable $ARCHIVE. The file is transferred to the archive server by copying it to the directory

$ARCHIVE/file name

To go to the archive area, write 

cd $ARCHIVE

You can handle the files with Unix commands 

cp, mv, rm etc.

The only difference is that when the file is only on tape, retrieving it takes longer. For example, if the command is

more file

it may take several minutes until the first row appears on the screen.

Various jam and error situations may prolong the retrieval time considerably. It is good to remember that the file server archive is on magnetic tape and not on disk - it can never work as fast as a disk.

See more Archive, $ARCHIVE.