Re: Store Taking Too Long



I'll definitely second that recommendation.

The other very useful bit of information offered by the "statistics" option - apart from tape output rate - is the number of tape i/o errors - which could significantly affect your times.

If you save every stdlist that included the "statistics" option, you could also easily compare no. of files stored, compression rate, i/o errors, i/o rate, total bytes output (compressed & uncompressed), etc.- and get a very accurate reading of exactly why your store times are increasing.


At 08:24 PM 08-10-03, Hawkins, Jim (ESS TCE Q&W) wrote:
No one seems to have yet mentioned the STATISTICS option, though it's mentioned often in connection with other threads. Looking at STDLIST and comparing the transferred bytes and I/O rates and elapsed time probably would give you some clues. If transferred bytes doesn't increase then check for new or changed jobs which might be stealing I/O or CPU bandwidth during back-up. There also can be subtle affects around compression features. Some folks with "extra" CPU can get better performance with STOREs COMPRESS options. Others get better results with enabling only the hardware compression -- enabling both historically could do weird things like actually EXPAND your data (and on 7980s actually corrupt it IIRC, no hard reference I can point to.) If my fingers remember correctly DEVCNTRL.MPEXL.TELESUP is one way to check on the hardware settings of the device.

-------------------------------------------------------------------------------------------------
Gilles Schipper
GSA Inc.
HP System Administration Specialists
300 John Street, Box 87651 Thornhill, ON Canada L3T 7R4
Voice: 905.889.3000 Fax: 905.889.3001
email: gilles@xxxxxxxxxx web: http://www.gsainc.com
-------------------------------------------------------------------------------------------------

* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *

.



Relevant Pages

  • Re: Table Compression, when to use it
    ... I frequently see compressed tables used on Linux boxes ... which have notoriously low I/O capacity. ... the analog boxes feeding the database were still too fast. ... compression happens only for the new blocks. ...
    (comp.databases.oracle.server)
  • New entry for Documentation for 2.4.25 (iostats.txt)
    ... +disk statistics have been introduced to help measure disk activity. ... +system-wide stats you'll have to find all the devices and sum them all up. ... as only one I/O. ...
    (Linux-Kernel)
  • Re: New means to compress random binary data, lossless and repeatable
    ... increase if we allow for 'flexible' number of cycles. ... size of this string we have then got for the counting the cycles. ... continiously elude compression. ... Using probabilities and statistics means we wont always have an ...
    (comp.theory)
  • Re: Huffmans use in various compressions
    ... > transmit the statistics (as in if they were some how known by the ... Example use a standard way to encode 4 symbols for compression ... optimal decompression this file must decompress into a file that makes ...
    (comp.compression)
  • Re: Arithmetic Assistance
    ... chance of having a non-compressible result. ... compression saved, I am now convinced of the 'no random recursive ... computed from that - in fact, it's the entropy of the source statistics. ...
    (comp.compression)