Re: [Networker] Compression Question
2013-04-19 07:13:46
This is correct. Just to elaborate a bit on this, many moons ago when there
were reel to reel 2400 ft tapes, when you got an error it did indeed skip over
the bad spot and attempt to write again. Back in those days, the end of the
physical tape was marked by a reflective spot such that once it passed under
the read/write head, the end of tape indicator was raised and after the write
of that block was successful, eot kicked in and the system dismounted the tape
and called for another one. If however, the block was really large and a good
piece of tape was not found, it would keep skipping and attempting to write
until ultimately, it sucked the end of the tape off of the reel and into the
vacuum columns and the job died. We had an application that wrote really large
blocks and we had a lot of bad tapes so we zapped the blocksize to 1/4 the
original size to heighten the chances that the write of the smaller block had a
good chance of success. It Did. Now, it is handled electronically and with less
manual tape handling, the tapes don't get damaged much, but if the backup
product writes the same bock sizes day after day and the same tapes are always
used it could be possible that they could wear and start seeing that skip
condition. Bill Now adays that same function is handled electronically.
> Date: Fri, 19 Apr 2013 01:09:53 -0700
> From: networker-forum AT BACKUPCENTRAL DOT COM
> Subject: [Networker] Compression Question
> To: NETWORKER AT LISTSERV.TEMPLE DOT EDU
>
> It was old tape practice that in case of a write error, a longer gap was
> created to pass that area.
> Although i do not know exactly how tape drives are controlled today, this
> procedure should still be implemented (in their firmware).
> This totally transparent to the user like handling a bad sector on disk.
> So many write errors will affect the tape capacity as well.
>
> NW will only assume a tape error and set the media to full if it has
> encountered 20 (default) CONSECUTIVE write errors.
>
> +----------------------------------------------------------------------
> |This was sent by carsten_reinfeld AT avus-cr DOT de via Backup Central.
> |Forward SPAM to abuse AT backupcentral DOT com.
> +----------------------------------------------------------------------
|
|
|