Veritas-bu

[Veritas-bu] LTO testing

2002-03-19 16:07:03
Subject: [Veritas-bu] LTO testing
From: Andrew Shinkarev <shinkara AT pprd.abbott DOT com> (Andrew Shinkarev)
Date: Tue, 19 Mar 2002 15:07:03 -0600
On Tue, Mar 19, 2002 at 03:02:19PM -0500, erbilgic.ym AT pg DOT com wrote:

LTO and SDLT are linear tape technology. However, 
LTO is claimed to be variable speed technology, while SDLT 
is still single speed. If SDLT is not feed with at least 
70% of nominal ( 0.7*11MB/s uncompressed ) it starts backhitching 
or shoeshining and trashing your tapes in big numbers. 
New cartridge for SDLT should address this issue, DLT IV
couldn't do any good with this issue.  

LTO can operate at 5/10/15 MB/s, so it should be much better 
for network backups. I do belive all LTO vendors implemented
this feature already. IBM was the last one. 

Quantum announced this feature for DLT8000 but never actually i
implemented it. I didn't see any announce for SDLT. 

It's just theory. In practice DLT7000 does trash a lot of tapes
when your tape drive is not a bottleneck. 


> 
> I have used/tested all 3 providers.  Up to now best performance was the HP 
> LVD.
> Seagate had a lot of time-out problems showing as 84 etc. on our SGI, fixed...
> I think we are one of the first to get Seagate's (or any lto) to work over 
> fibre
> on an SGI with Netbackup...  Took 3 months or so (immature drivers, firmware
> etc)
> 
> HP, does not certify (or recommend) their drives to be used over fibre/fibre
> bridges, but I was successful with the 3250 LVD.
> HP also had a problem that made direct attached servers reboot with the 
> firmware
> :)  That was a hard one to find...
> 
> IBM's seemed fine, tested on NT.  IBM's also have a neat display etc, can come
> handy, easier to figure out what is going on.
> 
> Seagate's 1mb/s additional speed claim, I do not think is anything to go by, I
> still do not think LTO is mature on all environments, has to be carefully 
> tested
> before put on production.  Every single time I installed an LTO drive for the
> first time it took me a week to 2-3 months to get it working right.  But 
> again,
> we started testing while the technology was not mature + using it on less 
> common
> configs.
> 
> But performance is definitely better than DLT...  Does anyone have DLT vs LTO
> (drive and media) reliability numbers, can not seem to find anything worth 
> while
> (e.g. Real user production data, rather than mtbf numbers).
> 
> I would not mind getting together and discussing (mail,phone etc) LTO stories
> and tests with other implementors.  Let me know if you are interested.
> 
> I hope this information helps.
> 
> Regards,
> 
> Yucel
> 
> ---------------------------------------------------------
> Yucel M. Erbilgic
> erbilgic.ym AT pg DOT com
> (513) 945-1859
> 
> 
> _______________________________________________
> Veritas-bu maillist  -  Veritas-bu AT mailman.eng.auburn DOT edu
> http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu

-- 
----
Andrew Shinkarev
Network Systems Specialist
Abbott Laboratories
(847) 9387559
shinkara AT pprd.abbott DOT com

<Prev in Thread] Current Thread [Next in Thread>