----- Original Message -----
From: <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Cc: <[EMAIL PROTECTED]>
Sent: Friday, January 18, 2002 7:17 PM
Subject: Re: tapetype entries
> It is my understanding that this drive was delivered with HW compression
> enabled. I have done nothing to disable it.
>
> After visiting their web site I see the problem...
>
> The tape is a V17. I did not recall, so I simply assumed that when you
> mentioned the V10, that it must have been the correct number.
>
> The definition should then read:
>
> define tapetype V17 {
> comment "V17 in ECRIX Model: VXA-1 Rev: 2959 w/HW compression"
> length 55000 mbytes
> filemark 1931 kbytes
> speed 2024 kps
> }
>
> Given this, Ecrix significantly overstates their capacity. For the V17,
> they advertise 33G uncompressed and I got 27.5G. That's almost 17%
> overstated.
>
The number you're stating is what tapetype gave you. I haven't looked at
it, but I assume its calculations deduct inter-record gaps and associated
overhead that it'd expect to incur when writing in normal "amanda use". As
such the capacity figure will be less than the figure Ecrix states which is
(probably) for one infinite length record. Whether or not 17% overhead is
correct I can't say.
> Is this common in the industry or is Ecrix out on a limb?
>
> Or am I misunderstanding the meaning of the numbers reported by
> tapetype?
>
Either you or me (or maybe both of us :-)). Usenix has been using an
autoPAK w/ V10 tapes for a while but I never had the time to let tapetype
run to completion to get an "accurate" entry. Instead I use a hack:
define tapetype VXA-V10 {
comment "VXA-1 V10 Cartridge (107 meters)"
length 20 gbytes # 40 GB with 2:1 compression
filemark 100 kbytes # pure guess
speed 3 mbytes # documented sustained transfer rate
}
and let amanda recover when it runs off the tape during multi-tape backups.
That's why your posting caught my eye; I thought I'd be able to drop your
entry in and eliminate this excess work.
> > FWIW I run w/o h/w compression; my Dell 2450 host is way faster at doing
> > gzip and I want the drive to stream whenever possible.
>
> Are you saying the drive does not stream with HW compression? How does
> one know when the drive is streaming and when it isn't?
>
No. I'm simply saying that I take no chances that the drive may not stream
by adding to the overhead of the controller. I assume the hardware is
designed to stream even with h/w compression but I've had experience with
some devices that slowed when doing compression. Besides, since gzip
compression tends to be as good or better than any hardware compression I
see no point in using h/w compression.
Oh, and you asked about my OS and how I disable hardware compression. I use
FreeBSD (4.3-RELEASE) and hardware compression is enabled or disabled by the
driver.
> Thanks,
> Dick
>
> P.S. Please reply to: [EMAIL PROTECTED]
>
> > BTW, was this really with h/w compression enabled? You state 26.86G of
> > writable space but the V10 cartridges are spec'd at 40G with
compression.
> > I'm curious because I have the same setup but never ran tapetype to
figure
> > out the actual capacity.
> >
> > FWIW I run w/o h/w compression; my Dell 2450 host is way faster at doing
> > gzip and I want the drive to stream whenever possible.
> >
> > Sam
> >
> > ----- Original Message -----
> > From: <[EMAIL PROTECTED]>
> > To: <[EMAIL PROTECTED]>
> > Cc: <[EMAIL PROTECTED]>
> > Sent: Friday, January 18, 2002 3:56 PM
> > Subject: Re: tapetype entries
> >
> >
> > > You're right. I wasn't thinking... I guess it should be:
> > >
> > > define tapetype V10 {
> > > comment "V10 in ECRIX Model: VXA-1 Rev: 2959 w/HW compression"
> > > length 55000 mbytes
> > > filemark 1931 kbytes
> > > speed 2024 kps
> > > }
> > >
> > >
> > > > > define tapetype vxa1 {
> > > > > comment "ECRIX Model: VXA-1 Rev: 2959"
> > > > > length 27513 mbytes
> > > > > filemark 1931 kbytes
> > > > > speed 2024 kps
> > > > > }
> > > > >
> > > >
> > > > The VXA-1 is a drive. This is specific to a cartridge; e.g. V10.
> > > >
> > > > Sam
> > > >
> > > >
> > >
> > >
> >
>
>
