SEO

Hello All,
We are encountering a situation where backups of our database take far to
long to complete (IMHO).

Upwards of two hours to tape with a two hour restore time.

This is fine from a protection standpoint as it is the most complete option,
however, I'm looking for something that would give us more agility.
For example, during a recent outage caused by running out of data pages on
one particular table one of our options was to restore from tape, at a cost
of two hours.

Optimally, I would have liked to have been able to restore simply the data
in the offending table to another disk/space etc without performing a full
restore from tape.

What option would give me this flexibility (hardware independent)?
I say hardware independent because I was recently in a data center with new
Net App gear capable (unbelievable as it sounds..) of fully restoring a 1TB
Oracle instance in approximately 10 seconds ("flashing" a snapshot if you
will...).

Thoughts?


Thanks.

Tam.

Tam O'Shanter said:
> Hello All,
> We are encountering a situation where backups of our database take far to
> long to complete (IMHO).
>
> Upwards of two hours to tape with a two hour restore time.

How big is it? (Your database, I mean!)

I have a client who backs up 110GB onto DLT in 20 minutes.

> This is fine from a protection standpoint as it is the most complete
> option,
> however, I'm looking for something that would give us more agility.
> For example, during a recent outage caused by running out of data pages on
> one particular table one of our options was to restore from tape, at a
> cost
> of two hours.
>
> Optimally, I would have liked to have been able to restore simply the data
> in the offending table to another disk/space etc without performing a full
> restore from tape.

You should attend one of the many IIUG sessions on this topic. ;o)

> What option would give me this flexibility (hardware independent)?
> I say hardware independent because I was recently in a data center with
> new
> Net App gear capable (unbelievable as it sounds..) of fully restoring a
> 1TB
> Oracle instance in approximately 10 seconds ("flashing" a snapshot if you
> will...).

Have you considered something like an IBM ESS array which will allow you
to fully restore a 1TB IDS instance in approximately 10 seconds? ?

> Thoughts?

onbar will allow parallel backups to improve speed of backup and restore,
but there are all sorts of consequences of this. What mechanism are you
using? What kind of tape are you backing up to? What block size are you
using?

--
Bye now,
Obnoxio

"... no bill is required as no value was provided."
-- Christine Normile

--
This message has been scanned for viruses and
dangerous content by OpenProtect(http://www.openprotect.com), and is
believed to be clean.

133943152 pg - 286.29 GB
6 physical disks mirrored.

Ulritum 2 Tape Drive
200/400 GB

ontape, nightly full dump.

Taper block of 512

I have investigated the quality of the data and so far isolated upwards of
70GB that can be removed from this system (time sensitive stuff that's over
2 years old when we only need 13 months...).

It would be nice to have a nonintrusive way of removing this data from our
system, and to this end I'm still investigating hpl (good idea right??).

Thoughts?


BTW...
You sir, are no clown.

Tam.


"Obnoxio The Clown" <obnoxio@serendipita.com> wrote in message
news:mailman.156.1162381899.29126.informix-list@iiug.org...
>
> Tam O'Shanter said:
>> Hello All,
>> We are encountering a situation where backups of our database take far to
>> long to complete (IMHO).
>>
>> Upwards of two hours to tape with a two hour restore time.
>
> How big is it? (Your database, I mean!)
>
> I have a client who backs up 110GB onto DLT in 20 minutes.
>
>> This is fine from a protection standpoint as it is the most complete
>> option,
>> however, I'm looking for something that would give us more agility.
>> For example, during a recent outage caused by running out of data pages
>> on
>> one particular table one of our options was to restore from tape, at a
>> cost
>> of two hours.
>>
>> Optimally, I would have liked to have been able to restore simply the
>> data
>> in the offending table to another disk/space etc without performing a
>> full
>> restore from tape.
>
> You should attend one of the many IIUG sessions on this topic. ;o)
>
>> What option would give me this flexibility (hardware independent)?
>> I say hardware independent because I was recently in a data center with
>> new
>> Net App gear capable (unbelievable as it sounds..) of fully restoring a
>> 1TB
>> Oracle instance in approximately 10 seconds ("flashing" a snapshot if you
>> will...).
>
> Have you considered something like an IBM ESS array which will allow you
> to fully restore a 1TB IDS instance in approximately 10 seconds? ?
>
>> Thoughts?
>
> onbar will allow parallel backups to improve speed of backup and restore,
> but there are all sorts of consequences of this. What mechanism are you
> using? What kind of tape are you backing up to? What block size are you
> using?
>
> --
> Bye now,
> Obnoxio
>
> "... no bill is required as no value was provided."
> -- Christine Normile
>
> --
> This message has been scanned for viruses and
> dangerous content by OpenProtect(http://www.openprotect.com), and is
> believed to be clean.
>

Tam O'Shanter wrote:
> Hello All,
> We are encountering a situation where backups of our database take far to
> long to complete (IMHO).
>
> Upwards of two hours to tape with a two hour restore time.
>
> This is fine from a protection standpoint as it is the most complete option,
> however, I'm looking for something that would give us more agility.
> For example, during a recent outage caused by running out of data pages on
> one particular table one of our options was to restore from tape, at a cost
> of two hours.
>
> Optimally, I would have liked to have been able to restore simply the data
> in the offending table to another disk/space etc without performing a full
> restore from tape.

VERSION INFO IS SOOOOO HELPFUL!

If you are running a later release of IDS 9.40 or and IDS 10.00 release the
include archecker utility is capable of extracting a single table's data and
restoring it using SQL commands so that it can be filtered and restored to a
temporary tablename on the same or a different server.

> What option would give me this flexibility (hardware independent)?
> I say hardware independent because I was recently in a data center with new
> Net App gear capable (unbelievable as it sounds..) of fully restoring a 1TB
> Oracle instance in approximately 10 seconds ("flashing" a snapshot if you
> will...).

Art S. Kagel

ontape has the ability to warm restore dbspaces (this requires that you
also backup
you log.logs!!!!!!!!!!!!!)
so if you make your dbspaces not too big you could look at that.



i assume you are using char mode raw devs, if cooked, it can really
hurt performance
when restoring. the filesystemcache is trashed in that case....

remove data based on time.... have a look at fragment a table based on
a col
with date info..... detach tables; dropping unneeded is far faster then
delete from

And yes use HPL (a.k.a. Nuke Obstacle tool)
if you need to setup a fragment strategy for big tables!!

Superboer.


Tam O'Shanter schreef:

> 133943152 pg - 286.29 GB
> 6 physical disks mirrored.
>
> Ulritum 2 Tape Drive
> 200/400 GB
>
> ontape, nightly full dump.
>
> Taper block of 512
>
> I have investigated the quality of the data and so far isolated upwards of
> 70GB that can be removed from this system (time sensitive stuff that's over
> 2 years old when we only need 13 months...).
>
> It would be nice to have a nonintrusive way of removing this data from our
> system, and to this end I'm still investigating hpl (good idea right??).
>
> Thoughts?
>
>
> BTW...
> You sir, are no clown.
>
> Tam.
>
>
> "Obnoxio The Clown" <obnoxio@serendipita.com> wrote in message
> news:mailman.156.1162381899.29126.informix-list@iiug.org...
> >
> > Tam O'Shanter said:
> >> Hello All,
> >> We are encountering a situation where backups of our database take far to
> >> long to complete (IMHO).
> >>
> >> Upwards of two hours to tape with a two hour restore time.
> >
> > How big is it? (Your database, I mean!)
> >
> > I have a client who backs up 110GB onto DLT in 20 minutes.
> >
> >> This is fine from a protection standpoint as it is the most complete
> >> option,
> >> however, I'm looking for something that would give us more agility.
> >> For example, during a recent outage caused by running out of data pages
> >> on
> >> one particular table one of our options was to restore from tape, at a
> >> cost
> >> of two hours.
> >>
> >> Optimally, I would have liked to have been able to restore simply the
> >> data
> >> in the offending table to another disk/space etc without performing a
> >> full
> >> restore from tape.
> >
> > You should attend one of the many IIUG sessions on this topic. ;o)
> >
> >> What option would give me this flexibility (hardware independent)?
> >> I say hardware independent because I was recently in a data center with
> >> new
> >> Net App gear capable (unbelievable as it sounds..) of fully restoring a
> >> 1TB
> >> Oracle instance in approximately 10 seconds ("flashing" a snapshot if you
> >> will...).
> >
> > Have you considered something like an IBM ESS array which will allow you
> > to fully restore a 1TB IDS instance in approximately 10 seconds? ?
> >
> >> Thoughts?
> >
> > onbar will allow parallel backups to improve speed of backup and restore,
> > but there are all sorts of consequences of this. What mechanism are you
> > using? What kind of tape are you backing up to? What block size are you
> > using?
> >
> > --
> > Bye now,
> > Obnoxio
> >
> > "... no bill is required as no value was provided."
> > -- Christine Normile
> >
> > --
> > This message has been scanned for viruses and
> > dangerous content by OpenProtect(http://www.openprotect.com), and is
> > believed to be clean.
> >

hum sorry forget the remarks on ontape since i forgot to read

that you ran out of 16 GB...--->> do a fragmentation strategy...
use HPL.

Superboer.

Superboer schreef:

> ontape has the ability to warm restore dbspaces (this requires that you
> also backup
> you log.logs!!!!!!!!!!!!!)
> so if you make your dbspaces not too big you could look at that.
>
>
>
> i assume you are using char mode raw devs, if cooked, it can really
> hurt performance
> when restoring. the filesystemcache is trashed in that case....
>
> remove data based on time.... have a look at fragment a table based on
> a col
> with date info..... detach tables; dropping unneeded is far faster then
> delete from
>
> And yes use HPL (a.k.a. Nuke Obstacle tool)
> if you need to setup a fragment strategy for big tables!!
>
> Superboer.
>
>
> Tam O'Shanter schreef:
>
> > 133943152 pg - 286.29 GB
> > 6 physical disks mirrored.
> >
> > Ulritum 2 Tape Drive
> > 200/400 GB
> >
> > ontape, nightly full dump.
> >
> > Taper block of 512
> >
> > I have investigated the quality of the data and so far isolated upwards of
> > 70GB that can be removed from this system (time sensitive stuff that's over
> > 2 years old when we only need 13 months...).
> >
> > It would be nice to have a nonintrusive way of removing this data from our
> > system, and to this end I'm still investigating hpl (good idea right??).
> >
> > Thoughts?
> >
> >
> > BTW...
> > You sir, are no clown.
> >
> > Tam.
> >
> >
> > "Obnoxio The Clown" <obnoxio@serendipita.com> wrote in message
> > news:mailman.156.1162381899.29126.informix-list@iiug.org...
> > >
> > > Tam O'Shanter said:
> > >> Hello All,
> > >> We are encountering a situation where backups of our database take far to
> > >> long to complete (IMHO).
> > >>
> > >> Upwards of two hours to tape with a two hour restore time.
> > >
> > > How big is it? (Your database, I mean!)
> > >
> > > I have a client who backs up 110GB onto DLT in 20 minutes.
> > >
> > >> This is fine from a protection standpoint as it is the most complete
> > >> option,
> > >> however, I'm looking for something that would give us more agility.
> > >> For example, during a recent outage caused by running out of data pages
> > >> on
> > >> one particular table one of our options was to restore from tape, at a
> > >> cost
> > >> of two hours.
> > >>
> > >> Optimally, I would have liked to have been able to restore simply the
> > >> data
> > >> in the offending table to another disk/space etc without performing a
> > >> full
> > >> restore from tape.
> > >
> > > You should attend one of the many IIUG sessions on this topic. ;o)
> > >
> > >> What option would give me this flexibility (hardware independent)?
> > >> I say hardware independent because I was recently in a data center with
> > >> new
> > >> Net App gear capable (unbelievable as it sounds..) of fully restoring a
> > >> 1TB
> > >> Oracle instance in approximately 10 seconds ("flashing" a snapshot if you
> > >> will...).
> > >
> > > Have you considered something like an IBM ESS array which will allow you
> > > to fully restore a 1TB IDS instance in approximately 10 seconds? ?
> > >
> > >> Thoughts?
> > >
> > > onbar will allow parallel backups to improve speed of backup and restore,
> > > but there are all sorts of consequences of this. What mechanism are you
> > > using? What kind of tape are you backing up to? What block size are you
> > > using?
> > >
> > > --
> > > Bye now,
> > > Obnoxio
> > >
> > > "... no bill is required as no value was provided."
> > > -- Christine Normile
> > >
> > > --
> > > This message has been scanned for viruses and
> > > dangerous content by OpenProtect(http://www.openprotect.com), and is
> > > believed to be clean.
> > >