ibm flash copy, hp business copy

[yls177]

[email]dfreybur@yahoo.com[/email] (Doug Freyburger) wrote in message news:<7960d3ee.0406080727.42c9875a@posting.google. com>...

> yls177 wrote:

> >
> > hi, basically, both of the above in the subject line achieves the same
> > objective using 2 san disks storage for backup/restore. any insights
> > to the pros and cons?

>
> IBM flash copy does a checkpoint of the current state into a copy.
> As changes are made, new storage must be consumed in the original
> to preserve that snapshot, or is it new blocks must be added to the
> snapshot. Anyways, to have a full copy that does not effect the
> new changing state, a mirror must be made. So flash copy is fast
> to create but it still needs a full mirror if you intend to keep
> it around for long. In particular a reverse copy might take a
> very short time to restore state or it might be a full reverse
> copy depending on the state when the reverse is done.
>
> HP business copy does a full mirror of the current state into a
> copy. As changes are made to the original, the business copy
> maintains a record of what blocks change. So a reverse copy
> only needs to rewrite the map of changed blocks. Also the next
> forward copy only needs to copy the changed blocks. Thus the
> initial copy will take a long time, but if your data sees 10%
> of its blocks change each day, subsequent syncs will take 10% as
> long as the first.
>
> The business copy mirror does not consume new storage as changes
> are made the way flash, but business copy requires the copy to
> be 100% the size of the original. So your trade-off is known
> storage requirements that are large vs unknown storage requirements
> that might be small.

but in either case, we still need the 2 same SAN storage available... isnt it?

[Doug Freyburger]

yls177 wrote:

> Doug Freyburger wrote:

> > yls177 wrote:

>

> > > hi, basically, both of the above in the subject line achieves the same
> > > objective using 2 san disks storage for backup/restore. any insights
> > > to the pros and cons?

>

> > IBM flash copy does a checkpoint of the current state into a copy.
> > As changes are made, new storage must be consumed in the original
> > to preserve that snapshot, or is it new blocks must be added to the
> > snapshot. Anyways, to have a full copy that does not effect the
> > new changing state, a mirror must be made. So flash copy is fast
> > to create but it still needs a full mirror if you intend to keep
> > it around for long. In particular a reverse copy might take a
> > very short time to restore state or it might be a full reverse
> > copy depending on the state when the reverse is done.

>

> > HP business copy does a full mirror of the current state into a
> > copy. As changes are made to the original, the business copy
> > maintains a record of what blocks change. So a reverse copy
> > only needs to rewrite the map of changed blocks. Also the next
> > forward copy only needs to copy the changed blocks. Thus the
> > initial copy will take a long time, but if your data sees 10%
> > of its blocks change each day, subsequent syncs will take 10% as
> > long as the first.

>

> > The business copy mirror does not consume new storage as changes
> > are made the way flash, but business copy requires the copy to
> > be 100% the size of the original. So your trade-off is known
> > storage requirements that are large vs unknown storage requirements
> > that might be small.

>
> but in either case, we still need the 2 same SAN storage available... isnt it?

In the HP business copy case you need 200% of the original space.
That makes it's space consumption predictable.

In the IBM flash copy case it depends on exactly what you do.
If you do make a complete extra copy from the checkpoint you need
very close to 200% of the original space. If you do not make a
complete extra copy and just depend on the checkpoint your
consumption depends entirely on your rate of change and it has a
worst case scenario of 200%. I have no idea what happens if you
notice that your average amount of change is 5% therefore you use
a flash pool that's 10% but one day you get 11% change and
overwhelm the pool.

[yls177]

[email]dfreybur@yahoo.com[/email] (Doug Freyburger) wrote in message news:<7960d3ee.0406100741.6f8be09d@posting.google. com>...

> yls177 wrote:

> > Doug Freyburger wrote:

> > > yls177 wrote:

>

> > > > hi, basically, both of the above in the subject line achieves the same
> > > > objective using 2 san disks storage for backup/restore. any insights
> > > > to the pros and cons?

>

> > > IBM flash copy does a checkpoint of the current state into a copy.
> > > As changes are made, new storage must be consumed in the original
> > > to preserve that snapshot, or is it new blocks must be added to the
> > > snapshot. Anyways, to have a full copy that does not effect the
> > > new changing state, a mirror must be made. So flash copy is fast
> > > to create but it still needs a full mirror if you intend to keep
> > > it around for long. In particular a reverse copy might take a
> > > very short time to restore state or it might be a full reverse
> > > copy depending on the state when the reverse is done.

>

> > > HP business copy does a full mirror of the current state into a
> > > copy. As changes are made to the original, the business copy
> > > maintains a record of what blocks change. So a reverse copy
> > > only needs to rewrite the map of changed blocks. Also the next
> > > forward copy only needs to copy the changed blocks. Thus the
> > > initial copy will take a long time, but if your data sees 10%
> > > of its blocks change each day, subsequent syncs will take 10% as
> > > long as the first.

>

> > > The business copy mirror does not consume new storage as changes
> > > are made the way flash, but business copy requires the copy to
> > > be 100% the size of the original. So your trade-off is known
> > > storage requirements that are large vs unknown storage requirements
> > > that might be small.

> >
> > but in either case, we still need the 2 same SAN storage available... isnt it?

>
> In the HP business copy case you need 200% of the original space.
> That makes it's space consumption predictable.
>
> In the IBM flash copy case it depends on exactly what you do.
> If you do make a complete extra copy from the checkpoint you need
> very close to 200% of the original space. If you do not make a
> complete extra copy and just depend on the checkpoint your
> consumption depends entirely on your rate of change and it has a
> worst case scenario of 200%. I have no idea what happens if you
> notice that your average amount of change is 5% therefore you use
> a flash pool that's 10% but one day you get 11% change and
> overwhelm the pool.

we did a flash copy everyday. its an exact copy of our production

cheers