|
Home > Archive > Data Storage > May 2005 > How Do You Back Up A Terrabyte Of Data ?
You are viewing an archived Text-only version of the thread.
To view this thread in it's original format and/or if you want to reply to
this thread please [click here]
| Author |
How Do You Back Up A Terrabyte Of Data ?
|
|
| Al Dykes 2005-05-23, 5:50 pm |
|
Other than second-site replication how do people with multi-TB of data
do disaster planning these days ? Do poeple just buy pairs of SAN/NAS
boxes ? I'd want three, actually. Sh*t happens.
I managed datacenters when disks went from 25MB to maybe 800MB and
9 track tapes were good enough for backup. I've been away from
this stuff for a long time.
I know that databases can also replicate updates buy that doesn't
make the problem any easier.
--
a d y k e s @ p a n i x . c o m
Don't blame me. I voted for Gore.
| |
| Thor Lancelot Simon 2005-05-23, 5:50 pm |
| In article <d6sob0$9s6$1@panix5.panix.com>, Al Dykes <adykes@panix.com> wrote:
>
>Other than second-site replication how do people with multi-TB of data
>do disaster planning these days ? Do poeple just buy pairs of SAN/NAS
>boxes ? I'd want three, actually. Sh*t happens.
Lots of tapes, or backup over the net to somewhere far away. Some people
use removable disk drives but the problem is that you can't be sure they'll
spin up when you need them -- they're a lot less physically robust than
tapes are.
--
Thor Lancelot Simon tls@rek.tjls.com
"The inconsistency is startling, though admittedly, if consistency is to be
abandoned or transcended, there is no problem." - Noam Chomsky
| |
| boatgeek 2005-05-24, 5:53 pm |
| We replicate around 8 TBs of data. Depends on your service license
agreement. If the client can afford to be down for up to a couple days
to recover the data from tape, then you could basically have two data
centers with a large pipe between them and then back up site A at site
B and backup site B at site A. We do that for non critical systems and
those that can't afford a full live DR scenario.
For the people who have to be online immediately but can accept a
performance degradation and don't want to have double the costs of a
single system for every DR device, we replicate primary NAS storage
(NFS and CIFS) to secondary nearstore storage. NetApp nearstores have
identical functionality to their top of the line filers, so you can
have a ATA nearstore device capable of sharing out the CIFS and NFS
data in a transparant manner to the primary in the event of a DR.
We've actually used this in a true disaster (15 minutes warning and the
primary data center is gone), and had everyone backup and working in
the DR in 10 minutes doing some quick aliasing on the DR filers.
For those who need to be up without and performance degradation, then
you replicate to another higher tiered storage device at the DR, just
let them know that it will cost more than twice the costs of just the
primary device alone (because you're paying for the snapshot license).
| |
|
| On 23 May 2005 10:11:44 -0400, adykes@panix.com (Al Dykes) wrote:
>
>Other than second-site replication how do people with multi-TB of data
>do disaster planning these days ? Do poeple just buy pairs of SAN/NAS
>boxes ? I'd want three, actually. Sh*t happens.
>
There are a few different alternatives for backups and recovery (note
the 2nd part there, backups are no good if your recovery plan doesn't
work). No one solution is right for every situation... and sometimes
a combination is required for different platforms within an
environment.
The main thing to start with is the business value of the applications
& data, and the amount of loss during an outage. Each business will
have different values for their infrastructure.
The key then is to invest the correct amount that meets the business
value, and the needs for recovery time objectives (RTO) and recovery
point objectives (RPO). Sometimes you might also need to factor in
the return recovery time objective (RRTO), since going back to your
production environment might actually take longer than the original
recovery (shutdown systems, backup, ship backups to site, restore,
test, etc).
Some environments do well just running a normal backups to tape.
Others can't take that much of a hit during backups, so they utilize
some form of replication (e.g., snapshots, replication, etc.).
For some environments, the time it takes to restore 1TB from tape is
too long for their service level objectives, so they backup to disk
(either directly to disk, or with one of the virtual tape libraries
that are really disk storage behind the scenes). This speeds the
backup as well as the recovery. The downside, though, is that you
can't send these backups offsite - usually a copy-to-tape operaton is
required for that.
Then there's getting into more full-featured replication scenarios
(SAN, NSI, etc).
--- jls
The preceding message was personal opinion only.
I do not speak in any authorized capacity for anyone,
and certainly not my employer.
|
|
|
|
|