Backup Central

I know you are busy and have better things to do than reply to this post. But, do you remember where you were a newbie too?

Come on, be my hero.

-----------------------------------------------------------------------------------

Hi,

I am new to Backup, DR and Dedupe and have a question about full backups, data dedupe and restore.

In the company I work, one of my coworkers is pushing the idea of implementing data deduplication. Currently, we are required to do daily full backups to send to Iron Mountain. We use a two weeks rotation on our backups and store a month end and yearly backup on permanent storage.

I agree about data dedupe as an option to reduce the backup time window but I have a few questions about it.

1) On a total DR, if I don't do outside replication, is it possible to recover from backup tapes? If so, how?
2) If I do data dedupe with my daily full backups, do I then copy the duduped data to tape to send to IM daily?
3) Is Outside replication my only option for DR purposes?

I have not been able to find answers to all these question, so your help will be really appreciated.

JD

I've done a lot of testing with De-dup devices over the last couple of years. Primarily I've tested a DataDomain DDR690, and a number of Quantum unites, including DXi5500, DXi7500, and most recently a DXi6550.

The number one biggest issue I've seen with all de-duplication technologies is the ability read the de-duplicated data. Most of them can read and de-dup data faster then they can re-constitute the data into a usable format. DataDomain by far does a better job of re-constituting the data compared to everything else I've tested. That's not to say that other will work well enough for your needs, but you'll have to test them out to see what will work for you.

Another big difference I've seen between Quantum and DataDomain is that Quantum's performance is reliant on disk while DataDomain is more tied to CPU and memory.

Quantum try's to off-set their shortcomings with reading from de-dup by using a caching area were they keep the most recently accessed data in a non-de-duplicated format. While this can give them a high level of performance while caching, you'll still potentially have issues with reading older data, or if you replicate off-site for DR, it could impact your RTO's. Of course an improperly sized DataDomain could give you the same issues.

The biggest issue with DataDomain right now is the big question of what is EMC going to ultimately end up doing with them.

Software bases solutions are also an option, but you have to consider the added overhead they will put on the system de-duping the data. There are too many variables in this area to give you any ideas what might work, other then to say that you'll need to test them out and see what works for your environment.

As for getting data off-site, you have different options, some of which you've already recognized:

1. Backup to disk/de-dup and tape simultaneously so you have an on-site and off-site copy.
2. Backup to disk/de-dup then replicate/duplicate/copy to tape.
3. Most de-dup devices and even some software has the ability to replicate to another device from the same vendor that is at an off-site location. There are even some "on-line/cloud" backup services that will do this for you and provide you with a device for use on-site.

Personally I've found the real advantage to de-dup is around having a lot of data stored on-site and similtanously replicated off-site for DR without needing to use tape. It's more efficient, but not necessarily more cost effective. It also most likely will not replace your long term retention tapes, unless you data is extremely static.

Hi Spaldam,

Thank you so much for sharing your experience in a very detailed way.

Josue