Amazon S3 for TSM storage

[dhs3]

Hi all,

I recently read an interesting implementation guide using Riverbed's Cloud Gateway appliance to store backup data in Amazon's S3 Cloud. The guide is located here:

http://www.riverbed.com/assets/media/documents/briefs/SolutionBrief-Riverbed-Tivoli.pdf

I was wondering if anyone has tried this, or a similar solution. I envision the solution as using the S3 cloud as the primary sequential access pool using a FILE type device class. This would keep disk as the primary random access pool with the S3 storage as the next pool. One of my concerns is what impact this would have on reclamations since the data would need pulled and pushed for this process. In the guide linked above, they suggest a reclamation threshold of 80 or greater.

Any thoughts would be appreciated.

Thank you,

Don

 

[dharr18]

My personal thoughts.

You will need to look at bandwidth. Can your company's internet link handle the traffic going to S3?

I am leery of keeping primary storage pools in the cloud, you are now dependent on S3's availability. Any outage, internal or external, will cause many volumes to be marked unavailable or read only.

As a copy pool, I am slightly more receptive to cloud storage. The same issues are there, but it will not affect the recover-ability of your clients.

Make sure you do discuss Disaster Recovery if used as a copy storage pool. You would need to plan on having another Whitewater box in your DR location. If your company does frequent DR tests, or any DR testing. Discuss with Riverbed how this can be accomplished without affecting production offsites. Reclamation can cause havoc when the copy storage pool is reclaimed and the production server does not know of the changes.

As a last thought, crunch the numbers based on GB or TB cost compared to local storage. A reclamation at 80% will use considerable cloud resources/storage. As pertaining to my environment, that would more than double the storage use/defined to TSM. Remember, even though TSM isn't tracking the space, you are still using the space.

If the numbers are justifiable, I would look at using smaller file sizes in the DEV type to reduce the actual consumption of cloud storage.

 

[moon-buddy]

Hi all,

I recently read an interesting implementation guide using Riverbed's Cloud Gateway appliance to store backup data in Amazon's S3 Cloud. The guide is located here:

http://www.riverbed.com/assets/media/documents/briefs/SolutionBrief-Riverbed-Tivoli.pdf

I was wondering if anyone has tried this, or a similar solution. I envision the solution as using the S3 cloud as the primary sequential access pool using a FILE type device class. This would keep disk as the primary random access pool with the S3 storage as the next pool. One of my concerns is what impact this would have on reclamations since the data would need pulled and pushed for this process. In the guide linked above, they suggest a reclamation threshold of 80 or greater.

Any thoughts would be appreciated.

Thank you,

Don

My two-cents:

DO NOT USE any storage that you can't control.

Your business survival depends on how quick you can recover WITH the least dependency on others.

Just remember - third party providers will always have fine prints in the contract wherein 'beyond their control' clauses and statements exist. As a Security Professional, I would not even entertain the idea.

 

[dharr18]

My two-cents:

DO NOT USE any storage that you can't control.

Your business survival depends on how quick you can recover WITH the least dependency on others.

Just remember - third part providers will always have fine prints in the contract wherein 'beyond their control' clauses and statements exist. As a Security Professional, I would not even entertain the idea.

You said it much more succinctly than I did.

 

[dhs3]

Thank you both for your feedback. I'm currently looking for a storage solution for our Tier-2 data which does not require disaster recovery or have a defined RTO. The cost, availability, and overall management are all valid considerations and will require further study. I agree that using the cloud as a copy pool is preferred.

Thanks again!

 

[skippet0306]

does TSM v7 support the amazon s3 or RESTAPI now as a backed backup repository?