HBS3 recommended on a Main/Backup server setup

[RayMcBe]

Since the last update in August with QTS, I've been having problems with backup which prompted me to rethink how I should be setting up HBS3. Asking for advice:

• I have a Main and Backup server. The Main server contains our standard file share - basically a single large 185TB network share. Should I setup the backup job on the Backup Server or the Main Server? Currently, the job is setup on the Main server and points to store the files on the Backup server.

• The Backup server is intended to hold only a copy of our Main server (currently the server has about 1.5x the amount of total storage than the Main). Right now for the Backup server, I have two volumes setup, one that's 235TB and the other that's 150TB - setup on an expansion unit (due to volume size limitations). How would you recommend I span the job across both units if it's even possible?

• What service should I be using? The servers sit in the same rack and are connected on the same switch, so speed isn't an issue for connectivity. Currently, I've been using RTRR.

• I'm also planning a version schedule of retaining the last 4 hours, 7 days, 8 weeks and 10 months. I imagine that if I run out of space, it'll remove one of these prior versions?

Other thoughts/advice to share?

Thanks in advance.

[P3R]

• I have a Main and Backup server. The Main server contains our standard file share - basically a single large 185TB network share. Should I setup the backup job on the Backup Server or the Main Server? Currently, the job is setup on the Main server and points to store the files on the Backup server.

You also mention versioning in your post and since that's only possible from the main unit I think you have your answer there.

• The Backup server is intended to hold only a copy of our Main server (currently the server has about 1.5x the amount of total storage than the Main). Right now for the Backup server, I have two volumes setup, one that's 235TB and the other that's 150TB - setup on an expansion unit (due to volume size limitations). How would you recommend I span the job across both units if it's even possible?

It have to be manually managed with some folder pairs pointing to the main unit and some to the expansion.

• What service should I be using? The servers sit in the same rack and are connected on the same switch, so speed isn't an issue for connectivity. Currently, I've been using RTRR.

I'm not aware of anything that's better so RTRR is what I would recommend.

• I'm also planning a version schedule of retaining the last 4 hours, 7 days, 8 weeks and 10 months. I imagine that if I run out of space, it'll remove one of these prior versions?

It won't remove anything automatically. It will try to save as many versions as you tell it and it's up to you to make sure the destination have enough capacity.

I did the mistake of configuring to many versions for my backup destination and had to remove some when I ran out of backup space. My advice is to start with less versions and work your way up to what you ultimately want. You could try 4 hours, 7 days, 4 weeks and 2 months and see how that work out to begin with. Then add weeks and months as long as you still have a healthy safety margin.

[RayMcBe]

Thanks P3R for your insight. Are you suggesting the healthy safety margin refering to space or the number of versions? If regarding storage, how much space would be a healthy safety margin? Could be a percentage...

[P3R]

Are you suggesting the healthy safety margin refering to space or the number of versions?

Space. The number of versions possible will be limited by what you have capacity for. If you need more versions, you'll need to increase capacity.

If regarding storage, how much space would be a healthy safety margin? Could be a percentage...

Hard to say and that's why I suggest that you increase the number of versions gradually to get a feeling for approximately how much each weekly and montly version use. When you feel confident about that, you could probably even go below 5 % free space if you have much static data.