Backup and Recovery Goals
-
@DustinB3403 said:
@Dashrender I am doing everything in my power to avoid Spinning Rust. The original proposal I made includes SSDs in RAID 5, 8TB in total.
Have you run something like a Dell Dpack to actually see how many IOPS you are using to know if you really will see any benefit from SSDs?
-
No, there's no need. We have nothing we "host" onsite that is IOPS intensive
But even with SSD's the price is still cheaper then what we've been quoted from others.
-
@Dashrender said:
Something to keep in mind. While every business wants as little downtime as possible and as little data lose as possible, while they may start with that goal, and give the times listed above, once you provide them numbers that provide that level of service, they may come back and change their minds and lower these requirements (i.e. make them longer).
Maybe with down time, but not so much loss of data, especially if you are a publicly traded company where you can not lose any data and have required data retention periods.
-
@DustinB3403 said:
The primary file ....... The server simply shares out the attached stored.
That's the case with any file server.
-
@Jason said:
@Dashrender said:
Something to keep in mind. While every business wants as little downtime as possible and as little data lose as possible, while they may start with that goal, and give the times listed above, once you provide them numbers that provide that level of service, they may come back and change their minds and lower these requirements (i.e. make them longer).
Maybe with down time, but not so much loss of data, especially if you are a publicly traded company where you can not lose any data and have required data retention periods.
Is there such a thing as zero data loss potential?
-
@Dashrender said:
@Jason said:
@Dashrender said:
Something to keep in mind. While every business wants as little downtime as possible and as little data lose as possible, while they may start with that goal, and give the times listed above, once you provide them numbers that provide that level of service, they may come back and change their minds and lower these requirements (i.e. make them longer).
Maybe with down time, but not so much loss of data, especially if you are a publicly traded company where you can not lose any data and have required data retention periods.
Is there such a thing as zero data loss potential?
Isn't that the dream?
-
@Dashrender said:
Is there such a thing as zero data loss potential?
Next to it when you have change monitoring, not just scheduled backups.
-
@DustinB3403 said:
The primary file server is directly connected to the Buffalo drive. The server simply shares out the attached stored.
The primary DC has multiple functions on it, Spiceworks, AD, and 2 network shares.
The 3rd server is acting as a HyperV host for 0365 / AD integration and a Reporting server.
That's an odd way to do it. I'd take the Spiceworks off the DC for sure and put in on the server with the o365 Azure sync. File server is better separate from a DC too.
-
@Dashrender said:
Considerations:
8 port 10 GigE switch with 10GigE in each server. Could probably be done with bonded 1 GigE ports instead, minimum of two per server.What kind of data are you pushing over this next work to need 10Gig? We don't even use 10Gig on our production network, we only have it for the iSCSI networks and some uplinks from the datacenter. All server connections are 1gig to the production network.
-
@Dashrender said:
Either 2 TB 7200 SAS drives in RAID 10 or 1 TB SSD drives in RAID 5 (8 TB per previous discussion, currently using 6 TB)
Have you calculated your data growth rate? that only leaves room for 1.33% growth.
-
@Jason You're telling me....
I've wanted to move it since I started....
-
@Jason said:
@Dashrender said:
Either 2 TB 7200 SAS drives in RAID 10 or 1 TB SSD drives in RAID 5 (8 TB per previous discussion, currently using 6 TB)
Have you calculated your data growth rate? that only leaves room for 1.33% growth.
1.33%? Don't you mean more like 33% growth, one third more than what you're currently using.
-
@Jason said:
@Dashrender said:
Considerations:
8 port 10 GigE switch with 10GigE in each server. Could probably be done with bonded 1 GigE ports instead, minimum of two per server.What kind of data are you pushing over this next work to need 10Gig? We don't even use 10Gig on our production network, we only have it for the iSCSI networks and some uplinks from the datacenter. All server connections are 1gig to the production network.
This speed was to get full backups accomplished in a reasonable timeframe. According to the last thread, Scott figured out that 1 GigE would take around 17 hours to backup the 6 TB of data. Currently Dustin is seeing that it takes 24+ hours considering their current setup. Moving to dual 10 GigE would reduce that to around 1 hour.
If Dustin moves to a different backup solution, he might move to synthetic full backups and not require the 10 GigE. But restore to meet the previously stated RTO might still require it.
But perhaps four 1 GigE bonded might be good enough as well.
-
From the other thread it was discussed that we could use LTO Tapes to create our weekly full backups. Is this the correct item?
Additionally how might I tie this into my plan for backup.
I still need to determine, to what device I'm going to backup our VM's and Data to either a R620 server in RAID 10 on SR with 3 or 4TB drives running a CIFS server for of management with the LTO Tape.
Any recommendations.
Unfortunately I'm outside of my experience with Tapes and how to tie them in.
-
@DustinB3403 said:
From the other thread it was discussed that we could use LTO Tapes to create our weekly full backups. Is this the correct item?
Additionally how might I tie this into my plan for backup.
I still need to determine, to what device I'm going to backup our VM's and Data to either a R620 server in RAID 10 on SR with 3 or 4TB drives running a CIFS server for of management with the LTO Tape.
Any recommendations.
Unfortunately I'm outside of my experience with Tapes and how to tie them in.
You might want an autoloader to keep a months worth of tapes in it. which will be networked for control then have SCSI for data. Whatever you do if you are virtualizing make sure it can be passed through, though really the backup server should be a separate physical box.
-
@Jason That's the plan, any recommendations on an Auto Loader. I'd really like to use LTO 6 in this write up. @scottalanmiller said LTO 6 isn't released yet though ....
-
@DustinB3403 said:
@Jason That's the plan, any recommendations on an Auto Loader. I'd really like to use LTO 6 in this write up. @scottalanmiller said LTO 6 isn't released yet though ....
Something like this should work http://www8.hp.com/us/en/products/tape-automation/product-detail.html?oid=5336447#!tab=features
-
Why are you looking at tapes again?
Can StorageCraft not provide you what you want? StorageCraft your backups locally, then replicate it to your offsite server - wasn't an offsite server your original plan?
-
@Dashrender Tapes only for the weekly take home portion. As uploading over the WAN is unrealistic.
Storage Craft is to be used to backup to a new local server with 24TB of onsite storage.
-
@Dashrender Off site was in the plan, but I doubt I'll be able to get them to use a COLO or even our sister office overseas.
It's just more reasonable when written down I think.
