Backup Strategy - 16x4Tb - Areca 1882-24-4gb - Supermicro SC846 - Hackintosh

P

PanaMax

n00b
Joined
Feb 2, 2015
Messages
4
Good day everybody,

I am a enthusiast Photo and videographer. i Shoot all my Fotos and since two years also all my videos in RAW. currently i have about 20Tb of data. since i expected my data to Grow fast in the near future i startet last year a Hackintosh project. This project is now in his final phase and so far working well...

here are the details of my setup:

Supermicro Superchassis 846 with 24 3,5"Hdd Trays
16x4Tb Seagate running in Raid6
Areca 1882-24 4gb Ram version
Currently 7x3Tb seagate + 1x5Tb External HDD for backup
OS: Mac OS X 10.9, disks formatted in HFS+
Ram 32Gb
Intel Devils Canyon
Gforce 770 4Gb


Current Backup Strategy:

currently i use Carbon Copy Cloner (a backup software) to backup all my data to the individual external Hdd`s, this i do roughly every 4-6 Weeks. afterwards i store the external HDD´s in the same house but on another place / floor then my computer.

What i don't like with this strategy:

I have to organize my to be backuped files by my self. that means i have for example a video folder organized by years. now i have jan12-jul13 on one 3tb Disk, then jul13-jan15 on the next and so on... with a growing Data this becomes more and more difficult to not loos the overview. also i either don't use the full HDD capacity for backup for organization, or i use the full capacities and its a mess to keep the overview... i hope you get the point.

On the other hand i would prefer to have automatic updates so i don't need to remember myself to get the external drives plug them in, search for the power adapters etc...

and i still have 8 available slots in my server Case and 8 available slots on my Areca card.


Questions i have:

- what do you think about running a Raid5 for backup and a Raid 6 for main data on the same card? what if the card fails?
- What you think about running JBOD instead of Raid 5 for the backup?
- if i create a pass through disc, can i connect the HDD to any other Sata port on any other computer (also not to the raid card) and read the data? or i need to always connect the card to my or any Areca Cards to be able to read it?
- How would you backup the data if you´d be me?


Things that can't be changed / done:
- i won't use another OS
- I can't use online backup due to a 1mbit upload max. speed.


looking forward to you're support on this!
Thank you,

Max
 
I'm unclear as to whether you are saying you want the backup drives to be in the same SuperMicro Chasis as the live drives. I would recommend against this.

Depending on your budget, a 16 bay JBOD connected to a PC running FreeBSD, Solaris, or FreeNas with ZFS as your file system. You can use RAIDZ2 (same thing as RAID 6). A few minutes googling will show you why ZFS is a great File System for backups. This leaves you 8 bays for expansion later on.

In regards to your RAID 5 suggestion for your backups, google "Raid 5 is dead" to see why this isn't feasible (unless you're using SSD's).

Then for backups you can run nightly rsync's over your 1Gb lan.
 
Thank you for your reply... ZFS and RAIDZ2 sounds very promising... however i won't look for an additional fileserver and then keep everything running and consuming electricity. i tested yesterday that a pass through disk can be read by a normal SATA port, so i guess it doesn't matter in case the raid card dies.

for now i bought a bunch of extra HDD Trays, and i will keep updating those individually, and then store them somewhere else. and with growing disks like the upcoming 8Tb disk for archives i think this would be the best and cost effective backup solution.
 
Don't think it really matters what you run on your backups so much in regards to RAID levels. They are backups after all and you'll have two copies of your data. Most of the arguments against RAID 5 are when that's used to house data that doesn't have backups. You could run RAID 0 if you really wanted to.

As for the Areca cards, pass-through disks are not tied to the Areca HBA and you can plug them in anywhere. If the card fails, as metadata for the RAID array is stored on the disks instead of the controller, it's just a matter of you obtaining another Areca card and plugging all the disks in.
 
Let's set a few things straight here, raid is not a backup and a backup stored in the same machine is not a backup. To have a proper backup strategy you need your data, a copy of you data in a separate machine and then a cold backup.

If you are serious about protecting your data you need raid6 on the main machine. Raid in this case is for convience. If you loose a disk you don't have to restore from backups. It's just easier.

The secone machine needs to be running some kind of file system that ensures data integrity, like zfs. You could use zfs compression on that machine because it's a backup target, not a high performance server. I would run as large of drives as you can with a minimum of raidz2 with no more than 6 drives (4 data 2 parity) in a vdev.

cold storage is going to be your problem, no real easy way to do that with as much data as you have. Bi-weekly clones to 8tb drives in removable enclosures could solve that. Have 6-8 8tb drives total and rotate disks out, so 3-4 on one cold backup and use the other disks the next time and rotate after that.

Do it right or don't do it. If you are serious about protecting your data you need to get realistic.
 
Last edited:
bds1904 said:
Let's set a few things straight here, raid is not a backup and a backup stored in the same machine is not a backup. To have a proper backup strategy you need your data, a copy of you data in a separate machine and then a cold backup.

Do it right or don't do it. If you are serious about protecting your data you need to get realistic.
Click to expand...

As someone with 20+ years in the backup/disaster recovery industry, truer words have never been spoken.
 
First of all thank you for your comments again.
And of course i want to make it right, thats why we are here :)

Current situation:

I run my main array (16x4) in Raid 6 + i have a copy of all that data spread on 6 external 3tb HDD´s stored at another place and only in use once i make the backup or restore my data...

i would think thats already not such a bad Backup strategy, please correct me if im wrong.

Improvements to the current situation:

If i understood right, you recommend 3 copies of the data...

1. the Master
2. a Permanent copy on a second machine
3. a cold copy.

the 1st and the 3rd is already done, so that leaves me with the question how to handle number 2. in a reasonable way. Would a Jbod Case be a possibility? how do you connect the backup machine to the master? via gigabit ethernet... that would take forever to copy the data?!? how you would solve this if you´re on a budget?

...
Max
 
Interesting problem you have. I am trying to find a way to back up about 1/3 the amount of data you have. Cloud backup is too slow, no budget for faster internet.
One more thing you should consider if it's important is an off-site backup. I know given what you are describing this might be tough, but what happens if there is a fire/natural disaster? (I don't have answers either - just throwing out something else to think about).
Most of my data is not important enough for me to lose sleep over. My important data is backed up to a separate machine and cloud backup. But, it's only about 2 TB and does not change often.
The rest (about 15T) can be replaced - just annoying and painful to do so.
 
Gigabit would only take a long time on the first backup. After that, rsync would only be copying the changes.

Assuming 100MB/s it would take around a week.
 
DO NOT USE RAID 5!!

Backups or nots, drives that large you are looking at days for rebuilds = crap performance + chance of datalose = more work to get back up and running.
 
MrGuvernment said:
DO NOT USE RAID 5!!

Backups or nots, drives that large you are looking at days for rebuilds = crap performance + chance of datalose = more work to get back up and running.
Click to expand...
RAID 5 is perfectly fine for this. It will not take days to rebuild and the chances of both arrays going out concurrently is very slim. Performance isn't poor either. You don't sound like you're that well versed on modern RAID cards seeing as pretty much every point you've made has been wrong.
leeleatherwood said:
Backblaze, unlimited data backups.

$5 a month.
Click to expand...
Best of luck uploading 50TB+ over a home internet connection. :rolleyes:
 
Yeah, Backblaze is too slow for a home network connection. I've been uploading 2TB for over a month now. I think I'm down to 10-12 days before its down.
If I back up my 16TB of data (not even close to the OP's system), it's going to take about a year to back everything up. A lot can happen in a year.
 
What about amazon s3 cold storage. If OP doesnt need regular access - perhaps this may be an option for a true back up repository?
 
You must log in or register to reply here.
Back
Top