Budget RAID 1 Recommendation

[Kelmon]

Hi Guys,

I'm in the process of changing a lot of my current setup because I need more storage space for my photographs, which means that the 1TB drive in the old iMac is going to be replaced. Oddly, it's being replaced by a Retina MacBook Pro with a 512GB SSD so the majority of my photo library will need to be stored externally, with the MBP holding just my active projects. Now, at the moment I have my photo collection on the internal 1TB drive, which is then cloned to an external USB 2.0 drive, and then backed up to Amazon Glacier as well. It's simple enough for me to transfer my photo library to an external drive in the new setup but I want it to be backed up so I think I need a RAID 1 solution.

The available capacity of the solution should be around 2 or 3TB. I don't (yet) make money from my photography so I don't currently need a professional level solution but I do want RAID 1 for some degree of piece of mind before the data gets to Amazon. Connection via USB 3.0 is likely all I need but the MBP will have Thunderbolt if there are some affordable solutions. Having just blown a load of cash on the MBP and a separate display I'd like to keep the costs down at this moment.

Any suggestions?

 

[ColdCase]

RAID is not a backup.

You only need RAID if you are dissatisfied with data transfer rates, or you want more than 4TB in a single volume. RAID used to make more sense as drives were expensive and efficiently using them saved some cash.

There are several USB3 4TB drives for less than $125. Buy two, use one for your photos, use the other as a backup for both the rMBP and photo drive. Continue to backup off site. By the time you fill that 4TB drive, there will be inexpensive larger ones available.

You only need to invest in TB drives/RAIDs if you are combining more than two rotational drives or are using SSDs.... or if you want to reduce cable clutter. SATA-TB adapters are available for some seagate USB3 drive enclosures if you want to go there.

 

[Kelmon]

My theory, so please correct me if I'm wrong, had been that with 2 drives in a RAID 1 configuration that my photos, when archived off the MBP, will automatically be backed up in case one of the drives in the array fails. I could use something like SuperDuper to backup a single external drive to another but that sounds a bit inefficient.

 

[paulrbeers]

RAID is not a backup.

You only need RAID if you are dissatisfied with data transfer rates, or you want more than 4TB in a single volume. RAID used to make more sense as drives were expensive and efficiently using them saved some cash.

There are several USB3 4TB drives for less than $125. Buy two, use one for your photos, use the other as a backup for both the rMBP and photo drive. Continue to backup off site. By the time you fill that 4TB drive, there will be inexpensive larger ones available.

You only need to invest in TB drives/RAIDs if you are combining more than two rotational drives or are using SSDs.... or if you want to reduce cable clutter. SATA-TB adapters are available for some seagate drives.

You are so far off I have no idea to begin with....

First I will grant you that RAID 1 is not a backup but rather a way to retain data in case of a hard drive failure. A true backup would be off site somewhere either "in the cloud" or at a friend/family member's house.


The rest I think you got confused with RAID 0.

RAID 0 (which really isn't RAID at all) is targeted at speed of transfer and for creating a large disk from multiple smaller drives. It also increases your failure rate!


RAID 1 is what any sane person would use. You buy two drives of the same size and give up half of the over all storage space so that if you ever lose a drive you don't lose any of your locally hosted data. There is some speed advantages in read only, but the true point is to have an exact copy of what you have regardless of a drive failure. All of my insure storage is in RAID 1. Anything mission critical is backed up to cloud storage and anything that would be annoying to lose (incase of fire or other disaster) but could be obtained (DVD and Bluray rips) is at least covered by a drive failure due to the RAID 1.

 

[Kelmon]

Thanks, Paul. For a moment there I thought I was barking up the wrong tree.

 

[adamneer]

why not just use Time Machine if you are only after a quick and easy "backup" solution? that way, you aren't eating possible (though probably not much) resources having your computer save duplicates every time something is written to the drive? i guess the benefit would be automatic drive restoration in the event that one goes bad, but other than that, i've never seen enough benefit of RAID 1 to bother. RAID 3 and 5 seem like much more beneficial solutions, although certainly not really necessary for your uses.

in any case, if you do go with a RAID setup, here is what I use, and as far as i know, its the cheapest setup out there (but its actually quite nicely built).

http://www.amazon.com/Vantec-3-5-Inch-External-Enclosure-NST-400MX-S3R/dp/B0089V4WOC/

Ive got 2 of them, both with 2tb Seagate Barracudas for a total of 4 drives/8tb of storage. One of them is setup in RAID 0 for speed as my media and/or cache drive for my video work. The other is just setup as JBOD, which I use as my Time Machine drive for backups of the RAID 0 and my system drive. The main benefit of the JBOD is that I only eat up 1 USB input on my machine instead of 2. Beyond that, it is potentially not the most reliable setup, given that if 1 drive fails, the entire TM backup will be taken out. But at the same time, I find it highly unlikely this would happen at the exact same time as my RAID going out, so unless they both get zapped by a huge power surge or something, i feel pretty safe.

 

[Kelmon]

Hi Adam,

The reason why I'm not using Time Machine here is simple: the data to be backed up will not be stored on the internal drive of the MacBook Pro (too small). The plan is to use that drive to store the OS, application and in-progress projects/documents. Anything that is finished is to be stored on the external drive. While I plan to have this data backed up to Amazon Glacier, the upload to Glacier takes ages so I'm really worried about the potential for a drive failure of the external drive before it has been backed up to Amazon, hence the desire to use RAID 1. RAID 5 sounds nice but my budget won't run to that. Another reason for the RAID solution would be to cut down on desktop clutter. In addition to the RAID array I want to maintain a Time Machine backup of the internal MacBook Pro drive and a clone of that disk.

The Vantec sounds nice but does not seem to be available in the UK. However, I do see other similar models that can be obtained, such as one by StarTech (SAT3520U3SR). I confess that something else I've started looking at is the LaCie 2Big Quanta.

 

[MCAsan]

I do LaCie 2Big 6TB running as RAID 1 3TB. That houses all my user libraries of photos, music, documents...etc. My 3TB Time Capsule backs up both the rMBP's internal SSD plus the LaCie 3TB. Of course Time Capsule is also generating my wifi network.

This is a good solution. If I wanted to improve it, I would also have backups offsite in a bank vault, cloud server....etc.

Each person needs to decide how critical some of their data is to them....and how much time and money they wish to invest in their backup solution.

 

[flynz4]

Hi Adam,

The reason why I'm not using Time Machine here is simple: the data to be backed up will not be stored on the internal drive of the MacBook Pro (too small). The plan is to use that drive to store the OS, application and in-progress projects/documents. Anything that is finished is to be stored on the external drive. While I plan to have this data backed up to Amazon Glacier, the upload to Glacier takes ages so I'm really worried about the potential for a drive failure of the external drive before it has been backed up to Amazon, hence the desire to use RAID 1. RAID 5 sounds nice but my budget won't run to that. Another reason for the RAID solution would be to cut down on desktop clutter. In addition to the RAID array I want to maintain a Time Machine backup of the internal MacBook Pro drive and a clone of that disk.

Personally... I think you are making a mistake. Your rationale is as follows:

So you will initially move completed projects onto the external... and since that is the only copy of your data... you want it to be RAID 1... particularly for the period of time until it is backed up to Amazon.​

The problem is that once you move your projects to your external drive, you still only have one copy of your data even though it is RAID 1. Yes... you have redundant copies on two physical HDDs... and RAID 1 does protect you from certain failure mechanisms (in particular, either HDD failing)... however (and this is the big deal)... RAID does introduce new failure mechanisms that do not exist on simpler systems. You could have a controller failure, data corruption errors, etc... that are unique to the more complex RAID subsystem.

My personal opinion (and I agree with ColdCase)... never consider RAID (of any flavor) to be backup. Hence... any data that resides exclusively on the array should be considered "at risk" just as if it resided on a single HDD.

Also... your rationale about Time Machine is also flawed. You claim that you cannot use Time Machine is because your projects will NOT be on the internal drive of your MPB. However, Time Machine is not limited to just backing up your internal drives. It can easily back up external direct attach drives as well. You just go into Time Machine preferences... and make sure that the external drive is not excluded (it probably is excluded by default).

/Jim

 

[Kelmon]

Hi Jim,

OK, that tends to change things quite substantially. I still think RAID 1 would be a backup solution unless there is a reason why the failure of a disk or the enclosure would mean that the data on both disks would be lost (if so then what's the point of it?). However, if Time Machine can indeed also backup external drives (documentation that I'd seen so far didn't talk about that but there's been enough responses to indicate that it does) then that might also be a solution, and a cheaper one at that. I should be able to get a reasonably priced USB 3 drive to act as main drive and continue using my current 2TB USB 2 drive as a Time Machine drive.

Thanks for the input, and to MCAsan.

 

[flynz4]

Hi Jim,

OK, that tends to change things quite substantially. I still think RAID 1 would be a backup solution unless there is a reason why the failure of a disk or the enclosure would mean that the data on both disks would be lost (if so then what's the point of it?). However, if Time Machine can indeed also backup external drives (documentation that I'd seen so far didn't talk about that but there's been enough responses to indicate that it does) then that might also be a solution, and a cheaper one at that. I should be able to get a reasonably priced USB 3 drive to act as main drive and continue using my current 2TB USB 2 drive as a Time Machine drive.

Thanks for the input, and to MCAsan.

My recommendation is to NOT think of a RAID solution as "two independent disks"... but rather consider it to be "a single point of failure". It is not only HDDs that fail. For an example... a failed controller can irreversibly corrupt data on multiple HDDs at the same time.

While I do have several RAID solutions at my disposal... I personally think that RAID is vastly overrated for consumer applications. It is useful if you absolutely need a single volume larger than the current size of HDDs (as of today... that is 4TB). It is also useful (or better said... it used to be useful) if you need more performance than a single HDD can deliver. The problem with that is that SSDs are so much faster than RAID HDDs... that SSDs are probably a wiser choice for IOPs sensitive workloads.

They key to having data safety... is to have multiple copies of the data, on as many different independent media as possible. Ideally... with the different media in different geographic locations... and using different software mechanisms to make the backup of the data... and also... with it requiring zero human intervention. RAID should not be considered a collection of independent media. It is just a single piece of media that in some cases is more reliable than a single drive... and in other cases having new failure mechanisms that are unique to RAID.

Finally... Time Machine will back-up any direct attach drives (such as those using Thunderbolt, USB, or FW. It will not back up network attached drives. Personally, I think putting any picture libraries on a NAS is a mistake. I believe a DAS is a vastly superior solution.

/Jim

 

[paulrbeers]

and RAID 1 does protect you from certain failure mechanisms (in particular, either HDD failing)... however (and this is the big deal)... RAID does introduce new failure mechanisms that do not exist on simpler systems. You could have a controller failure, data corruption errors, etc... that are unique to the more complex RAID subsystem.

Jim -

These might be true if you are using RAID 5 and a hardware RAID. We are talking about RAID 1 which is the simplest of RAID's and OSX has a SOFTWARE RAID 1 built into the OS. I've moved my OSX SOFTWARE Raid from one enclosure to another on entirely different machines with no issue. In fact, I just recently moved 4 - 3TB hard drives in RAID 1 from an eSATA Enclosure hooked to a Mac Pro to a USB 3.0 Enclosure hooked to my Mac Mini with NO issues. Further, RAID 1 is direct data copy so it is not a complex RAID subsystem. Theoretically data corruption can be had whether you use independent disks or not because if the main disk you copy to ends up with a data corruption, then guess what gets copied to the duplicate?

I used to use a RAID 5 system with hardware RAID and your points are exactly why I moved to a simple RAID 1 with SOFTWARE RAID. It can be easily moved. There is no "controller failure" that can't be overcome by just moving them to another enclosure. Data corruption is only slightly more likely to happen with RAID 1 (although I have yet to experience it).

Again, I think people keep missing what is actually being asked in this thread....

This isn't RAID 0 and this isn't a complex hardware based RAID 5 or 6. Are we not talking about Simple Software RAID 1 that is controller independent?!?!?!

 

[flynz4]

Jim -

These might be true if you are using RAID 5 and a hardware RAID. We are talking about RAID 1 which is the simplest of RAID's and OSX has a SOFTWARE RAID 1 built into the OS. I've moved my OSX SOFTWARE Raid from one enclosure to another on entirely different machines with no issue. In fact, I just recently moved 4 - 3TB hard drives in RAID 1 from an eSATA Enclosure hooked to a Mac Pro to a USB 3.0 Enclosure hooked to my Mac Mini with NO issues. Further, RAID 1 is direct data copy so it is not a complex RAID subsystem. Theoretically data corruption can be had whether you use independent disks or not because if the main disk you copy to ends up with a data corruption, then guess what gets copied to the duplicate?

I used to use a RAID 5 system with hardware RAID and your points are exactly why I moved to a simple RAID 1 with SOFTWARE RAID. It can be easily moved. There is no "controller failure" that can't be overcome by just moving them to another enclosure. Data corruption is only slightly more likely to happen with RAID 1 (although I have yet to experience it).

Again, I think people keep missing what is actually being asked in this thread....

This isn't RAID 0 and this isn't a complex hardware based RAID 5 or 6. Are we not talking about Simple Software RAID 1 that is controller independent?!?!?!

Paul,

I still consider it to be a single point of failure... because that is exactly what it is. BTW: I also consider "clones" a single point of failure... for exactly the same reasons. This is particularly true for clones that run automatically on a schedule.

BTW: I do use clones... and I do use RAID. However... I think it is important to understand their weak spots. That does not mean they are useless... but I firmly believe it is a mistake to put too much credence in their use.

I personally believe that all data (that one desires to keep) be double backed up.

1) Offsite with deep versioning - I use Crashplan+
2) Local with deep versioning - I use Time Machine/Time Capsule

I also use clones and such... because they are useful... especially if you notice corruption immediately, or if you have a device crash. Still... their use is temporal in value... once a clone is overwritten by corrupted data, it is pretty much useless.

Also... I have a number of RAID systems in use. They also have value, although they are diminishing in value. They were especially valuable in the days where disk drives were small... and RAID was the best method to create large volumes. For the most part, with 4TB drives... I just don't need more in a single volume. Some do... and RAID is probably the best technology to get such large volume.

I also put some data directly onto NAS RAID arrays with no backup... but that is only for data that I just do not care about anymore. An example would be my ripped personally owned movies. Honestly... I am not even sure why I keep them anymore. With streaming movie services such as Netflix... the value of having a home movie collection is dubious for me.

So bottom line... I don't have anything against RAID. I use the technology. However, I think its consumer value is diminishing over time. Most importantly... I think it is a dangerous trap for people to think that RAID is offering significant data security... and choose to use it instead of backup. RAID is not backup... period.

/Jim

 

[Kelmon]

Just to confirm, in the end I went with:

• LaCie 3TB d2 Quadra v3 as primary storage
• Seagate 2TB Backup Plus as Time Machine drive
• Seagate 1TB Backup Plus Portable as clone drive for rMBP and photo import copy from Lightroom
• Amazon Glacier for off-site backup

The LaCie ought to be relatively dependable with a 3-year warranty and the rest is largely there in case of problems. As a solution it likely could be stronger (certainly the drivers take up space on the desk) but I think it will do the job.

* crosses fingers *