Backup Strategy...

[sidewinder]

Folks,

My 8 core Mac Pro should be here in a few days and I have not spent much time considering a backup strategy. I want to keep it simple and reliable.

I am not sure what size drives I am going to get. I will not be doing a lot of graphics work on the system. I will be using it for everyday business productivity, iPod music storage, and application development. I will also be using Fusion 2 with a Vista Home Premium VM. ON my current PC, I am using less than 100GB. I suspect 750GB will give me plenty of headroom.

My first thought is to setup a three drive system using software RAID 1 (mirroring) with Time Machine using the third drive. I know hardware RAID 1 is faster, but I don't think software RAID 1 will stress my Mac Pro too much. I would like to do this with all drives internal. Will Time Machine support that? If I don't think I am going to come close to filling up the 750GB array, do I really need to get a bigger drive for Time Machine?

I am not looking for the highest availability possible or the most data security possible. I just want to make sure that I have a reasonably solid strategy that will keep me up and running through a single drive failure and allow me access to older files that I may have deleted for whatever reason.

Thoughts?

S-

 

[fredr500]

SearchT

his has been discussed numerous times before.

but let me say again, any drive that is connected to power should not be considered backup. I had a direct lightning hit that took out 5 computers, all on UPS/surge protectors.

My current solution is a 1 TB external drive for TM connected all the time. Every weekend I take my second 1TB external drive from the fireproof safe, disconnect one drive, unplug it from the wall, then plug in the second and let TM do it's thing on it. Then I reverse it all and put the drive back in the safe.

It doesn't take that much of my time, and it is something I can live with. I suppose I really should be taking the drive to work and locking it in my desk instead.

And to prove that I am truly paranoid I am also a beta tester for Carbonite mac.

Fred

 

[sidewinder]

Fred,

I didn't find anything that matched what I was talking about doing when I did a search. Of course, the search capabilities of these forums is weak at best.

Also, did you read the part where I said "I am not looking for the highest availability possible or the most data security possible."?

My main questions revolves around Time Machine which I have never used before. I run a rather large data center so I know about RAID and off site storage of backups. As I said, that's not what I am trying to do....

S-

 

[sidewinder]

Well, I found out that I will have no problems using Time Machine to back up to an internal drive. The only question left is what size drives to buy. I'll probably buy two 500GB or two 750GB drives to run the RAID 1 array and a 1TB drive for Time Machine.

S-

 

[Semiopaque]

I haven't used Time Machine so I can't speak to that, but I would use a two drive strategy. Partition each drive the same way, one partition on each drive is your system and application drive and one partition is your data partition. You can software mirror the data partitions if you want.

The OS/app partition I would image using cloning software and only clone to the "backup" drive when you're sure your system is flawless. (I actually use a third drive to create a third 'working' clone as well so I don't always have to jump back too far if needed).

I think a third drive for Time Machine may help with your 'accidental' deletions of data more than RAID 1, though, again, haven't used it myself.

 

[sidewinder]

Semiopaque,

RAID 1 will do nothing to help "accidental" deletions. That is what Time Machine will help with. RAID 1 keeps me up and running if a drive fails. That's why I want to use both.

I have read some posts here that suggest that you can't boot from RAID arrays. When using 10.5.6, Is that true for software RAID, hardware RAID, both, or neither?

Is there really any issues with using a single drive (or RAID 1 array) for the OS, apps, and data? How much of a performance is there really if fast drives are used?

S-

 

[Semiopaque]

Semiopaque,

RAID 1 will do nothing to help "accidental" deletions. That is what Time Machine will help with. RAID 1 keeps me up and running if a drive fails. That's why I want to use both.

I have read some posts here that suggest that you can't boot from RAID arrays. When using 10.5.6, Is that true for software RAID, hardware RAID, both, or neither?

Is there really any issues with using a single drive (or RAID 1 array) for the OS, apps, and data? How much of a performance is there really if fast drives are used?

S-

Sorry if I wasn't clear in my third paragraph, but yes mirroring will not help with accidental deletions which is why I mentioned Time Machine there. I'm assuming it does some sort of time stamped back-up but I have not used it so I'm not sure.

The main point of my suggestion was to avoid putting your OS and apps into a RAID with your data at all. Separating your OS and Apps from your data will allow you to make clean, bootable clones of your OS and Apps without affecting your data, and is the quickest way to get back up and running if something goes wrong with your primary drive and/or OS as you can just boot from your backup drive immediately. It didn't have anything to do with performance as your main issue seemed to be surviving a drive failure quickly.

I have not had any issues booting from a software RAID. I have not set-up a Mac hardware RAID myself, but the issues I have heard about seem to be hardware specific so I think it may be possible to set that up depending on the hardware you choose.

 

[Full of Win]

Another option to consider is off-site backup, in addition to Time Machine.

S3 from Amazon is one of many examples: http://aws.amazon.com/s3/

The cost is not that much, about the same as an external drive would cost, when you project its cost out for 3 to 5 years.

 

[nanofrog]

I have read some posts here that suggest that you can't boot from RAID arrays. When using 10.5.6, Is that true for software RAID, hardware RAID, both, or neither?

In software RAID, you can boot from the array. In hardware RAID, it depends on whether or not the card supports in. They do exist for Macs, but have a tendancy to not be cheap. The other features however, usually make up for this. Apple's own RAID card is the exception. Junk. Slow, expensive, and has problems.

I would truly hope you take Semiopaque's information on separating the OS from the data to heart. Apps can be placed on either. The reasoning is simple. If the OS fails, and it's on the array, it has a nasty habit of taking the data with it. It also takes more work to fix. Separated, it can't.

Conversely, if the array goes, you still have a working OS. This makes it much easier to recover, or in a worst case scenario, completely rebuild the array. (All data lost, and start from scratch). Hence the need for a separate backup system. Failures do happen, and the cost associated with data recovery is beyond expensive.

Is there really any issues with using a single drive (or RAID 1 array) for the OS, apps, and data? How much of a performance is there really if fast drives are used?

S-

Performance depends on multiple factors.

• Software vs. Hardware

• Type of RAID

• Drives: type/model and quantity

A good setup that can fit internally in a Mac Pro, consists of:
Separate OS drive, physically mounted in the empty optical bay.
The 4 drive bays are then used for RAID. RAID 5 offers a good balance of speed and redundancy. Not quite as fast as RAID 0 (offers no redundancy at all), but faster than RAID 1, or 0+1. It requires a minimum of 3 drives to build a RAID 5 array, but using the 4th will increase your throughput speeds.

Use a separate backup system, and has several practical reasons.
1. The data is still there if the Mac dies*.
2. Doesn't eat up drive bays, which are limited in the Mac Pro.
3. Will allow you to keep the array completely internal, assuming you won't exceed what can fit.
4. Assuming 3, it keeps the array from being spanned between internal and external drives. Simply put, keep the internal drives on a separate array, even if you also have an external RAID solution.
This is a can of worms that will cause endless hell if you don't.

*PSU failures in particular.

 

[kudukudu]

his has been discussed numerous times before.

but let me say again, any drive that is connected to power should not be considered backup. I had a direct lightning hit that took out 5 computers, all on UPS/surge protectors.

My current solution is a 1 TB external drive for TM connected all the time. Every weekend I take my second 1TB external drive from the fireproof safe, disconnect one drive, unplug it from the wall, then plug in the second and let TM do it's thing on it. Then I reverse it all and put the drive back in the safe.

It doesn't take that much of my time, and it is something I can live with. I suppose I really should be taking the drive to work and locking it in my desk instead.

And to prove that I am truly paranoid I am also a beta tester for Carbonite mac.

Fred

Fred, I have a similar backup strategy, but I keep my second backup off-site and I swap them every few months. A word of advice on the safe strategy: fireproof and waterproof are not the same thing. I know of someone who lost a ton of documents in a fireproof safe from a fire because they all got soaked (water form the fire hoses went through the seams). In their new house everything in the safe is in ziplock bags.

 

[nanofrog]

In their new house everything in the safe is in ziplock bags.

I would avoid these.

LDPE (Low Density Polyethylene) has a low melting temp (105 - 115C typically). It's only rated to 80C sustained, and 95C for a short duration, before it flows (oozes all over anything inside).

 

[sidewinder]

Guys,

I hear and understand what you are saying. I help manage a data center that requires the kind of reliability and redundancy that you are trying to achieve here. I am not looking for super never fail data redundancy in case of a huge natural or unnatural disaster! If I were, I would have separate OS/app and data drives and all would have some form of mirroring. I would backup to offline storage that would be on site and off site. Any important data that I really want to keep will be burned onto DVD's and stored safely.

I am not to terribly worried about lightning strikes here in Northern California. Nor am I worried about floods or fires. The only thing I really need to worry about are earthquakes and power outages. I have a UPS......

Anyway, assume I RAID 1 my single drive that contains the OS, apps, and data while using Time Machine to a third drive. The two drive RAID 1 array has doubled the chance of a disk failure over a single disk, but the chances of both drives going out at the same time are MUCH lower than a single drive failing.

Time Machine gives me some things the RAID 1 array does not. I get access to files that I have deleted, either by accident or on purpose, that I need again. And it provides me with a backup that I can use to restore my system after reinstalling the OS. I won't RAID 1 the Time Machine drive because the files stored there are not "must have".

The odds of both RAID 1 drives AND the Time Machine drive going bad at the same time are ridiculously low. No, this setup is not perfect. But, after some research since I started this thread, I am comfortable with what it will provide in regards to keeping my data safe and me up and running.



What I really want to understand is how having separate OS/apps and data drives affects the performance of the Mac Pro. Coming back to Macs from the Windows world is not a simple transition. But, I used to hear the same "you need separate OS/apps and data drives" talk that I hear here. Yet, in actual practice, single drive systems ran just as well and darn near as fast as multi-drive systems. They were much simpler to maintain, support, and to backup.

Yes, systems that access a lot local data, especially randomly, benefit from multiple drives. Servers absolutely benefit from multiple drives. But how much of performance hit am I going to take using a single RAID 1 array with OS, apps, and data versus splitting them up?

Again, I will not be doing a lot of graphics work on the system. I will be using it for everyday business productivity, iPod music storage, and application development. I will also be using Fusion 2 with a Vista Home Premium VM. ON my current PC, I am using less than 100GB.

S-

 

[nanofrog]

What I really want to understand is how having separate OS/apps and data drives affects the performance of the Mac Pro. Coming back to Macs from the Windows world is not a simple transition. But, I used to hear the same "you need separate OS/apps and data drives" talk that I hear here. Yet, in actual practice, single drive systems ran just as well and darn near as fast as multi-drive systems. They were much simpler to maintain, support, and to backup.

Separating them onto different drives will increase the throughput. Simply because you aren't making the same drive to do everything at once. Data flows through different interface ports simultaneously, to their respective drives, reducing the load per drive, thus increasing throughput.

Better than a fast, single drive any day.

But how much of performance hit am I going to take using a single RAID 1 array with OS, apps, and data versus splitting them up?

The actual throughput reduction shouldn't be that great, as the processing for RAID 1 isn't too significant. Nothing like RAID 5 or 6.

Again, I will not be doing a lot of graphics work on the system. I will be using it for everyday business productivity, iPod music storage, and application development. I will also be using Fusion 2 with a Vista Home Premium VM. ON my current PC, I am using less than 100GB.

S-

How much application development do you do?

I ask, as this is something a different RAID type can really help with for compile time. If you use this professionally, or at a similar level, it would make sense to go such a direction. (For me, it's circuit simulation. I couldn't get things done reasonably, unless I have all kinds of time to wait a day or two for the results to be completed for each sim).

If for occasional use, why did you go for the Mac Pro, given the cost?

 

[shayn]

Guys,

I hear and understand what you are saying. I help manage a data center
S-

Why are you asking questions about backup strategy if you manage a data centre? You should already know the answers

 

[sidewinder]

Why are you asking questions about backup strategy if you manage a data centre? You should already know the answers

Because managing scores of servers and various forms of network attached storage that require five 9's uptime, high availability, automatic fail over, and data redundancy is significantly different than a managing a desktop system that does not require any of those things.

Plus, I pretty much new what I wanted to do. But, since I lack experience in the recent Mac world, I thought it would good to ask. I thought I laid out the situation upfront. But I guess I was not clear.

S-

 

[sidewinder]

Separating them onto different drives will increase the throughput. Simply because you aren't making the same drive to do everything at once. Data flows through different interface ports simultaneously, to their respective drives, reducing the load per drive, thus increasing throughput.

Better than a fast, single drive any day.

Yes, I understand that.

The actual throughput reduction shouldn't be that great, as the processing for RAID 1 isn't too significant. Nothing like RAID 5 or 6.

This is what I was trying to get at. Yes, the system will perform faster with multiple drives. But is it faster enough to be worth the extra costs and maintenance?

How much application development do you do?

I ask, as this is something a different RAID type can really help with for compile time. If you use this professionally, or at a similar level, it would make sense to go such a direction. (For me, it's circuit simulation. I couldn't get things done reasonably, unless I have all kinds of time to wait a day or two for the results to be completed for each sim).

Currently? Very little. In the future? Who knows. If I did end up doing a lot of development, I would consider hardware RAID 0+1, 5, or 10.

If for occasional use, why did you go for the Mac Pro, given the cost?

The Mac Pro will be my every day all day system. I will have to keep a Vista Home Premium (32-bit) VM running all the time for one application.

My other option was an iMac. Not too appealing when you you consider I don't want to have a bunch of external enclosures on my desk on on the floor. I didn't want to have only 2 cores. Finally, I wanted more flexibility in my choice of displays.

Buying a refurbished Mac Pro was not that expensive.

S-

 

[nanofrog]

This is what I was trying to get at. Yes, the system will perform faster with multiple drives. But is it faster enough to be worth the extra costs and maintenance?

Depends on use. (see next section)

Currently? Very little. In the future? Who knows. If I did end up doing a lot of development, I would consider hardware RAID 0+1, 5, or 10.

This is what I really needed to understand.

It seems like you won't be placing a heavy demand on the drives, so skip the extra drives for now, and just go with SW RAID 1. You can always upgrade if you need to.

The Mac Pro will be my every day all day system. I will have to keep a Vista Home Premium (32-bit) VM running all the time for one application.

My other option was an iMac. Not too appealing when you you consider I don't want to have a bunch of external enclosures on my desk on on the floor. I didn't want to have only 2 cores. Finally, I wanted more flexibility in my choice of displays.

Buying a refurbished Mac Pro was not that expensive.

S-

I can understand that. A single CPU (quad core) of the current would have sufficed, but it's a good idea to get the second processor at the time of purchase. Last I looked, the prices are still way too high to try and do it yourself later.

Refurbs are wonderful, aren't they?

 

[shayn]

Because managing scores of servers and various forms of network attached storage that require five 9's uptime, high availability, automatic fail over, and data redundancy is significantly different than a managing a desktop system that does not require any of those things.

Plus, I pretty much new what I wanted to do. But, since I lack experience in the recent Mac world, I thought it would good to ask. I thought I laid out the situation upfront. But I guess I was not clear.

S-

In that case, check out SuperDuper It makes a clone of your drives and is bootable. Use it instead of a raid setup, you can set it to run every day at a specific time.

A combination of Superduper ( for bootable backup ) and TM ( for recovering deleted contacts, emails or photos ) is a great combination. Set the macpro up with a system - apps drive and a data drive. Clone both of these to system backup and data backup. This uses up your 4 drives in the macpro. Get a 1tb external drive and use it for TM for your system, apps drive. Your iTunes and photos will also be on the system drive so any photos, songs, emails, contacts, etc you accidentaly delete will be recovered by TM. You have a nightly backup of both your drives with SuperDuper. If your system drive fails, you can put a new drive in, clone it from the system backup and be back in action straight away.

Get smartreporter and install it. It will warn you if your drives are going to fail.

Get Macaroni for automated system maintenance.

 

[sidewinder]

shayn,

Thanks for the links to "smartreporter" and "Macaroni". Both look quite interesting. Any other "must have" software I should look into getting?

In regards to "SuperDuper", I understand the concept but why would I do that instead of using software Raid 1 for both drives? The RAID solution is transparent and more up to date. If a drive fails, the other drive has all the current data. No need to go boot a drive that is many hours out of date which means there is lost data, even with Time Machine running. Yes, the software RAID 1 will slow writes down a bit, but reads should be faster.

I am going to start with a 4 drive system. Two 750GB Seagate NS drives in a software RAID 1 as my boot/OS and data drive. I have a third 750GB Seagate NS drive that will be the cold swap drive for the array but I will use it for Time Machine to start. This drive will eventually be replaced with a 1.5TB drive but I will wait until I fill the RAID array up a bit. The fourth drive is a 500GB Seagate AS drive that I will use as a temp/scratch/play drive that will not be backed up.

That should give me plenty of storage and decent, but not awesome, data security. It will be better than what most people have, that's for sure.

S-

 

[sidewinder]

Refurbs are wonderful, aren't they?

They are even better when your wife works at Cisco and gets a discount and you have friends that work at Apple and get an even better discount!

The price difference compared to a new iMac 24" made the Mac Pro decision a "no brainer"!

s-

 

[shayn]

shayn,

Thanks for the links to "smartreporter" and "Macaroni". Both look quite interesting. Any other "must have" software I should look into getting?

In regards to "SuperDuper", I understand the concept but why would I do that instead of using software Raid 1
S-

As I said, if you know the answers, why are you asking the questions?

 

[sidewinder]

As I said, if you know the answers, why are you asking the questions?

I think I asked a question, didn't I? I wanted to know what advantage using SuperDuper had over using software RAID 1. I laid out my thoughts on why I felt RAID 1 was the better option. My thoughts could be all wet. I might not be considering something that you know and I don't.

Why do people get so uptight when they are asked to explain their suggestions when compared to another point of view?

S-

 

[millerj123]

I think I asked a question, didn't I? I wanted to know what advantage using SuperDuper had over using software RAID 1. I laid out my thoughts on why I felt RAID 1 was the better option. My thoughts could be all wet. I might not be considering something that you know and I don't.

Why do people get so uptight when they are asked to explain their suggestions when compared to another point of view?

S-

I've been bouncing ideas with a guy at work, and I'm going to put RAID 1 drives in my wife's computer, and use an external drive or two for periodic backups. I don't have any experience with either SuperDuper or the other software, so I'm also interested in other people's choices.

 

[jamin100]

Have you every thought about storing most of the data in the cloud?
I know dropbox do 100GB plans and then this would / could sync to any pc you wanted.

As for RAID1, you would probably be better off getting a NAS with at least 2 drives.

Or Western Digital do a mirroring one that connects five firewire or USB. Then should you want to move this to another location its easy to do.

 

[philosopherdog]

Mozy for off site backups. You get 2 gb free! $5 a month for unlimited space. It's also one of the most reliable off site outfits.