Internal Storage HD Size - What's enough?

[dpavid]

I just checked my 2008 MacPro Applications, System, and Library Folder sizes and it use 35 GB, 6 GB, and 63 GB for each folder. That comes out to roughly 100+ GB. If I store all my files and media on an external Thunderbolt RAID, and strictly use the internal just for the OS and applications, a 256 GB Internal would be sufficient.

Any drawbacks of why I should consider the 512 GB model?

 

[Marty62]

I just checked my 2008 MacPro Applications, System, and Library Folder sizes and it use 35 GB, 6 GB, and 63 GB for each folder. That comes out to roughly 100+ GB. If I store all my files and media on an external Thunderbolt RAID, and strictly use the internal just for the OS and applications, a 256 GB Internal would be sufficient.

Any drawbacks of why I should consider the 512 GB model?

No, for you that seems to make sense.
I would go bigger if you want to import lots of "working files" like film/photo
or audio for editing.
You have not mentioned your usage but based on a 100gb system then the
256 should be fine.
Just remember not to over fill it, perhaps 200gb maximum.
M.

 

[dpavid]

I plan to keep all my working files on an external RAID 5 striped with 6 drives minimum. Speed should be sufficient.

 

[flatfoot]

Sounds good.

 

[AidenShaw]

I plan to keep all my working files on an external RAID 5 striped with 6 drives minimum. Speed should be sufficient.

You'd probably be better off with 3 or 4 larger drives, 6 is a lot for RAID-5.

 

[dpavid]

You'd probably be better off with 3 or 4 larger drives, 6 is a lot for RAID-5.

True, but we'll be editing multi-cam 4k in less than a year. The more drives, the faster disk speed.

 

[Jim-H]

You'd probably be better off with 3 or 4 larger drives, 6 is a lot for RAID-5.

Curious why you say this. Aren't the new Pegasus2 R6 and R8 configured with RAID5? If you have 6+ drives, what should the config be?

 

[AidenShaw]

Curious why you say this. Aren't the new Pegasus2 R6 and R8 configured with RAID5? If you have 6+ drives, what should the config be?

Writing 1 KiB to a 6 drive RAID-5 array requires the system to read 320 KiB and write 65 KiB.

Writing 1 KiB to a 6 drive RAID-50 array requires the system to read 128 KiB and write 65 KiB.

Writing 1 KiB to a 6 drive RAID-10 array requires the system to write 2 KiB.

(For really big sequential writes with a RAID-5 controller with nonvolatile writeback cache - no reads would be needed and the writes would only stretch by the one extra write to the parity drive.)

In practice, this means that RAID-5 arrays lose write performance as you increase the number of drives in the array.

There is also a serious problem in that the expected error rate from disks is approaching the size of the disk.

A Seagate 4TB NAS drive has an expected unrecoverable error rate of 1 bit per 10^14.

IT CONTAINS 3.2 * 10^13 BITS!

As you add drives, the probability of a bit error while reading the drives to rebuild the array gets closer and closer to 1. Statistically, a 12 drive RAID-5 array of these Seagates could not be rebuilt - any attempt would get an error which would cause the loss of all data.

The is compounded by the fact that drive failures are not independent. One of the best predictors of whether a particular drive will fail is if drives with nearby serial numbers have failed. (Failure due to bad runs at the factory, material problems, etc.)

Therefore, when you buy a batch of brand new disks, you're setting yourself for co-related failures.

 

[VirtualRain]

I just checked my 2008 MacPro Applications, System, and Library Folder sizes and it use 35 GB, 6 GB, and 63 GB for each folder. That comes out to roughly 100+ GB. If I store all my files and media on an external Thunderbolt RAID, and strictly use the internal just for the OS and applications, a 256 GB Internal would be sufficient.

Any drawbacks of why I should consider the 512 GB model?

What do you use your nMP for? Is it practical to keep your work-in-progress on your internal SSD? If so, this would probably make a huge difference in the performance of your workflow at various stages. For example, in my case, I keep my active photo library on my internal SSD which makes a lot of functions much faster than if I stored it on an external (loading library, generating versions, exporting, etc.)

At the very least, my advice would be to buck up for at least one size more than you think you need because even if it's enough today, it probably won't be tomorrow.

Writing 1 KiB to a 6 drive RAID-5 array requires the system to read 320 KiB and write 65 KiB.

Writing 1 KiB to a 6 drive RAID-50 array requires the system to read 128 KiB and write 65 KiB.

Writing 1 KiB to a 6 drive RAID-10 array requires the system to write 2 KiB.

(For really big sequential writes with a RAID-5 controller with nonvolatile writeback cache - no reads would be needed and the writes would only stretch by the one extra write to the parity drive.)

In practice, this means that RAID-5 arrays lose write performance as you increase the number of drives in the array.

There is also a serious problem in that the expected error rate from disks is approaching the size of the disk.

A Seagate 4TB NAS drive has an unexpected unrecoverable error rate of 1 bit per 10^14.

IT CONTAINS 3.2^13 BITS!

As you add drives, the probability of a bit error while reading the drives to rebuild the array gets closer and closer to 1. Statistically, a 12 drive RAID-5 array of these Seagates could not be rebuilt - any attempt would get an error which would cause the loss of all data.

The is compounded by the fact that drive failures are not independent. One of the best predictors of whether a particular drive will fail is if drives with nearby serial numbers have failed. (Failure due to bad runs at the factory, material problems, etc.)

Therefore, when you buy a batch of brand new disks, you're setting yourself for co-related failures.

Book worthy.

I believe that parity RAID really has no business in a home user environment. It's designed for business continuity and if that's not critical to your storage, then parity RAID is taking on more complexity and risk to solve a problem you don't have.