What to buy for a RAID 0

[johannesj]

They are connected using 3gig, never had any problem using it.
I use esata for external single disk´s and for raid enclosure.

[GrassValley_BH]

It's statistics vs probability.

Say you roll a six-sided die (1d6) 10 times and you never get a 6.
Probabilistically, your odds of rolling a 6 have not changed. It's still a 1-in-6 chance.

However, statistically, your odds of rolling a 6 increase with every roll that isn't a 6.

This is all assuming the die is properly weighted and such.

[hdvideo]

I never use usb.
I use esata for external drive´s.


my best
Johannes

Is the Esata the same as those 1 gigabyte ethernet cables i am hearing about?

[johannesj]

I never use usb.
In the past i used firewire but now I use esata for external drive´s.


my best
Johannes

[hdvideo]

This is for HDV footage.

As far as that usb 2.0 casing i don't think the transfer rate of USB is fast enough.

[Jerry]

Ok Great thanks for the explanation.

What do i need to buy for a RAID 1 or 5 that is fast enough to allow me to edit HD footage?

You also need to be a little more specific on the HD footage.
Is this HDV footage( I guess it is) or are you going to be using full raster uncompressed 10bit as well.
If you are building a raid for HDV it will be a lot less expensive than one that will do full raster 10bit uncompressed.
On newer notherboards, such as the X38 chipset, you can get pretty fast speeds with the onboard controller. It will do raid, 0, 1, 5. So you could purchase four 1tb drives and be good to go.

[Zorro]

You can easily set up your own Raid 0 or 1 setup with a enclosure like that:

http://www.mapower.com.tw/mapower/Product.asp?CateID=277&LineID=3

[THoff]

thats like saying that a computer with two or more sticks of ram is more likely to fail sooner than one with one large stick

If all the RAM modules have the same MTBF rating, then yes, chances are that the computer with two modules will fail first.

Imagine a deck of 52 cards that has been shuffled and presented to us fanned out and upside down. I take one card, and you take two. Whoever has the lowest card pays the other person $1. Do you think you'll gain or lose money in the long run? If this still doesn't sink in, let's say you take ten cards and I take one, and again the person with the lowest card pays the other person $1. Do those game rules seem fair to you?

[Blast1]

It's the Mean Time Before Failure, not an absolute number.

Nobody said it was, its a guideline, a yardstick to give a estimated hardware or data lifetime.

but the fact is that one drive will fail before the other does, and when that happens, none of the data will be recoverable.

Really, what determines that fact? thats like saying that a computer with two or more sticks of ram is more likely to fail sooner than one with one large stick, Harddrives like most electro-mech devices have a reasonably predictable lifespan, infancy failures in the begining, a fairly consistant lifespan, with increasing failures toward the end of life, the event points can be reduced by a burn-in period in the begining, and removing the drives from service a reasonable time before predicted end of life is reached, the lifetime of course can be shortened by intervention by the operator, shock, cooling, and unstable power being the greatest perps.

[hdvideo]

Thank you this was helpful.

[LightPrism]

The decision on any Raid Array has to do with

1. the average read/write bandwidth that most of your projects (do you do a lot of 8bit 1080i/60i multi-stream - multi-video track compositing such as for commercials ... or do you maybe usually only need 1 or 2 high def video tracks and long-form narratives)

2. how long are your average projects? do you have several projects going at one time - how much storage do you need? and remember that most raids, as they fill up with data, the bandwidth starts to slow down. So for example, after a raid is about 1/2 to 3/4 full, they start slowing down in bandwidth read/write. Fibre raids can sustain bandwidth longer as they fill up, but they are starting to be surpassed for overall speed capability.

3. all of the above factored in to a safety scheme of raid configuration. As stated before, Raid 0 is the fastest, but no protection -but if you are just working a short form 30 second spot with tons of tracks, you could always backup to cheapo firewires and still mantain the super fast speed needed for more RT of multiple layers. Raid 3 and 5 are a compromise between safety and speed. But then they also usurp storage space and therefore speed as well. But if you are working on a large long form narrative that will stay on the raid for many months, then safety is more important for recovery. And remember that the more drives in a raid, while faster -- this also increases the chance of a drive failure.

It really needs your thoughtful look at your average projects to decide what you need.

There are some great solutions out there from Dulce, CalDigit, Ciprico, and others for some pretty high powered raid solutions. Keep in mind that most of them want a PCIe slot for the controller card.

Go to these folk's websites and you can learn a lot about what your raid need would be, included calculators for typical project types/video format types and the resultant bandwidth needed.

[hdvideo]

I would like a ready to go RAID 1

[THoff]

It's the Mean Time Before Failure, not an absolute number.

If both drives died at the exact same time, then a RAID 0 array would not increase the risk of data loss, but the fact is that one drive will fail before the other does, and when that happens, none of the data will be recoverable.

If you could have a theoretical RAID 0 array with 10,000 drives, would you trust your data to it for even one day, knowing that a single drive that failed would mean the loss of all data? Using your logic, it would be perfectly safe because a failure won't happen for several decades.

[Blast1]

In fact, it increases the chance of data loss because the crash of a single drive will take out the entire array.

The idea that having two drives in a raid 0 does not increase the chances of drive failure over having a single drive of the same size as the raid, people often confuse the Hardware MTBF with the Data MTBF by which you can increase the chance of data lose by raid 0, but the Data MTBF is usually measured in decades, with most drives like 30-40yrs, if the raid fails you lose your data or if the single large drive fails the same happens, the advantage of the raid is the increase of the through-put over a single drive the same size, There used to be a large cost savings but today you can get TB drives for about the same price as two 500gigs.

[THoff]

Are you looking for a ready-to-go RAID array or just an enclosure that you are going to drop your own drives into? What interface do you want?