How long does it take to empty a 64gb red ssd card?

[Brian Iannone]

Hi Brian,

Where did you get that from, please ?
As far as I know, LaCie is a company with a very good reputation and high prices.

LaCie is a good company with a very good reputation among consumers. However, it's well-known in in the workstation/server enterprise IT industry that LaCie products have a failure rate above any other similar hard drive enclosure and SATA controller manufacturer.

 

[Kemalettin Sert]

The card does indeed make a big difference. If you're referring to LaCie's SATA II ExpressCard/34 card, I've never tried it, but I've heard it works well. I've never heard of anyone receiving the 300 MB/s speeds that LaCie states the product supports on the product page, but you should get between 125-200 MB/s out of it. Another good choice is Sonnet's Tempo SATA ExpressCard/34. I personally would go with Sonnet's card as LaCie products tend to have a high failure rate.

Sonnet Expresscard34 has been problematic with me on macbook pro.it keeps disconnecting while doing transfer

 

[Brian Iannone]

Just to make shure people dont get a unrealistic point of view..
MOST harddrives dont have a transferrate around 120-150 MB/s. Good 7200 ones have around 80 MB/s in reality.. 120 MB/s i havn't seen yet on a single drive (especially if it is filled above 10%)
To get 260 MB/s with raid 0, you need at least 4 drives to have sustained rate throughout the whole storage space.

There is a big difference getting a few MB's on an empty disk and copying SSD after SSD on set..

I would account for not less than 15min for an SSD to a single drive

Since most people's desktop/workstation setups tend to be configured in unideal ways, I suppose 80 MB/s would be plausible, however that sounds unrealistically slow. As you say, it all depends on the disk. I would typically expect to hear that sort of speed coming from a 5400 rpm drive, though. I get 130 MB/s from each of the Seagate 7200 rpm 3 TB disks that I use on my editing workstation. I have them configured in an 8-disk RAID 0 array. One interesting (slightly off-topic) note that I would recommend to anyone building an array is that hard drives, even ones of the same model, vary in speed. For example, a Hitachi HD103UJ. The model number is "HD103UJ." Therefore, you would expect every HD103UJ model hard drive by Hitachi to be identical in speed. That's not true. Depending on which of the manufacturer's facilities that disk was made and the firmware running on the drive, the speed can vary ± 20 MB/s. Which is why when you build an array, it's best to test the speed of each individual drive before creating the array to ensure that all the disks are similar is speed. By similar, I mean ± 5 MB/s. Another note: try to always use 7200 rpm drives with as large as cache and capacity as possible. Due to platter densities on each of the disks, a drive that's 75% full with data will operate slightly slower than a drive that's 25% full. Access times also decrease.

Because there's so many factors that affect disk speed, I typically like to give a wide range like 120-150 MB/s. Of course, if your setup is flawless in terms of configuration, you'll get faster speeds. If you use lesser quality components such as 5900 rpm 100 GB disks with 4 MB cache, then speeds are going to decrease dramatically.

Also, if you're using single drives to store data and not RAID arrays, use drives with high platter densities. Platter density is the amount of data each platter in the hard drive can store. For example, take a 1 TB drive. You could have 1 TB drives with 5, 4 or 3 platters.

A 1 TB drive with 5 platters would mean that each platter has a density of 200 GB (1,000 GB / 5)
A 1 TB drive with 4 platters would mean that each platter has a density of 250 GB (1,000 GB / 4)
A 1 TB drive with 3 platters would mean that each platter has a density of 333 GB (1,000 GB / 3)

Because the sectors on the platters are packed more closely together, a 1 TB drive with 3 platters will be significantly faster than one with 5 platters.

 

[Julien Deka]

Here are the maximum theoretical speeds of the interfaces on the REDMAG reader when emptying a 64 GB SSD.

eSATA: 2 minutes, 51 seconds
FireWire 800: 10 minutes, 40 seconds
USB 2.0: 17 minutes, 47 seconds


Factoring in drive overhead, here are some more realistic speeds.

eSATA: 5 minutes
FireWire 800: 12 minutes
USB 2.0: 20 minutes

Hi Brian, thanks for this very useful information... So, a 128GB SSD should take about 25min max to get emptied in a RAID0, right?

It is reassuring a bit, because I read 2 days ago on Phillip Bloom's blog that it takes 90 minutes to empty a 128GB SSD from is Epic-M, using FireWire 800...

It seemed a lot to me. But the fact is: Phillip bloom is an Epic user and owner, when I never had a chance to play with this camera so far.

Those cards take an age to offload! Firewire 800 will take around 90 minutes to offload to one drive…

http://philipbloom.net/2011/11/20/scarlet/

His math sounded off to me, and from my experience with Redmags (and other regular HDD), your post seems more accurate.

Any Epic user to confirm by practice rather than theory?

Thanks!

 

[Brian Iannone]

Hi Brian, thanks for this very useful information... So, a 128GB SSD should take about 25min max to get emptied in a RAID0, right?

It is reassuring a bit, because I read 2 days ago on Phillip Bloom's blog that it takes 90 minutes to empty a 128GB SSD from is Epic-M, using FireWire 800...

It seemed a lot to me. But the fact is: Phillip bloom is an Epic user and owner, when I never had a chance to play with this camera so far.



http://philipbloom.net/2011/11/20/scarlet/

His math sounded off to me, and from my experience with Redmags (and other regular HDD), your post seems more accurate.

Any Epic user to confirm by practice rather than theory?

Thanks!

Ninety minutes to transfer 128 GB almost sounds ridiculous. That would be a speed of 24 MB/s (190 Mbps). I have no idea what he meant by that, but 24 MB/s sounds frighteningly slow. The throughput capability of FireWire 800 is four times that speed and the REDMAG SSD itself is capable of more than ten times that rate. Even if you are transferring to a single hard drive, the HDD should be able to handle at least 100 MB/s.

It would be great to hear from some RED users here about what type of speeds their getting.

 

[Mike P.]

I think bloom may have been transferring to an external, and incredibly slow, drive (5400rpm? USB?) I think he uses MBP for the most part, and they only have one FW800 connector (although you could daisy chain them... would the throughput get cut in half?) No way does an external plattered 5400 or 7200rpm get sustained 100MB/s... Maybe burst... I seem to recall it being closer to ~25-33MB/s sustained (which would coincide with the 90minutes for 128gigs..)

 

[John Fairstein]

I use R3D Datamanager http://www.r3ddata.com/ to transfer files from my RED One drives. I don't have any numbers for transfer rates, but it verifies transfers with checksums so it may be slower. Any thoughts on the value of checksum verifcation for R3D file transfers?

 

[Brian Iannone]

I think bloom may have been transferring to an external, and incredibly slow, drive (5400rpm? USB?) I think he uses MBP for the most part, and they only have one FW800 connector (although you could daisy chain them... would the throughput get cut in half?) No way does an external plattered 5400 or 7200rpm get sustained 100MB/s... Maybe burst... I seem to recall it being closer to ~25-33MB/s sustained (which would coincide with the 90minutes for 128gigs..)

If he had the REDMAG station connected to his MBP over FireWire 800, and was transferring the data to another hard drive connected to the computer over USB 2.0, then it's possible the speeds could slow down to 25 MB/s. I would have to assume that the external drive he was using was a 2.5" disk as they are much slower than 3.5" disks. You are correct, that would make sense. Also, when you daisy chain FireWire drives, the throughput will only decrease by half if you are transferring data to or from both connected drives simultaneously.

It's highly unlikely that you'd get 100 MB/s from an external drive (single disk) even if it's connected over FireWire 800. HDD/HDD enclosure manufacturers typically use 5900 rpm or 7200 rpm low-performance disks in order to keep the costs down and more affordable to consumers. Some companies like G-Technology or LaCie's enterprise enclosure division use server-grade 7200 rpm drives, however.

 

[Brian Iannone]

One additional note: avoid USB 2.0 whenever possible. The maximum speed is 480 Mbps, however, due to it's architecture when transferring large amounts of data, it often operates at half that.

 

[Jeff Kilgroe]

Seems there are several armchair IT / DIT guys in this thread. I would not recommend either the Sonnet or LaCie solutions for eSATA connectivity at this time. The Sonnet cards currently have an incompatibility with OSX Lion (10.7.x) in conjunction with the RED SSD media. The LaCie cards are just OK. Some of their products are great, some not so much. I put their SATA controllers in the not so much category.

The CalDigit cards are great for both ExpressCard-34 and PCIe. I would start there. Also consider an ATTO H6xx series card if you're installing in a Mac Pro or PC.

In a proper setup with good eSATA host and fast drives to copy to, the RED SSD's can saturate the 3Gbps SATA channel. They can peak at up to the full 280MB/s or so, holding about 250~260MB/s for reads in my experience. Should be about 4 minutes to unload a completely full 64GB REDMAG in ideal conditions. I don't have any 64GB SSD's myself, but the guys at RED say they're all the same speed. My 256GB cards often get filled to nearly full and they unload in 15 minutes or less.

FW800 takes nearly 4X as long. And USB2.0 is only for those who want the transfer to happen slowly overnight...

Transfers will only be as fast as your slowest connection / device being used. So on a Macbook Pro, your best bet is the CalDigit eSATA ExpressCard. The latest model has dual 6Gbps ports and supports EC2.0 for the latest Macbook Pro 17". It obviously still doesn't have enough bandwidth in that ExpressCard slot to fully supply both of those ports, but still enough to unload a REDMAG at full speed and dump it to equally fast storage. You can also connect the RED Station via eSATA and your storage via Thunderbolt. I can get the rates I mentioned above with this CalDigit ExpressCard adapter and writing to a Thunderbolt Pegasus RAID. I get about 60% of that speed if I swap the Pegasus RAID for the LaCie Thunderbolt little big disk. Which, BTW, is one of the LaCie products that I consider very good. They hit it out of the park with this little Thunderbolt drive. Expensive, heavy, but compact and fast. Not sure about rugged. The housing sure is, but I have my doubts about the innards.

Always connect the RED Station via eSATA where you can. A good dual-port eSATA ExpressCard is a good way to go if you can. Because even in older 250MB/s ExpressCard setups, connecting both the RED SSD and your storage there is going to be a bit faster than connecting one of them to eSATA and the other to FW800. Remember, FW800 real-world speed tops out around 76MB/s... It just can't go any faster. 800Mbps with 10bit/byte overhead gives a peak bandwidth of just over 76MB/s... You'll need a very good 2-disk RAID or better to consistently sustain that. Most single-drive units won't swallow 75MB/s sustained for lengthy times, as has been pointed out here already.

Some people do the RED Station via FW800 and their storage via eSATA. It's totally workable and is reasonably fast. For 128GB and larger SSD's you can probably start an offload and have it finish in time to swap media at the EPIC. Run 3 SSD's and you should be fine on all but the most demanding productions. I personally recommend 5 or more SSD's, but I'm a bit obsessive/ compulsive when it comes to fault tolerance and planning for failure. You'll want more SSD's if you have smaller ones because you'll cycle them faster.

 

[Deanan]

The Atto cards are also great.

 

[Jeff Kilgroe]

Sorry everyone, it wasn't my intention to act like a show-off. In the IT world, an "armchair IT" person means someone who "shows off and constantly tries to show their tech expertise." That is absolutely not what I was trying to do. I was just trying to help.

You're fine. The armchair comment wasn't referencing you.

I mentioned the ATTOs above and Deanan just did as well. To be more specific, the H680 and H6F0 are awesome. And being full SAS/SATA cards, you can attach LTO tape units and other devices to them.

A handy one from CalDigit is the 6GU3. It has dual 6Gbps eSATA ports and dual USB3.0 ports.

 

[Laco Zamba]

And what about some real numbers from REDMAG owners?

 

[John Marchant]

I use 256gb cards - a simultaneous offload to two HD103SJ or HD204UI of a typically filled card takes under 20 mins.

Thats on a macpro using the ATTO H680 - four esata breaking out from one port, and an LTO5 on the other via SAS. The Red Station is on esata, and the drives are esata too in very cheap and cheerful trayless Startech enclosures. Cheap fast solution.

I had someone in my studio just two days ago citing Bloom's mention of transfer speed as a major issue for them. Safe to say I could put his mind at rest with a little demo

 

[Brian Iannone]

I use 256gb cards - a simultaneous offload to two HD103SJ or HD204UI of a typically filled card takes under 20 mins.

Thats on a macpro using the ATTO H680 - four esata breaking out from one port, and an LTO5 on the other via SAS. The Red Station is on esata, and the drives are esata too in very cheap and cheerful trayless Startech enclosures. Cheap fast solution.

I had someone in my studio just two days ago citing Bloom's mention of transfer speed as a major issue for them. Safe to say I could put his mind at rest with a little demo

Twenty minutes to offload 256 GB in a real-world situation is great!

 

[Daniel P Rudin]

Seems there are several armchair IT / DIT guys in this thread

Great to know, you can differentiate that based on ability to read..

Omar asked about dumping 64GB SSD via FW800 or USB 2.0.. still you MasterDIT's throw 250+ MB/s with dumptimes of 4min around... that is simply not possible.. so much for armchair..

I still stand by my answer of 80MB/s and 15min absolutely best given Omars parameters.. and yes, this is tested by 64GB SSD via FW800 to a 6 Disk Raid..

But what do i know??

 

[Dave_S]

Omar, Don't get me wrong on this but Its unbelievable that you can download 64GB with in 2 or 3minutes using e Sata and actually its laughable that 17 minutes for USB 2 .... unless you are using a super computer. if you are using top range i7 gaming computer, 64GB card= 30 minutes using usb3, about an hour using usb 2. yes, in theory if you calculate the speed it seems very fast but, in practice, that speed/bandwidth is only maintained for first few 4 or 5GB of data after that it slows down. please improve my knowledge by telling me that i am fully wrong.

 

[Dave_S]

just one more thing... if you ever wish to test download speed..please test it with full 64 gb data.

 

[Daniel P Rudin]

Because the sectors on the platters are packed more closely together, a 1 TB drive with 3 platters will be significantly faster than one with 5 platters.

Totally agree Brian, because the 3TB / 3 Platter Seagates you are using, are quite the fastest harddrives you can get at the moment, you do get wonderful results with a good raid controller..

But the 'normal' user will not get near that performance with using FW800 to external disks, especially if powered via FW or USB (these single 5400 drives are targeted towards powersaving, not speed, they will get you to times as Phillipe Bloom has experienced)

And that is what Omar asked in his post...

 

[Stephen Matthews]

Should be about 4 minutes to unload a completely full 64GB REDMAG in ideal conditions. I don't have any 64GB SSD's myself, but the guys at RED say they're all the same speed. My 256GB cards often get filled to nearly full and they unload in 15 minutes or less.

FW800 takes nearly 4X as long. And USB2.0 is only for those who want the transfer to happen slowly overnight...

Great to know, you can differentiate that based on ability to read..

Omar asked about dumping 64GB SSD via FW800 or USB 2.0.. still you MasterDIT's throw 250+ MB/s with dumptimes of 4min around... that is simply not possible.. so much for armchair..

I still stand by my answer of 80MB/s and 15min absolutely best given Omars parameters.. and yes, this is tested by 64GB SSD via FW800 to a 6 Disk Raid..

But what do i know??

Seems he agrees with the figure you stated in fact his is more conservative, 4x4=16. And on the 4min being not possible front he seems to be speaking from experience, well in comparison to his 256GB times(it's still a guesstimate but better than nothing).

We asked for someone with real world experience and he gave us a frame of reference form his own personal experiences, I don't see anything wrong with that.

Play nice!

edit: and there is this as well "Most single-drive units won't swallow 75MB/s sustained for lengthy times, as has been pointed out here already"