3.5" or 2.5" HD more reliable?


Status
Not open for further replies.
Ignoring the price and size difference, does anybody know whether the normal 3.5" HD or the 2.5" notebook HD is more reliable.


So far I have been storing my photos on internal as well as external 3.5" HD and they usually give problems after some time. WD, Maxtor, Seagate -- the brand does not matter.. they will all ultimately fail.

I have to buy a new HD soon and am thinking of getting a 2.5" one but I do not know whether they are better than the bigger ones or not.

If they are always failing, other possibilities could be

a) for internal drives, it could be a location or system configuration issue.
b) for external drives, handling or usage issue.

E.g. For external 3.5" drives, are the power adaptors designed to run for long hours over a long period of time? For external 2.5" drives, after they are disconnected from the PC or laptop, was a couple of secs given for the drive to spin down?
 

just go for your needs?
mobile --> 2.5"
cheap and good --> 3.5"
got extra money --> NAS
 

Just want to point out that many 3.5" drives were never designed to be used in a mobile environment regardless of whether they are enclosed in an external casing or not. Serious knocks can easily compromise your data. On the other hand, 2.5" disks are designed as portable and laptop drives and often come with at least some sort of shock dampening that will come in useful.
 

If you are talking about reliability and want to get the most reliable HDD out there in the market now, get a solid state drive.

A good SSD will outlast HDD by many, many times, and allow much faster access speed too. But get ready to pay the price; SSDs are not cheap.

But if you are just comparing just the 2.5" and 3.5" HDD, there's no real difference between the two, in terms of reliability and lifespan, as far as I know.

Really interested in where you got your info on SSD from. As far as I know SSD are prone to MTBF way before harddisk, due to limited write cycles, and even more so when using MLC type of SSD.
 

Jeez... the level of misinformation floating around is just incredulous...

That thing which you call a "surge protector" is a ferrite core, which is a choke to reduce electromagnetic interference. It does NOT protect against electrical surge!


A good external HDD should have surge protection.

DSC09127.jpg
 

Really interested in where you got your info on SSD from. As far as I know SSD are prone to MTBF way before harddisk, due to limited write cycles, and even more so when using MLC type of SSD.

Most flash memory used in SSDs are rated for 1 million+ cycles and have MTBFs in the range of 1 million hours. Of course this MTBF greatly depends on your write frequency... But as a storage for photos, doubt you're going to write to the same cell over and over again.

Still, at the current cost of SSD, I don't think its a wise choice to use them as storage devices for photos/videos...
 

Last edited:
Haha, this is a joke innit..An SSD is never proven for reliability. Wonder why the enterprises are very very slow in adopting SSD's in enterprise environments even though SSD's provide insane IOPS?

A good SLC based SSD should technically be superior than a traditional HDD (in theory). But SSD is not only about the Flash, What's rather more important is the controller being used and the firmware (Ever heard of the infamous JMicron controllers and the stuttering issues?)

Ever heard of the slow performance after a half full SSD? Do you know how TRIM works and which SSD supports it now?

Coming back to TS' question..A 2.5" HDD was earlier limited by the speed (RPM), onboard cache and the platter density. With huge advancements in platter technology and with abundant availability of 7200RPM HDDs & Huge Caches the gap between the mainstream 3.5" and 2.5" drives have shrunken a lot. I have about 5 - 6 external HDD's which are all 2.5" and they are working fine for years now (None of them are branded)

Not really a joke, but I did forget to mention a few points.

I know of TRIM; not sure how it works and all its technical details, but I know it's about SSD controllers knowing which data blocks are truly in use, and which had been discarded. That helps with access speeds, as far as I know. Correct me if I'm wrong.

As for which SSD supports it now, I admit I was not entirely sure.

Thanks for raising up the few points though. Did some searches and learnt a few new things. Like how SSDs now still don't support TRIM. I'd read of TRIM a long way back and thought it would be supported by now. :dunno:

And I can't believe I forgot RAID. :bsmilie:
 

Jeez... the level of misinformation floating around is just incredulous...

That thing which you call a "surge protector" is a ferrite core, which is a choke to reduce electromagnetic interference. It does NOT protect against electrical surge!

Thank you for highlighting my misconception; learn something new today.
 

Not really a joke, but I did forget to mention a few points.

I know of TRIM; not sure how it works and all its technical details, but I know it's about SSD controllers knowing which data blocks are truly in use, and which had been discarded. That helps with access speeds, as far as I know. Correct me if I'm wrong.

As for which SSD supports it now, I admit I was not entirely sure.

Thanks for raising up the few points though. Did some searches and learnt a few new things. Like how SSDs now still don't support TRIM. I'd read of TRIM a long way back and thought it would be supported by now. :dunno:

And I can't believe I forgot RAID. :bsmilie:

No worries dude..Intel's second generation SSD's now support TRIM and some of the Indilinx based controllers with the revised firmware now support TRIM. Do watch out for data corruption / loss during firmware upgrade and make sure you have a solid backup if you're intending to upgrade the firmware.

SSD's today are made of millions of NAND gate flash cells (This can be further classified as single bit / cell or multi bit / cell). When it comes to writing data, it needs to be written in pages (Typically a 4 Kilo Byte chunk). Things are different when it comes to erasing / deleting the data as it can only be done in blocks (Typically 512 Kilo Bytes or 128 Pages).

SSD's will have to keep track of every last bit of data of an address to know whether it's written or free. This typically slows down the performance of the SSD when you have half filled or near filled data. This is why you find a huge difference in performance between a newly formatted SSD and a used SSD.

To combat this ATA-TRIM commands are required (This needs support from the OS too like Windows 7). When you permanently delete a file / data block, the addresses are sent along with the ATA-TRIM commands to the SSD controller and it wont track these addresses for their state anymore which typically reduces the load for the controller.

There will definitely be huge variances on the algorithms of each controller though they use the same ATA-TRIM command set.

Do note that there are totally different approach in SSD Algorithms (Write less) which benefits TRIM pretty much and improves the overall performance (Example the SandForce Controllers on OCZ's Vertex 2 Pro)
 

you are turning us into geeks too innit?
 

No worries dude..Intel's second generation SSD's now support TRIM and some of the Indilinx based controllers with the revised firmware now support TRIM. Do watch out for data corruption / loss during firmware upgrade and make sure you have a solid backup if you're intending to upgrade the firmware.

SSD's today are made of millions of NAND gate flash cells (This can be further classified as single bit / cell or multi bit / cell). When it comes to writing data, it needs to be written in pages (Typically a 4 Kilo Byte chunk). Things are different when it comes to erasing / deleting the data as it can only be done in blocks (Typically 512 Kilo Bytes or 128 Pages).

SSD's will have to keep track of every last bit of data of an address to know whether it's written or free. This typically slows down the performance of the SSD when you have half filled or near filled data. This is why you find a huge difference in performance between a newly formatted SSD and a used SSD.

To combat this ATA-TRIM commands are required (This needs support from the OS too like Windows 7). When you permanently delete a file / data block, the addresses are sent along with the ATA-TRIM commands to the SSD controller and it wont track these addresses for their state anymore which typically reduces the load for the controller.

There will definitely be huge variances on the algorithms of each controller though they use the same ATA-TRIM command set.

Do note that there are totally different approach in SSD Algorithms (Write less) which benefits TRIM pretty much and improves the overall performance (Example the SandForce Controllers on OCZ's Vertex 2 Pro)

i repeat -
i would trust someone by the name of geekbrains on a photography forum. :bsmilie:
 

No worries dude..Intel's second generation SSD's now support TRIM and some of the Indilinx based controllers with the revised firmware now support TRIM. Do watch out for data corruption / loss during firmware upgrade and make sure you have a solid backup if you're intending to upgrade the firmware.

SSD's today are made of millions of NAND gate flash cells (This can be further classified as single bit / cell or multi bit / cell). When it comes to writing data, it needs to be written in pages (Typically a 4 Kilo Byte chunk). Things are different when it comes to erasing / deleting the data as it can only be done in blocks (Typically 512 Kilo Bytes or 128 Pages).

SSD's will have to keep track of every last bit of data of an address to know whether it's written or free. This typically slows down the performance of the SSD when you have half filled or near filled data. This is why you find a huge difference in performance between a newly formatted SSD and a used SSD.

To combat this ATA-TRIM commands are required (This needs support from the OS too like Windows 7). When you permanently delete a file / data block, the addresses are sent along with the ATA-TRIM commands to the SSD controller and it wont track these addresses for their state anymore which typically reduces the load for the controller.

There will definitely be huge variances on the algorithms of each controller though they use the same ATA-TRIM command set.

Do note that there are totally different approach in SSD Algorithms (Write less) which benefits TRIM pretty much and improves the overall performance (Example the SandForce Controllers on OCZ's Vertex 2 Pro)

A very enjoyable read on ur post geekbrains!! I wished i consulted ya before my ssd puchases previously :bsmilie:. Actually the MTBF of some of the SSDs now including the MLC ones like OCZ turbo/vertex approx 1.5 mil hrs

The OCZ Vertices are a godsend when running my Win 7 pro. My boot up/shut down of win 7 with OCZ Vertex on my old Core 2 simply pwn my harddisk booting win 7 pro with i7 core..

But the price of the SSD per byte is pretty ex. for large capacity storage, the hdd is still the more economical option. But as portable storage, the ssd is more immune to the usual knocks and bumps

ryan
 

A very enjoyable read on ur post geekbrains!! I wished i consulted ya before my ssd puchases previously :bsmilie:. Actually the MTBF of some of the SSDs now including the MLC ones like OCZ turbo/vertex approx 1.5 mil hrs

The OCZ Vertices are a godsend when running my Win 7 pro. My boot up/shut down of win 7 with OCZ Vertex on my old Core 2 simply pwn my harddisk booting win 7 pro with i7 core..

But the price of the SSD per byte is pretty ex. for large capacity storage, the hdd is still the more economical option. But as portable storage, the ssd is more immune to the usual knocks and bumps

ryan

Thanks giantcanopy! I'm glad that I managed to explain this in simple terms. SSD's without TRIM support will be faster in the initial stages and you will start to notice some degraded performance once data fills in.

The MTBF is one thing that I'm least bothered about (Though it means a lot in enterprises for reliability). In reality MTBF means nothing. The drive can fail anytime. Especially NAND wear out is very fast if you're updating / deleting files in a large scale. That's where the wear leveling algorithms play a major part. (60GB drives will have actually more than 60GB, typically 64GB)

NAND technology is improving at a tremendous pace and MLC NAND with intelligent Controller algorithms can match the speed of SLC NAND these days (SandForce controllers). Let's wait and see what we can get in 2010 on SSD :d

I'm yet to see any fruitful outcome of WD's acquisition of the SSD Controller maker SiliconSystems Inc.
 

really enjoyed reading some of the articles in your geekbrains.com website

thanks for sharing your knowledge........ appreciated
 

Back to TS's question.......
If the objective is to get a reliable storage, then consider this setup:
1) A NAS (Synology, QNAP and many other brands) or external storage that has
a) minimum two-bay (meaning it can take two disks) config, for 3.5" disks
b) built-in RAID-1 (and RAID-5 for 3 or more bays) support.
2) Decide on disk size. 1 tera-byte is a good start. Decide on a brand.
3) Go to two different shops to get disks of the same model. Why?
- Avoid getting disks of the same manufacturing batch.
- For most of the time when there is a manufacturing problem, it affects all products made in the same batch.
4) If you are willing to spend more, get a UPS (Uninterruptible Power Supply), and have your computer and storage plugged into it.

Reliable storage is never going to be cheap. But prices have fallen greatly to be quite affordable. For priceless photos, I do not mind spending on it.
 

well, get a mirror instead of RAID 5. WD Mirror is one of the choices. when the NAS in RAID 5 failed, it may not be possible to access the data at all.
 

really enjoyed reading some of the articles in your geekbrains.com website

thanks for sharing your knowledge........ appreciated

Thanks Ed! I'm glad that my articles are of some help:D

Back to the TS,

Have you decided on your HDD purchase yet?
 

Status
Not open for further replies.
Back
Top