NAS Disk Platters

[Bidule0hm]

It seems to be totally unnecessary to waste an SATA channel just for booting where theoretically each NAS would experience only 1 boot in its lifetime.

This is why we often use USB sticks instead of SATA drives for the boot drive(s) ;)

Also, from the RAID Reliability Calculator, does "meantime to data loss" mean a complete (thus catastrophic) failure of the entire RAID with unrecoverable data?

Yes, "data loss" means "data loss" :D

By chance, I found the following report. Notice Hitachi fares much better than others. Amazingly the Hitachi 4TB have 5 platters.

https://www.backblaze.com/blog/best-hard-drive/

Backblaze stats are not valid for our case because of the drives environment they use (high vibrations, very low temps).

According to the RAID calculator on failure rate, the MTTDL is associated with data failure not drive failure. As I understand it, data failure can be corrected by ZFS.

http://www.servethehome.com/raid-calculator/raid-reliability-calculator-simple-mttdl-model/

Yes, their results are false for RAID-Zx because ZFS can correct the data corruption.

The idea is that when one fails, one must replace it immediately. The probability of another failure in the meantime becomes the issue. Given the analyses, I have to agree with all of you that RAID-Z2 is the de facto best way to go.

In fact if you want the best reliability RAID-Z3 is better ;)

 

[zambanini]

a four way mirror would be better ; )

 

[Bidule0hm]

Yes but if you go that way we can say a thousand way mirror is better... :D

 

[zambanini]

that read speed, wow that would be great!

 

[joeschmuck]

To Ahjohng:

When it comes to reviews on any seller website such as NewEgg, you will never see 100% of sales commented on. So if NewEgg sold 10,000 units, and there is any failure rate (of course there is), then you would expect most people who had a failure to make a comment of an issue and not being satisfied. You really only see positive reviews from people who are more on the Geek side of things and want to give some feedback to the community. Most of the satisfied customers will not make a single post that they are satisfied. You must put every review into perspective. Also, if you have been stung by NewEgg, I'm sorry to hear that. They are generally top notch with customer service. I've have an issue myself where it took them 3 times to send me an UPS, the first two broke due to mishandling by the delivery company, but I still felt like NewEgg should have packaged it better in the first place.

As for 2, 3, or 4 platters, who cares because it will be under the warranty you purchased it under, meaning you should purchase your drives based on the warranty period. Grant it, if there is a particular drive model which has a high failure rate, no one wants to be replacing drives several times during the warranty. I have no issues with the WD Reds, nice warranty period too of 3 years, the Pros have a 5 year warranty if you want to spend the money, and the Greens are 2 year but I suspect they could last as long as 5 years too, just adjust the head loading cycle timer. I honestly expect to get 5 years of use out of my Reds before I have a failure, or really before I consider replacing them one at a time and then they could be my emergency drives. Hum, maybe a script to let me know when I've hit 43,800 hours for each drive, good milestone. I've got some time to make the script. I think you should also consider that some of those Red failures are due to the folks delivering the product to your door. You see it all the time of a package being tossed on the porch and not nicely placed on the porch, or of a driver just tossing packages to the front or rear of the truck. Some of these folks just don't care. Keep things into perspective it the main thing. If you have another complaint about WD Red drives or something along this road, I'd think you are just fishing for an argument, which I hope you are not.

 

[Ahjohng]

When it comes to reviews on any seller website such as NewEgg, you will never see 100% of sales commented on. So if NewEgg sold 10,000 units, and there is any failure rate (of course there is), then you would expect most people who had a failure to make a comment of an issue and not being satisfied. You really only see positive reviews from people who are more on the Geek side of things and want to give some feedback to the community. Most of the satisfied customers will not make a single post that they are satisfied. You must put every review into perspective. Also, if you have been stung by NewEgg, I'm sorry to hear that. They are generally top notch with customer service. I've have an issue myself where it took them 3 times to send me an UPS, the first two broke due to mishandling by the delivery company, but I still felt like NewEgg should have packaged it better in the first place.

I don't trust 100% of the reviews but do appreciate a large feedback base, and my bad experiences with Newegg is swamped by good ones. At the end of the day, it is the product that I have procured that matters. Believe me. I am enough problems in my life. With the future NAS build, I would wish it to be as smooth as possible. That is why I am gathering as much information as possible. With that said, I have also noticed bad reviews usually come in clusters according to dates inputted. Whether if whole bunch of disinformants choose to get together or a bad production lot is hard to tell, but it is something interesting to pay attention to to make my life less of headache to come.

As for 2, 3, or 4 platters, who cares because it will be under the warranty you purchased it under, meaning you should purchase your drives based on the warranty period.

I dislike to deal with RMA's.

I'd think you are just fishing for an argument, which I hope you are not.

I find this hardly to be an argument but an exchange of opinions and information. You don't know me. If this were an argument, I would present mathematical outcomes and experimental data that you would be totally tongue tied. My handle is 'Koobee Wublee' in the 'sci.physics.relativity' newsgroup. If you wish to see me in action, that is where you will find me.

In the meantime, I am not judging your decision of going with 3TB drives with an extra platter that I am not comfortable with from all these field reviews. They have worked well for you, and that is the bottom line. You have chosen wisely for that build, and that is all it matters for now. I am just trying to be as good as you are in making that decision.

 

[Bidule0hm]

I think you are overthinking all of this, the number of platters per drive should be one of the least of your concerns...

There's already a ton of things to think about on a FreeNAS server that are more important for reliability, durability, etc. by many orders of magnitude than the number of platters ;)

 

[nightshade00013]

Significantly reduced rate of "oops, I hit the USB drive and now all that's left is the PCB barely hanging on to the USB plug" situations.

There's also the question of why a SATA device instead of USB - it's simply because USB drives tend to be very unreliable. Some people have been through several USB drives since 9.3 was released and failing boot deices became easy to identify.

A lot of the newer boards are also being made with internal USB ports so that you can avoid the oops factor and not lose a sata port in the process. Otherwise I would actually think about something like this http://www.frozencpu.com/products/5...le_Port_Cable.html?tl=g2c539s1558&id=GETf2hQW and just make sure that it is secured with a zip tie to help make sure it is not inadvertently unplugged from the header .

I do have a question though, if you are going to build out an array using Raid-z3 array would it be better to use more 2tb drives or less drives of a larger capacity to end up with less drives in the array?

 

[cyberjock]

I do have a question though, if you are going to build out an array using Raid-z3 array would it be better to use more 2tb drives or less drives of a larger capacity to end up with less drives in the array?

That depends. Will more spindles in a vdev help your workload type? There are advantages to more spindle vs less. The question is whether it help YOU and YOUR workload. If you are like most people here, you are using this for storing media and such, and more spindles really don't matter too much. If this is a production environment where you have dozens or hundreds of simultaneous users, you not only want lots of spindles, you want lots of vdevs too!

 

[nightshade00013]

That depends. Will more spindles in a vdev help your workload type? There are advantages to more spindle vs less. The question is whether it help YOU and YOUR workload. If you are like most people here, you are using this for storing media and such, and more spindles really don't matter too much. If this is a production environment where you have dozens or hundreds of simultaneous users, you not only want lots of spindles, you want lots of vdevs too!

This is going to be a home system but I am also looking about the reliability aspect, a lot of the transfer will be pictures and small files by two to three users which doesn't really mean a whole lot speed wise. The rest will probably be streaming video in the house and maybe a little remotely to my step daughter or myself if I get stuck somewhere with nothing much to do.

I know smaller drives will be faster to rebuild but at what point is an array too large to consider safe/sane. I figure I will run zfs3 with either 2tb or 3tb drives with 7 drives in the array. I am planning to replace all the drives after about 6 years rotating them out at about one drive per month which will give me the ability to have newer drives as well as a large pool should it be needed. The old drives can then be relegated to some other duty that I am less concerned with loss. I don't know if I am going overkill on this thought process but integrity is one of the things I am wanting to insure as best as possible.

 

[cyberjock]

I'm doing 10x6TB in RAIDZ2. Does that answer your questions about reliability?

 

[nightshade00013]

Yeah sounds like it's not something to really worry about then.

 

[Ahjohng]

I'm doing 10x6TB in RAIDZ2. Does that answer your questions about reliability?

If you accept the manufacturer's specification of 1M hours between failures, there is a 29% chance that 1 drive, out of 10, would fail in the next 5 years and 6% chance that two would fail. There is also 1% chance that three disks would fail.

Maybe Western Digital pays more attention to quality control of their flagship product which results in lower DOA's, but as you have acknowledged, the 5-platters of 6TB disks would suffer a higher failure rate than lower-platter devices carrying forward. This means theoretically the numbers would go up.

So, good luck with that build. I just wish there is a way to monitor the reliability of your system.

 

[zambanini]

ahjong....lol

you can do a 4way mirror, or ten way mirror. there will be always a risk from different sites. that is also a reasson why you need a backup and where replication can help you. mtbf in the IT is often based on assuming and...