Bigger drives or more drives?


TiGuy

Recommended Posts

Hey guys

 

Currently im having 7x1TB including parity and a 160GB SSD as cache drive, each of them is in a Thermaltake Max4 for easy accessiblity. But now im kinda at a turning point, with less than 200GB of free space and adding more data each day i will max my raid really soon.

 

So i have 2 option, begin to replace few drives including parity for something like WD 4TB or changing my setup to be able to stack something like 16 drives and add 8 more 1TB (We got tons of unused at work so i can get them really cheap). My questions is whats the ideal ammount of drives? Is there any drawback to always add more drives for parity or anything else?

 

Any hint would be more than welcome

 

Regards

Link to comment

If the extra drives and your electricity are free, and you have the physical room and free SATA ports, then it's pretty much a no brainer. However... every drive you add increases the odds of multiple drives having issues at the same time. Also, adding slots, either ports with an HBA card or extra drive bays isn't free, and the added cost of those will probably outweigh the cost of a couple 4TB drives to start you off with your expansion in the existing setup. I'll be honest, even with free 1TB drives, I'd probably still go with the 4TB. Your power supply may also need to be upgraded to handle twice the number of older 1TB drives.

 

Gary will say you would be silly not to go with the 6TB while you are at it, but Gary's made of money, and has a fairly high tolerance for risk since he keeps scrupulous full backups. I'd rather not get a bunch of 6TB drives until they've been on the market for a while and the longer term failure rate is better known.

 

The ideal amount of drives depends on you and your tolerance for risk and amount you want to spend. Fewer spindles tends to be cheaper and less hassle in the long run.

Link to comment

I agree as well.

 

The largest risk I see with your 1TB solution is that if these are retired drives from work you don't know how close to end of life they are, and as jonathanm mentions you increase your risk of multiple drive failures.

 

I would definitely start moving to 4TB drives, or 6TB drives if you have the money. I have a mix of 3 & 4 TB disks, but my next purchase will be a 6TB for parity and every drive after that will be 6TB as well. I only buy WD Reds now and don't have real concern of failure, though it's always an option. Either way, be it 4TB or 6TB I think that is a smarter approach.

Link to comment

Gary will say you would be silly not to go with the 6TB while you are at it ...

 

Given that the current array only has 1TB drives in it, moving to a 4TB parity, and then using 4TB drives isn't a bad choice.    But yes, I would personally buy a 6TB parity drive in this situation, and then start replacing 1TB units with 6TB units.  Except for the one-time expense of the parity drive, which is clearly higher if you go with 6TB, the cost/TB of the additional data drives is very nominal.  At today's Newegg prices, 4TB WD Reds are $43.25/TB, 6TB units are $49.50/TB.    So for a $6.25/TB premium, you can have 50% higher per/SATA-port storage;  33% lower power consumption/TB; and faster drives (the 6TB units are 1.2TB/platter units, so they have a 20% higher sustained data rate than the 4TB units).

 

But if the initial outlay is too high for you, just use the 4TB units.

 

 

... Gary's made of money

 

8) 8)    I wish.  My wife would certainly disagree with that !!

 

 

Gary ... has a fairly high tolerance for risk since he keeps scrupulous full backups.

 

Actually I have a very LOW tolerance for risk ... that's why I ensure I'm always backed up.  Running WITHOUT backups is by far a larger risk.

 

 

I absolutely agree that if you can get a bunch of 1TB drives for free (or nearly so), I'd still pick those up, and use them for backups.    Nothing protects your data better than a current set of backups.

 

 

Link to comment

Gary ... has a fairly high tolerance for risk since he keeps scrupulous full backups.

 

Actually I have a very LOW tolerance for risk ... that's why I ensure I'm always backed up.  Running WITHOUT backups is by far a larger risk.

Let me clarify a little. You have a higher tolerance for risk in your primary array, because you can always restore from backup. People who don't have full backups get nervous when 1 drive fails. You are covered in almost any scenario, so running on the bleeding edge with your primary array is fine.
Link to comment

The older 1TB drives can be used to keep your backups of he unRAID data.  If you have one of those USB adapters that allows you to plug in bare 3.5" drives they should not take up too much space.

This is what I do with the smaller drives after I replace them with larger ones. Use them to backup critical files then store them offsite.
Link to comment

Larger drives vs more drives ... the choice is not so clear.

 

Although fewer larger drives means lower risk of any data loss, more smaller drives risk losing a much smaller amount of data if it does happen.

 

Larger drives tend to cost less per TB than smaller ones because of the inherent cost of any drive. So 4 1T drives would cost a lot more than 1 4T drive. Drive manufacturers charge a premium based on market conditions for the largest drives, so this may not always be true. But once a drive has been on the market for 6 mo. to a year, it is typically true. My cheapest 4T drive cost $130, which is $32.50 per TB. If you are getting the 1T drives for less than that (e.g., free), growing the array (more slots) may be cheaper. But even doing it cheaply, slots are going to cost you at least $15 a piece (see bottom of this post), and the expandability limits are much less. Personally 1T drives have gotten too small for me to worry about in my primary array (although I still have quite a few in my backup array). Even if I got them free I wouldn't use them in my main array. But my storage needs may be much higher than yours, so you need to make the decision for yourself. Lots of things to consider. See below ..

 

I suggest adding drives over time, based on the most economical / reliable. Historically that has meant users are getting a variety of sizes over time.  When it comes time to upgrade you are replacing the smallest disk with one that is much larger. Upgrading a 6 year old 500G drive to 4T doesn't hurt much. But this has been more challenging as we got stuck at 3T and 4T for quite a while. Upgrading a 4 year old 3T drive to 4T would hurt a lot. So, IMO, more slots should not be viewed so much as a way to expand to a much higher capacity array, but instead a way to extend the investment of your drives you have and ultimately reduce the number of disk upgrades you have to buy in the long run.

 

People will get themselves into trouble (IMO), if they use all of their limited data slots over a 2-3 year time period, and then are forced to upgrade large drives with larger drives. For example, imagine a user with 5 data slots and already have 5 4T drives = 20T. That user would be looking at swapping out 4T for 6T (at a huge price per T since he's only adding 2T/drive).  It is false thinking for someone with such an array to say that their array is expandable to 30T because, although technically true, the cost would be prohibitive. I think we have a lot of users with smaller arrays that will get into this crunch. For that reason, for people that insist on a small server with few slots and consume space at a brisk pace, I always recommend investing in the largest disks possible, even it if it is at a premium (within reason). Because it is so much cheaper than upgrading.

 

Of course there are options beyond upgrading disks. You can setup a second array, or you can do what I did on my backup server:

 

Tethered DIY Addon Enclosure

 

My backup server, an Antec 900, was maxed out with 3 5in3s = 15 drives. But I wanted more to hold more leftover disks from upgrades over the years for backups.

 

I found THESE 4in3s from Newegg for $45 each. $90 for 2.

 

16-132-037-02.jpg

 

 

And THESE SAS to SATA breakout cables I bought were pretty long and worked well at $15 each=$30 for 2.

 

So for $120 I was able to add 8 more slots to my array=23 total. I screwed the two 4in3s together and ran the breakout cables out the back of my server along with an existing power supply run using a spiraling wire wrap so it looks like a conduit. Voila - a small external drive enclosure tethered to the server supporting 8 more drives. (Note that I had an extra BR10i that I used which is limited to 2T drives, but that is not a problem in my backup array).

 

BTW - although not the quality of the Supermicro 5in3s, these Rosewills are not bad. 120mm fans on the back cool well. Bottom (hot part) of drive well ventilated. No heat issues at all. The bright blue LEDs on each slot (8 of them) are a little annoying but it sits in my basement so I don't care. 4 is actually a good size for putting these outside of the array because one SAS connector is 4 SATA ports, and 2 (a full 8 port controller card) is two.

Link to comment

... so running on the bleeding edge with your primary array is fine.

 

... I suppose you may think that, but I don't.  I do NOT run "... on the bleeding edge ..." with any of my servers or PC's.    Very conservative equipment; no overclocking;  never more than 2 memory modules on boards with unbuffered RAM;  high quality hard drives;  UPS's on every computer I have; etc.    Hardly what I'd call "high risk" or "bleeding edge"  :)

 

Link to comment

... so running on the bleeding edge with your primary array is fine.

 

I think the reference was in that you have already adopted 6TB drives, while others are waiting a little bit to see if any issues crop up over the initial few months as they get an initial population in the field vs a handful in various labs.

 

I know my next purchase will be the 6TB Red, but I am going to wait 2-3 more months, I hope.

Link to comment
I think the reference was in that you have already adopted 6TB drives, while others are waiting a little bit to see if any issues crop up over the initial few months as they get an initial population in the field vs a handful in various labs.

 

I know my next purchase will be the 6TB Red, but I am going to wait 2-3 more months, I hope.

Exactly what I was getting at. I was burnt as a computer reseller by the IBM deathstars, and a few other random models of various brands over the years, so I always wait a little while to see if a particular model has any problems in wide distribution instead of jumping on the first batch of anything.
Link to comment

Understand.  If you read my various posts, I've said frequently that "my next drives will be the 6TB Reds".    I have not, in fact, bought any yet ... as I simply don't need more storage at the moment.  The reality is it'll probably be Oct or Nov before I actually buy any -- not because I have any concerns about them, but simply because I don't need them.    I suspect by that time if there are any significant issues, they'll be well publicized, and I'd likely change my mind  :)

 

... but given the superb history of the WD Reds, I seriously doubt there are any issues.  As you may know, WD originally projected those drives would be available by the end of last year ... but it was over 6 months later before they actually shipped them.    I suspect this time was put to good use in ensuring the manufacturing facilities were in fact producing quality products (and perhaps redesigning any flaws they had discovered in testing).    I think there's very little risk in buying these drives -- and they do, of course, have a 3-year warranty "just in case."

 

Link to comment

WD 1T greens (EACS/EADS) were awesome.

WD 2T greens (EADS) sucked.

 

Seagate 300G, 500G, 750G, 1T Barracuda were awesome

Seagate 1.5T, 2T sucked.

 

Hitachi (IBM) Deskstar (Deathstar) sucked.

Hitachi 2T, 3T, 4T awesome.

 

Each drive needs to prove itself.

 

Gary - you are smart to wait (by design or accident).

 

There are currently 4 reviews on Amazon and 1 on Newegg for the 6TB REDS.

Link to comment

Understand.  If you read my various posts, I've said frequently that "my next drives will be the 6TB Reds".    I have not, in fact, bought any yet

So why do you advocate that others buy them right now when asked for advice on upgrades?
Gary - you are smart to wait (by design or accident).
Yet he's pushing hard to get others on here to buy the 6TB drives. I suppose you would rather others get burned if there happens to be an issue?

 

I'm not saying there will be an issue with any particular model, I hope your advice turns out well. It just seems a little wrong to enthusiastically tell people to jump on the 6TB drives right now when you don't have any experience with them. I (wrongly) assumed you had already used them in your array since you were advocating for them.

 

As a moderator on this forum, your advice is probably taken more seriously than the average poster.

Link to comment

I HAVE bought 8 of them for other friends systems -- and they've had absolutely no issues with them.    I just haven't bought any for myself.  As I noted, I have NO problems recommending them, as I have complete confidence that WD did a thorough QC before they started shipping them.

 

I just haven't needed any for my personal systems.    I've also not seen ANY negative reviews ... at Newegg, Amazon, Tiger Direct, Provantage, CDW, etc.  ... nor at any of the independent review sites.

 

I have no WD stock, nor any vested interest in these drives.  I'm simply a fan of the WD Reds, and think the new 6TB units are a superb choice because they're the only 1.2TB/platter drives.    The 6TB Seagates and Hitachi's are probably just as good -- especially the Helium-filled units from Hitachi, but those are 7-platter units that will be appreciably slower than the Reds.  If money's not an object the 6 platter Seagate Enterprise class 1TB/platter 7200 rpm units should be the fastest available., but I don't recommend 7200rpm drives for UnRAID due to the higher power consumption and heat generation.

 

I'm building another system for a friend in late September/early October (when we get back from a 3-week vacation in September) ... and it will also use the 6TB Reds (probably 5 of them).

 

Link to comment

I agree with jonathanm and others that the 6TBs are awfully new and not proven. As I've stated over and over, and Gary ignores over and over, there is history of the exact same drive series from WD and other manufacturers having issues with the next greater drive size. The WD 6TB, in particular, are worrisome to me exactly because they push the drive density to 1.2 TB/platter. No other drive manufacturer has done this. HGST, the MOST reliable drives IMO, explicitly keeps the density lower than other manufacturers. This does mean lower performance, but contributes to the higher reliability IMO. 1.2TB/platter was reported as problematic as the density was too high to reliably store data. If WD found a way to do it - and it appears they have based on early reports - more power to them. But I want it to prove itself reliable for a little longer.

 

Anyway, I don't fault Gary or anyone for using them for himself or friends. Or fault others that feel they are too new and expensive to invest in right now. I just try to give the factual pros and concerns and let people make there own decisions. I also give my personal opinion which you can take for what it is worth.

 

And my personal opinion is - 6TBs are riskier than 4TB and more expensive. We are seeing deals on the 4TBs right now and I for one have been taking advantage of them. 6TB will wait until a really good deal pops up - maybe black Friday timeframe.

 

And my personal opinion is - for folks with arrays almost full with 3TB+ disks, and adding more 4TBs is going to mean that you are going to get squeezed for space soon (you define soon) - for you I would suggest jumping to the 6TB. I wouldn't buy 5 of them, but 1 or 2 to tide you over until prices fall.

Link to comment

Certainly agree that if you're worried about the density, then avoiding the 1.2TB/platter units makes sense.  Clearly I'm not concerned about it -- I just don't think WD would have released the drives if their testing had shown issues with it.

 

But if you ARE concerned about that, then I still wouldn't buy 4TB drives.  One thing that's been visibly absent in the discussion of these is that there is also a 5TB WD Red, which uses the same 1TB platters as 1-4 TB units do.  The cost/TB is the same as the 6TB unit ... there's just one less TB  :)  [Currently $242.46  ($48.49/TB) at Amazon]

 

Not what I'd do (as is abundantly clear from my other posts) ... but perhaps a "less risky" move for those concerned about the reliability of 1.2TB/platter units.

 

 

Link to comment

Certainly agree that if you're worried about the density, then avoiding the 1.2TB/platter units makes sense.  Clearly I'm not concerned about it -- I just don't think WD would have released the drives if their testing had shown issues with it.

 

But if you ARE concerned about that, then I still wouldn't buy 4TB drives.  One thing that's been visibly absent in the discussion of these is that there is also a 5TB WD Red, which uses the same 1TB platters as 1-4 TB units do.  The cost/TB is the same as the 6TB unit ... there's just one less TB  :)  [Currently $242.46  ($48.49/TB) at Amazon]

 

Not what I'd do (as is abundantly clear from my other posts) ... but perhaps a "less risky" move for those concerned about the reliability of 1.2TB/platter units.

 

True - but yesterday I bought a 4T HGST NAS drive for $144. That is $36/TB. So that one extra TB costs $100. And for those 5T to be useful, you'd actually have to buy two of them (with one you're just upgrading parity and still only get 4TB of data (your old parity) for your $243 = $60.75/TB). Its an easy decision for me.

Link to comment

$48 vs. $36 per TB is just not a big deal to me.  I'd much rather have the higher capacity per SATA port.  Less power;  less heat;  and a higher net capacity for the system.

 

Prices in the $50/TB range are simply not at all bad for storage.  I DO balk a bit when they hit $100 or more (as with the Hitachi helium-filled units).    Although the first hard drive I ever bought cost me over $173 MILLION per terabyte  8)  [$173,076,923.08 to be precise.]

 

Link to comment

Damn i come back from a long weekend and my post got many reply

 

After reading all of it and searched for some price i decided to order 3 x 4TB. 6TB are too expensive at the moment, the best price i can get is 320$ while 4TB are at 175$ (Both are WD Red Edition) and im not sure if 6TB can be trusted for reliability (We had so much troubles with WD 3TB at work when they released it). With this purchase i will double my NAS space so it's will take some times before having to worry about free space, well i hope so :)

 

Thanks for all the reply guys :)

Link to comment

"Gary will say you would be silly not to go with the 6TB while you are at it, but Gary's made of money, and has a fairly high tolerance for risk since he keeps scrupulous full backups"

Ha Ha I love this statement especially as Im now doing the same thing and far from being made of money. I do take risks though but I have a few friends who are using these WD 6TB Reds and they have had no issues with them, not even a D.O.A. drive which is better than I got with the 3TB WD Reds and now loads of people use and recommend them. I think a lot of the issues with my D.O.A. 3TB Reds was down to poor packing. Like Gary I do have an offsite backup of my media though. I don't think a NAS or Server is a suitable backup system if you value your data or like me don't want to spend months re-encoding all your movies.

Link to comment

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.