Toshiba vs WD for NAS

chrislip · 22 Jan 2020 at 14:15

Hi all

I have just brought my Microserver 8 into production with UNRAID and am looking at addng some disks to it. For my nested ESXi I have 4 * 3 TB WD Red drives in a SHR configuration in Xpenology. Now I'm looking at moving my videos across to the UNRAID system an want to know the best drive to install in this. Looking at the OCUK page I can get a Toshiba 4TB X300 drive for less than an WD 3 TB Red, which is obviously quite a difference! If the Toshiba drive was not on sale it would be same price as a WD 3TB Red which makes me think it is not as reliable, however I'm planning on having 2 parity drives in my system so should I go for the Toshiba which would give me more space for the same money?

Avalon · 22 Jan 2020 at 20:49

In my personal experience of running larger disk pools, WD Red’s fail - as often as anything else/to live up to the hype. Toshiba have historically been pretty reliable storage wise, but if the content is relatively static (eg media) and you have a (reasonably) fast internet connection, then you really should be considering cloud based storage mounted locally + a VPS or dedicated server, in many cases it’s cheaper than running something older locally in terms of power and the connectivity is massively better. Obviously some things are better kept local though.

Firejack · 22 Jan 2020 at 21:47

Hey,
Originally I setup unRAID using some old Samsung F1 & WD Blue HDD's I had spare. Performance was bad - especially writes. Turbo Write helped but then all the drives were spinning constantly. At that time I was reading about other setups and the advice seemed to be to have WD Reds as the Data drives and WD Blacks as the Parity drive. In multiple simultaneous writes situations, having the faster Parity drive would help performance. Combined with the fact the WD Reds were quieter and designed for long life even in systems with many HDDs all creating vibrations.

I swapped out my old HDD's for 2x WD 4TB Reds and performance was better. I then added a VM with my Dockers and unRAID was grinding to a slow crawl again. This is when I added a cache drive to unRAID (currently 1x Corsair MP510 240GB NVMe) . Suddenly write performance wasn't an issue and I was getting near enough full Gigabit Read/Write over my network.

My Array is 88% full and I've been looking around all the options for new HDDs like you. Since I keep backups on another system and performance is no longer an issue. I'm going to expand the Array using the best value in terms of capacity for my money.

There is a temptation to over engineer the drives in unRAID. Really though you don't need to paying that 25%+ premium as like Avalon said above, there is little difference in the failure rate between the cheaper and most expensive drives.

Avalon · 23 Jan 2020 at 11:52

Firejack said:
Hey,
Originally I setup unRAID using some old Samsung F1 & WD Blue HDD's I had spare. Performance was bad - especially writes. Turbo Write helped but then all the drives were spinning constantly. At that time I was reading about other setups and the advice seemed to be to have WD Reds as the Data drives and WD Blacks as the Parity drive. In multiple simultaneous writes situations, having the faster Parity drive would help performance. Combined with the fact the WD Reds were quieter and designed for long life even in systems with many HDDs all creating vibrations.

I swapped out my old HDD's for 2x WD 4TB Reds and performance was better. I then added a VM with my Dockers and unRAID was grinding to a slow crawl again. This is when I added a cache drive to unRAID (currently 1x Corsair MP510 240GB NVMe) . Suddenly write performance wasn't an issue and I was getting near enough full Gigabit Read/Write over my network.

My Array is 88% full and I've been looking around all the options for new HDDs like you. Since I keep backups on another system and performance is no longer an issue. I'm going to expand the Array using the best value in terms of capacity for my money.

There is a temptation to over engineer the drives in unRAID. Really though you don't need to paying that 25%+ premium as like Avalon said above, there is little difference in the failure rate between the cheaper and most expensive drives.

Writes to the storage pool were slow because it was writing to slower drives with lower platter density and in simple terms it had to write to them, calculate parity and write that to the parity drive. Implementing a cache drive means all data is written to cache and then at pre-defined intervals to the pool. In practical terms for most people not running a managed switch and doing multiple client writes or 10Gb, that's 100MB/s ish allowing for overheads, once you move to an SSD you still have the same remote write cap, but that still leaves more than enough IO capacity for dockers and VM's, a mechanical drive would be horrible at this point. Once you have a large enough SSD for cache, pool speed for writes becomes largely irrelevant and reads are always going to outpace reads, unless you have a large number of clients doing reads from the same drive.

As someone who went down the 'i'll add lots of local storage using high capacity drives' route, before sinking large amounts of cash into physical storage, for things like 'media' which is static and easily replaced, look at using inexpensive cloud storage mounted locally, combine it with a cheap VPS and it's a game changer.

FreeStream · 24 Jan 2020 at 19:06

Get two/three 8/10/12/14TB WD Elements/MyBooks and shuck them.

Got four myself 8 months ago to replace 9x4TBs in my UnRAID server. Going to get myself from 14TBs when they are next on offer.

Slowed down enterprise drives, so perfect for NAS!

Rroff · 25 Jan 2020 at 00:33

My main NAS is running a bunch of Seagate Barracuda 7200.12s in there as I had them sitting about (sadly about half of the drives seem to throw up a small number of reallocated sector counts after awhile but none has failed yet) externally I then backup to Toshiba drives - not intentionally just how it worked out. So far I've not had any data loss or corruption that I'm aware of with this setup and been running it for many years without so far any critical drive failures.

We've not had great experiences with WD in as NAS setting in this household though - my brother has had several of them fail entirely and/or creeping corruption that wasn't detected at first and none of my WD drives used in other setups - mostly with less use than my main NAS - continue working to this day. I'm a little hesitant to recommend Seagate though, despite being a big fan for many years, as they aren't half the company they used to be.

pc-guy · 29 Jan 2020 at 03:27

I wouldn’t get normal desktop HDD to go into NAS without some kind of redundancy and backup. NAS drives do have better reliability.

consider all manufacturers. Why limit yourselves to Tosh and WD?? Get whatever is in you budget for the best performance and endurance.

Avalon · 29 Jan 2020 at 07:09

pc-guy said:
I wouldn’t get normal desktop HDD to go into NAS without some kind of redundancy and backup. NAS drives do have better reliability.

consider all manufacturers. Why limit yourselves to Tosh and WD?? Get whatever is in you budget for the best performance and endurance.

If you’re going to claim NAS drives are more reliable, I assume you have a source to back that up? I’d love to see the data as frankly it disagrees with pretty much everything I’ve seen published or personally. Enterprise drives are rated with higher MTBF, but consumer NAS drives tend to be little more than you paying more for an enhanced warranty/RMA process, and at least in WD’s case, it’s a joke.