c3

They are 7200rpm, hence the statement of being 25% faster than other brands. http://www.storagereview.com/seagate_nas_hdd_8tb_review

Please don't don't ban him for being on topic, but either new to online posting (thus not knowing all the features [edit message]) or posting a link to a source. The information is correct. - You need to use new batteries. Old batteries will have short life. - A source is mentioned (is this ban worthy?) - The source location is given, a good thing. - A UPS with a dead battery should not be discarded. Batteries are typically much cheaper than a new UPS, especially is sourced by companies like the one linked. - and a comment that the source can help with selection and even take old batteries. Overall the posts give me a sense that english may not be the first language of the writer.

OCZ vectors had problems. There was a firmware update, which might help. Depending on which version of Windows, some settings may be needed (sleep/wake).

This type of post is NOT frowned upon.

I would check the manufacture date. The naming indicates it is the newer version, but that may just be packaging.

So, you mentioned all the heads touching down, which is likely if the motor fails (drops rpms). But it is also possible that one head crashed and the resulting debris got into all the remaining heads. I'm not sure anything can get the dust back in the right places, or sort it.

uhm, one small difference, I really shouldn't even be mentioning it, the slicestor has CPU, memory, NIC, etc, and the HGST has none of these.

The 8TB discussion on them http://lime-technology.com/forum/index.php?topic=52274.0

The IronWolf (ST8000VN0002) is not SMR (Shingled), thus very different. Is it worth $50 is dependent on workload, etc.

uhm, pretty sure a made in Japan drive would not come from China, and since I can get same day delivery where ever they were bulk shipped from, they are now sitting in Amazon warehouses like the ones down the street from my house. I wish they were made in China as another serious manufacturer is needed. These look to be rebranded Toshiba, very much Hitachi split off.

Sometimes debated, here is solid dataset showing extended drive life, reduced failure rates. http://storageconference.us/2016/Slides/KirillMalkin.pdf

You might be looking for the $/TB yield. The attached spreadsheet might help. Blue areas are input; your target capacity needed, and the current prices for drives. Cost-per-TB.zip

There is plenty of work on the other filesystems. But you wont see much until after 4.8

This information is incorrect. The version 2 is ST8000AS0002, which should be pretty obvious. Yes, there was a version 1. While it is true the ST8000AS0022 can be used as a drop-in replacement for the ST8000AS0002, the ST8000AS0022 is capable of so much more (with the right software).

http://lime-technology.com/forum/index.php?topic=50478.0 and older news http://lime-technology.com/forum/index.php?topic=45608.0

Does not ship to US.

It's not the helium, it is the tall platter stack that forced the screw to move.

4TB drives hold price because they are already damn cheap (<$25/TB). There is a significant install base, so they wont go the way of the 3TB as fast as the 5TB and 6TB. As far as the other articles, seems those sources don't understand the paper. http://www.semiconductors.org/main/2015_international_technology_roadmap_for_semiconductors_itrs/ First the roadmap is only the plan, and only out to 2030, not 2040. Moore's law, depending on what you think it is, has already been broken, or has a very long way to go. The transistors and connections will continue to increase thanks to 3d semiconductors. Compute power will continue to increase as a result. New technologies keep giving it life. It is not unusual to say by 2030 something new will be required to go further. To be on topic, in the disk drive world, things like SMR and HAMR. The processor design game has been processing power per watt for several years new. And energy production continues to climb. The combined chart may show an intersection, but again, that is meanly to point out the area for new development. I'd like to see the chart of the total solar energy reaching the planet consumed by computers by 2040, and supporting math. As for aircraft carriers, no recent constructed aircraft carrier runs on oil.

I am not sure what the question is. It is true 4TB drives are not changing price much ($25/TB is pretty low/good already, routinely seen on Fry's promo code email). The archive drives (SMR) are different, and it is reasonable to use or not use as you see best for you workload. The 8TB, and now 10TB drives, have put pressure on the 6TB pricing, but again the 4TB is already good.

According to this post, it says the m1015 is compatible with unraid. Do you mean that unraid recognized the card but won't act like an HBA? https://lime-technology.com/wiki/index.php/Hardware_Compatibility#PCI_SATA_Controllers The wiki indicates the m1015 must be cross flashed to be supported. It's on the right side.

Considering the per GB is lower on the larger, it is pretty easy choice. In the past the two did not align normally.

+1

What does losing ZFS filesystem mean to you? ZFS has many features. You mentioned scrubbing, which unRAID does not do, but is can schedule a parity check. unRAID currently has only single parity (more is in beta). ZFS has a much more robust data integrity offerings, like more parity. But if you did not use those, that's not want you are losing. ZFS has several tiers of caching. unRAID has caching, but it is a very different architecture. You already know the biggest advantage of using unRAID, no forklift upgrade. Which is a big deal for most home users.

This is actually pretty easy to recover from. 443 bytes are just the MBR, and you only touched the first copy of the partition table.

There is no doubt that the rapid falling price of SSDs makes these valid questions. SSDs are the top of any performance/dollar chart. Large SATA drives are the top of any GB/dollar chart. 1a) No, the VMs should run from SSDs, and the NAS should be bigger+slower HDD. Cheap SSDs are not a good place to start, as they are low on the performance curve. You said put money into speed and reliability, that means no cheap SSDs. 1b) True, but it is likely to be less costly to have a fast set of cache SSDs, than a full set of SSDs for everything. 1c) I think this is the route to look into, solid fast SSDs for cache, and Dockers, and VMs. Then HDD for expansive storage. 2) What is not clear is your space requirement for VMs, or even cache. As mentioned above, this is the area to focus on for performance. 3a) I would not mix SSDs and HDDs in the array, even if supported. The SSD will perform for reads, but not writes. The BTRFS based cache pool will allow the VMs to continue to run after an SSD failure. 3b) Mentioned above, I don't think it makes sense to mix SSDs and HDD in the array. 5) Potentially the RAID1/5 (don't use RAID0) could be used for the cache drive instead of BTRFS. Roughly, I would suggest a pair of 900+GB SSDs in RAID1, for cache and VMs, and three HDDs (3TB-6TB each). But that vastly oversizes the 2-4TB as 6-12TB. If you really mean 4TB max, not going to need to grow it, you might look at dropping unRAID and use a fast controller with battery backed cache and 6 or more SSDs. Without need to grow, and using all SSD, unRAID's major features (easy expansion and spin down) are not needed. Also expect more real offerings in the NVMe space this year.