That's because there is no actual latin letter C in the title. For whatever reason, there is the kyrillic uppercase letter "Es" (U+0421) in the title, which basically looks like a latin C, but is a different character.
It seems the font used for the title (Rokkitt) does not provide a glyph for character U+0421 (the kyrillic uppercase "Es"), making the browser fall back to a serif font (serif is mandated by the title's CSS style) that provides a glyph for U+0421. Which happens to be Times New Roman (not sure what the fallback font would be on Linux, Mac or mobile browsers...)
I recall just reading a few short months ago that manufactures/customers had no desire to move beyond 16TBs in a single drive as they thought that was just too much data to rebuild should they have a hardware failure. Guess, that concern is nill.. as 20TB represents a 25% increase over 16TB. That's allot of consumer purchasing data to store..... Orwell would be proud.
It depends on ops. Ours have no desire to go beyond around 4TB for HDD. Recovery from single drive failure takes hours or even days. Fundamental problem of HDD is that capacity is increased with number of tracks, density of data on tracks (and number of platters to very limited extent) while speed is increased only with density of data on track and that is only relevant for linear speed. Ie speed is increasing with square root of capacity. HDDs are literally outgrowing their usefulness.
Yeah, even a 4-disk RAID5 makes me nervous with 2-4TB as it takes about 8-24 hours to rebuild. Saw a good 4x12TB offer from OVH recently but I'm not sure I'd go for it if I expected to use all 12TB.
I suspect redundancy levels also play a factor. If you can only survive a single drive failure without data loss your need to load the replacement asap is a lot higher than if you're able to survive 2 or more drive failures.
Do any of the big cloud providers provide public information on how redundant their data storage setups are?
I'm hoping this means NAS/enterprise drives finally move the needle in cost/TB. We've been at $25/TB for half a decade (if not longer?). 2014 to 2019 came & passed us by with essentially zero price changes per TB.
Surely 6 TB / 8 TB drives should've dropped in price by now, but nope.
i dont believe these enterprise growth projections for one nano second. this largely assumes the growth of content (online video) will go to hard drives. thats where they are dead wrong. hard drive bandwidth and performance is not good enough for these use cases requiring RAID and replication games of all sorts. SSDs can serve this traffic with much lower replication factor.
cold storage and not content is what these drives will be useful for. their growth estimates are pure fantasy
Lots of content is cold. We have some 8-year-old content. Much of it *could* be accessed at any point, but probably won't be - if it is, it needs to be available within a second or so. HDD's good for this.
Don't get me wrong, we'd love to go full-SSD, but today we can't really do that on a single server. It's close, but not quite the default yet, and not at our price point. Soon, yes. For now, NVMe+HDD.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
18 Comments
Back to Article
keg504 - Monday, December 2, 2019 - link
What's with the Times New Roman "C" in the title?jordanclock - Monday, December 2, 2019 - link
I am so glad I'm not the only one thrown off by that.TheSkullCaveIsADarkPlace - Monday, December 2, 2019 - link
That's because there is no actual latin letter C in the title. For whatever reason, there is the kyrillic uppercase letter "Es" (U+0421) in the title, which basically looks like a latin C, but is a different character.It seems the font used for the title (Rokkitt) does not provide a glyph for character U+0421 (the kyrillic uppercase "Es"), making the browser fall back to a serif font (serif is mandated by the title's CSS style) that provides a glyph for U+0421. Which happens to be Times New Roman (not sure what the fallback font would be on Linux, Mac or mobile browsers...)
Ryan Smith - Monday, December 2, 2019 - link
Thanks. I'm not sure how that happened, but it's been fixed.HardwareDufus - Monday, December 2, 2019 - link
I recall just reading a few short months ago that manufactures/customers had no desire to move beyond 16TBs in a single drive as they thought that was just too much data to rebuild should they have a hardware failure. Guess, that concern is nill.. as 20TB represents a 25% increase over 16TB.That's allot of consumer purchasing data to store..... Orwell would be proud.
qap - Monday, December 2, 2019 - link
It depends on ops. Ours have no desire to go beyond around 4TB for HDD. Recovery from single drive failure takes hours or even days.Fundamental problem of HDD is that capacity is increased with number of tracks, density of data on tracks (and number of platters to very limited extent) while speed is increased only with density of data on track and that is only relevant for linear speed. Ie speed is increasing with square root of capacity. HDDs are literally outgrowing their usefulness.
GreenReaper - Tuesday, December 3, 2019 - link
Yeah, even a 4-disk RAID5 makes me nervous with 2-4TB as it takes about 8-24 hours to rebuild. Saw a good 4x12TB offer from OVH recently but I'm not sure I'd go for it if I expected to use all 12TB.DanNeely - Monday, December 2, 2019 - link
I suspect redundancy levels also play a factor. If you can only survive a single drive failure without data loss your need to load the replacement asap is a lot higher than if you're able to survive 2 or more drive failures.Do any of the big cloud providers provide public information on how redundant their data storage setups are?
sheh - Monday, December 2, 2019 - link
https://www.backblaze.com/blog/vault-cloud-storage...rpg1966 - Monday, December 2, 2019 - link
* a lot/pedant
ikjadoon - Monday, December 2, 2019 - link
I'm hoping this means NAS/enterprise drives finally move the needle in cost/TB. We've been at $25/TB for half a decade (if not longer?). 2014 to 2019 came & passed us by with essentially zero price changes per TB.Surely 6 TB / 8 TB drives should've dropped in price by now, but nope.
azfacea - Monday, December 2, 2019 - link
i dont believe these enterprise growth projections for one nano second. this largely assumes the growth of content (online video) will go to hard drives. thats where they are dead wrong. hard drive bandwidth and performance is not good enough for these use cases requiring RAID and replication games of all sorts. SSDs can serve this traffic with much lower replication factor.cold storage and not content is what these drives will be useful for. their growth estimates are pure fantasy
TheinsanegamerN - Tuesday, December 3, 2019 - link
SSDs are WAYYY too expensive for large scale video storage, and online video projects already have issues turning a profit.GreenReaper - Tuesday, December 3, 2019 - link
Lots of content is cold. We have some 8-year-old content. Much of it *could* be accessed at any point, but probably won't be - if it is, it needs to be available within a second or so. HDD's good for this.Don't get me wrong, we'd love to go full-SSD, but today we can't really do that on a single server. It's close, but not quite the default yet, and not at our price point. Soon, yes. For now, NVMe+HDD.