Data centre disk use is spinning down – Wikibon report

Decades of disk domination are coming to a close: Wikibon sees flash replacing disk entirely for primary data storage enterprise and hyper-scale cloud data centres over the next ten years. Disk will still be used as the cheapest and fastest – for the price – bulk-capacity media for archival and other low-access rate data. The …

[Trevor_Pott]

Nice to see Wikibon more or less in line with my own thoughts on the matter. All companies having a "Tier 0" flash tier by 2020 that handles potentially up to 50% of workloads that would normally be considered "Tier 1". By 2025ish the whole of Tier 1 moved to Flash (or a post-flash technology).

I think it's possible we'll be able to build enough foundries to handle that.

It's where people start saying stupid things like "nearline storage will be replaced by flash" or "Tier 1 will be all flash within the next [insert very short timeframe here]" that I start getting tetchy. No, we damned well don't have the fab capacity for either of those scenarios.

I don't think nearline storage will ever be replaced by flash. It will probably take a post-flash technology to do the job. Meanwhile, Tier 0 applications will be on post-flash technologies by the time we've moved "all" Tier 1 applications to flash (2025ish).

Tier 2 and lower applications will still likely be on hybrid storage for a long time to come.

[Rob Isrob]

What about total PB?

Wait a second...

Some of us are paying attention out here, by the way.

What about this, doesn't HD growth still outpace SSD in the future (see chart in link below):

http://www.theregister.co.uk/2014/12/09/no_flash_datacentre_takeover/

Is it no accident WikiBon study doesn't touch on total PB on the floor and ratios SSD<->HD?

[Anonymous Coward]

Re: What about total PB?

Nope, the future got here early. I sell storage for a living and it's rapidly getting to the point where I can't even give away spinning rust but SSD appetite seems endless. One big bonus to SSD is the "one pool to rule them all" management structure. It really resonates with customers.

[Anonymous Coward]

Useless graph

This graph needs a log scale on the vertical axis.

Everything looks the same if it all looks like zero. But maybe flash will still be 10 times as expensive as disk - and maybe that will still matter, since the volume of data we store will have gone up by a correspondingly huge factor as well.

[DavCrav]

Re: Useless graph

"This graph needs a log scale on the vertical axis.

Everything looks the same if it all looks like zero. But maybe flash will still be 10 times as expensive as disk - and maybe that will still matter, since the volume of data we store will have gone up by a correspondingly huge factor as well."

I read that graph as saying "Look! A graph to prove my point! Since I have graphs I can't be wrong!"

[Bob H]

In the grand scheme of things most companies don't generate PB of data and capacities are generally more boring than we expect. The idea that capacities would grow exponentially is a little flawed.

[ntevanza]

It's not about the hardware.

It's not about the hardware.

It's not about the hardware.

It's not about the hardware.

[Anonymous Coward]

You'll still be buying spinning...

Because no one is going to invest a $trillion to go and build sufficient flash capacity ijn the next 10 years.

Thus, flash will reach a base price based on supply-demand and the remainder will be satisfied with spinning rusty stuff.