Wednesday, February 27, 2013
Here's a summary of worthwhile news for the last couple news in the realm of storage, let's start with hard disk drives (HDDs) and potential ways to push the envelope for more capacity beyond perpendicular recording and GMR heads:
Shingled writes (picture below), HAMR (heat-assisted recording) and Helium-filled drives seem to be three proposed solutions to the challenge! While HAMR and Shingled writes would help to increase aerial density of HDD platters, the Helium used inside HDDs aims rather at allowing more platters per real-estate by lowering air-friction. Combining the two would yield the best results in terms of capacity increase but also obviously raise manufacturing costs.
Aerial density growth for HDDs has been down at the 20% level (per year) in the last couple years as you can read in this article by Hu Yoshida here, so to maintain capacity expansion, new technologies will be needed!
That same article also has some interesting trends regarding flash disk cost levels: expectations are that the per capacity costs of SSDs will reach SAS HDD levels in a 2019 time frame.
1.2TB SAS drive from WD are still based on traditional architectures!
The reasons why we desperately need more capacity are discussed in the most recent "Digital Universe" study: IDC predicts that "...the digital universe will reach 40 zettabytes (ZB) by 2020, an amount that exceeds previous forecasts by 5 ZBs, resulting in a 50-fold growth from the beginning of 2010". I personally would argue that the kind of snapshots taken all over the world by cell phone users are not an "act of creating information"... but obviously these pictures still use up storage space in various places. Find the complete report here!
The related comment in this Storage Networking article here: "Digital hoarding: do we have a problem"?
And an outlook for the HDD market thru 2017 can be found here.
On the Wall Street side of our industry, the Q3 results of VMware generated some mixed echos: "VMware sees trouble ahead" was the title of an article in the financial press. But nevertheless, VMware continues to build out it's in-house storage ecosystem and announced the intent to acquire Virsto, the maker of "..the first VM-centric Storage Hypervisor".
Which closes the loop: Jon Toigo has his very own and distinct opinion about storage hypervisors (see graph above): "Then, when VMware engineers climbed on stage a year or so ago and said that they were going to add another microkernel they called a storage hypervisor to the clusterf#@k of microkernels that they fancy to be a OS, I couldn’t believe my ears. Imagine if each of the leading hypervisor vendors decide to do a storage virtualization microkernel too: that would suck."