Advertisement
Current Issue
October 2014
Issue: January 1, 2013

2013 Storage Forecast

By: Steve Modica
I get asked to do articles and talks like this pretty regularly. Like most pundits, I’m not often asked to review my own predictions. I just keep looking forward and ignoring all of my past prognostications. 

To change it up a little this year, let’s look at my predictions last year and see how they are holding up.

With the spindle shortage due to flooding in Thailand, I suspect we'll see an acceleration of SSD based products. It's always cheaper to "print" something like a chip or a circuit board than it is to machine and assemble it. The CRT is dead and the spindle is not far behind.

I think I get to claim WIN here. SSD prices are now hitting $1/GB, which is a very reasonable price. I’m expecting to see 2TB SSD drives hit the market early this year. Granted, some would consider $1,000 for a 1TB disk steep, but the performance benefit for your laptop or video storage toaster is amazing. It’s well worth the money.

FCoE is starting to become a reality. Linux and OpenSolaris are both offering targets now, so the likelihood of proprietary offerings in FCoE targets is going up quickly.  

I’ve been watching FCoE vendors for the last year and I have to say I’m a little disappointed. Cisco remains the only FCoE switch vendor that’s putting in the effort to work with developers. Qlogic has yet to apply any pricing pressure to the market with a new, lower-priced switch. 

So given the lack of a reasonably-priced switch and the lack of a well-supported, commercial FCoE target, I think FCoE is probably another year out and maybe more. It’s just not moving as quickly as I’d like to see. 

The danger of course is that Fibre Channel vendors start pushing hard on 16Gb infrastructure that locks everyone in for another technology cycle. It will require 40Gb Ethernet to convince people to switch and that will drag things out even further. 

10GbaseT to the desktop is coming fast. Motherboards are already releasing with 10GbaseT onboard and it's only a matter of time before Macs have something as well…perhaps on the new Mac Pros we're all hearing rumors about!

Not withstanding the fact that I mentioned a “New Mac Pro” (It was a rumor!), I think it’s fair to say that 10Gb to the desktop is happening now by way of Thunderbolt. Small Tree already has a 10Gb Thunderbolt product for Macs and we’ve already seen 10Gb motherboards from server vendors. We may never see 10Gb chips directly on an iMac or laptop, but I like this new model. Why carry around a 10Gb Ethernet chip when you’re on the road? Better to have a dock on your desk at work.

File system based RAIDs are starting to become fast and reliable. Now that Oracle and Network Appliance have made up, I expect to see things that are based on software RAID and clustering become widely available.  

BSD Unix has ZFS and Linux has BTRfs. Both of these file systems attempt to offer RAID–like redundancy without the need for a hardware RAID chip. Our own testing has shown that these file systems seem to perform extremely well and offer the ability to expand on the fly without rebuilding the file system.

There are vendors today offering these capabilities using ZFS under the covers. I expect to see this trickle down into lower priced offerings from more vendors this year. I’m calling this one a WIN.

So now that we’ve considered my last year’s guesses, what does this year look like? In short: 

1 - New 7200RPM spinning disk that are dual ported SAS rather than SATA
2 - Advanced format sectoring on all devices for a much better latency profile from all RAID controllers
3 - The first rumblings of 40Gb Ethernet running on PCIE 3.0 busses
4 - Some late year hints at what lies beyond Thunderbolt

Steve Modica is the Chief Technology Officer at Small Tree (www.small-tree.com) in Oakdale, MN. The company specializes in storage solutions for Mac workflows.