A Lesson in Flying and Another Reason Why Solid State Disk Matters
August 3rd, 2011 | 
Author: 
HDDs in Arrays are Like Fighter Jets Flying in Formation
For more than 20 years, storage system suppliers have been using software to compensate for the fact that spinning hard disk drives fail. The software, which can be embedded in a hardware device like a RAID controller or running as software on a server, is designed to recover data, when data is lost. And given the challenges of hard disk drive designs, without some sort of protection, data will be lost.
In a hard disk drive the read/write head literally flies on a cushion of air above the surface of the platter. The height and speed of the flight was once compared to flying a fighter jet three inches off the ground. But when disk drives are put into a storage system, where vibration, heat, and other interference can be transferred from one disk drive to another, it’s actually more like trying to fly multiple fighter jets in formation three inches off the ground. And all of that assumes that the storage system is in a clean, temperature-controlled environment, with conditioned power, and limited floor vibration. Using the fighter jet example and with magnetic heads flying over disk drive platters at a height of 3 microns, a single dust particle, which averages 500 microns, would appear over 40 feet high to the fighter jet, and a typical smoke particle would appear over 80 feet tall.
Now imagine you want to build a storage system that can protect the data on your disk drives, in the event that there is a fire in the data center, the building floods, smoke fills the room, the floor shakes from an earthquake or explosion, or the ceiling collapses. Today’s spinning disk drive technology is not designed to survive these types of physical conditions. Fortunately, flash memory and solid state disk, which are much more tolerant of a wide range of adverse conditions, have become much more affordable, and now can be applied to solve some of these complex data-protection challenges. That is the subject of Dr. Alex Winokur’s speech entitled “Flash Forward for Reliable Data Protection and Recovery” at the upcoming Flash Memory Summit in Santa Clara, California on August 9th. We hope that you can attend.
Posted in 
Tags: 
Twitter
YouTube
LinkedIn
Facebook
Using SSDs for disaster protection is a lot less cost effective than duplicating a data center off shore…
Yuval, we agree that data should be duplicated off shore, and the storage for the duplicate data at the off-shore site will be spinning disk for the foreseeable future. But the way companies get their data off-shore is with asynchronous replication, which leaves some data unprotected. The challenge that we address is protecting the rest of the data; the data that isn’t yet at the off-shore site. We keep customers’ most recent data in our black box for the data center, called the Phoenix System. If storage for our Phoenix System was spinning disk, it would not be able to withstand the effects of vibration and violent shock, and it would require too much power and cooling, while we transmitted the data to the remote site. Solid state disk is perfect for this application: protecting a small amount of data through the worst of environmental conditions, while generating very little heat and requiring very little power.
Thanks Liat, that was the explanation I needed. I would also mention less power consumption means reducing the amount of emergency power needed to run your critical data.
That’s correct. We keep the last of your critical data on SSD. While the data is at rest, our system consumes no power, but when we transmit it over the cellular network, we need a small amount of power for the processor, modems, and SSDs within the Phoenix System. Our system has battery power sufficient to power the unit while the data is transmitted. If we used spinning HDDs, the power requirement would be much higher.