Guest Post by Chuck Dufresne, Sr. Technical Marketing Manager, EMC Backup Recovery Systems
A governor, or speed limiter, is a device used to measure and regulate the speed of a machine, such as an engine.
Just as you probably would not appreciate speed limiters on your personal automobile, you don’t want your backup solution to impose limitations on meeting your backup windows. With explosive data growth pushing data volumes into zettabytes and backup windows shrinking with 24 by 7 global data access, something has to give. You cannot afford unexpected bottlenecks slowing down your backup, particularly if you are about to invest in disk based, backup storage. Next-generation backup solutions must be extremely fast, just to keep up with this unparalleled growth in data, but also must be affordable.
So what is the obvious bottleneck that exists when designing a disk-based deduplication backup appliance? Answer… the disks. Reading and writing to disk takes milliseconds, which is forever in the computer world. But what can be done to overcome this overt speed limiter?
Intel co-founder, Gordon Moore’s axiom, often quoted as Moore’s law, states that, “the number of transistors on a chip will double approximately every two years.” Over the past 20 years CPUs have indeed improved in speed by a factor of millions. However, over the same time the speed of hard disk drives has only improved modestly (~ 10x).
The Data Domain engineers at EMC have leveraged Moore’s law to overcome the speed-limit bottleneck of disk I/O performance characteristics. By partnering with Intel, the CPU-centric design of Data Domain systems takes advantage of this continuously accelerating processing power.
The Data Domain Stream-Informed Segment Layout (SISL) scaling architecture delivers high-speed, inline, data deduplication that keeps getting better with every advance in CPU design. Intel continues to push the speed limits of CPU processing power, which is ideal for Data Domain systems because they are not bound by disk spindle speeds. This means that the throughput of a Data Domain system increases as CPU performance increases.
The Data Domain system architecture optimizes deduplication performance by minimizing disk accesses. SISL identifies duplicate data segments in RAM, inline, before storing to disk. It also stores related segments and fingerprints together, so large groups can be read at once. These patented techniques enable Data Domain systems to utilize the full capacity of large SATA disks for data backup and recovery without increasing system memory and the number of disks needed to deliver high throughput performance; overcoming the speed limits of disk.
EMC breaks through the disk speed limit bottleneck and transforms your backup into sizzling fast backup, recovery and disaster recovery.
For more information on Data Domain Stream Informed Segment Layout read, “EMC Data Domain SISL Scaling Architecture – A Detailed Review”.