Byte and Switch ran a story recently titled, Responding to Poor Response Time. It is helpful to see the issues surrounding storage performance coming to light. Massive increases in storage capacity, fueled by the amount of data created and reduced drive prices, is actually making the storage performance worse.
This doesn't always make immediate sense, but upon further examination it becomes clear. Disk drives have two basic characteristics, capacity and access time. Capacity is relatively straightforward. Access time is the amount of time it takes for a drive to receive and respond to a request.
While capacity gains have been enormous, access time has not increased significantly in recent years.
Henry Newman at Enterprise Storage Forum wrote another great piece, A Historic Moment for Storage I/O, that outlines this dilemma perfectly. He shows how Fibre Channel disks have increased in capacity by almost 4000x while seek times for disks (access times) have increased by a factor of 3x!
I/O Performance Increases of Different Technologies since 1977
Now, with capacity increasing even further, the problems are only getting worse. If it takes a drive a certain amount of time to scan across a 100GB disk, it will only take longer when it has to scan across a one terabyte disk.
So, keeping up with the cache becomes even more important. But today's storage systems are often fully populated with cache leading to a need for alternate solutions.
Gear6 sees centralized storage caching as a way to alleviate this performance bottleneck, allowing customers to keep their existing storage systems, but add a scalable caching resource in the network to provide a performance gain that matches the incredible capacity gains of the last several years.