Solid State Disks are the right storage for ClustrixDB. What’s not to love about low latency and high throughput?
You have lots of choices for data storage these days. The most familiar data storage device is the spinning or hard disk drive (HDD): spinning disks are very affordable and can store huge amounts of data. Another option is Solid State Drives (SSD) which are incredibly fast and durable. And Storage Array Network (SAN) products can provide vast amounts of storage to be shared amongst all of your enterprise applications. Each data storage technology has its good points and bad.
However, only one of these is the best solution when you desire high performance from your relational database. Relational databases like ClustrixDB work best with SSDs because they provide near-instantaneous access to the information in storage and have significantly higher rates of data throughput than spinning disks. SSDs can be 100 times faster than HDDs when used with relational databases like ClustrixDB.
ClustrixDB, unlike applications like streaming media, accesses random (usually small) pieces of information from storage very frequently. In human terms, ClustrixDB might need to go to storage and get a zip code. And then it might need to go to a different storage location and get a product ID. Seek Time is the amount of time it takes a hard drive’s read/write head to find the physical location of a piece of data on the disk. Latency is the average time for the sector being accessed to rotate into position under a head, after a completed seek. Both metrics are significantly higher on a spinning disk than a solid state disk. Thus when your relational database requires 50,000 or 100,000 (or millions) of IO requests per second, this small interval of time built into each IO will add up and impact your database performance significantly.
A third data storage option is SAN: While it’s true that SANs can be configured with SSDs and their latency can be very low– SANs are usually a shared resource (several enterprise applications share the device at the same time) and have typically higher latency due to their need to move the data over the network. But even the fastest SAN can quickly degrade the performance of ClustrixDB when it is being shared with another application that is acting as a ‘noisy neighbor’… which can be a common occurrence in the cloud.
Dedicated Solid State Drives are the best storage medium for relational databases like ClustrixDB. They offer the best performance, consume the least power, produce the least heat and contain no mechanical parts.
Please note that all SSDs are not created equal. Clustrix recommends using SSD products that are designed for enterprise use and have power loss protection.