Software-defined storage (SDS) is storage software that runs on ordinary servers and joins their drives into one shared storage system, called an array. The company buys the software and the servers separately. It does not buy a sealed storage box with the software already inside.
A traditional enterprise storage system is sold as one product. The drives, the controllers and the software all come together from one maker. The customer must then buy that maker's own drives and upgrades. Software-defined storage separates the software from the hardware. The same split is happening across the technology industry, and the page on hardware-software disaggregation explains why.
How does it work? The software is installed on several standard servers. Each server holds its own hard disk drives and flash drives (SSDs). The software joins all these drives into one pool of space. Applications see one large disk, or one shared folder. The software decides on which server each piece of data is written. It writes every piece to at least two servers. So if one server fails, the data is still available on another. This is redundancy. The software also keeps the most-used data on fast flash and moves the rest to cheaper hard disks. This is called tiering. Inside each SSD, an SSD controller spreads the writing evenly, so that no flash cell wears out early.
Take, for example, a district cooperative bank in Kolhapur with 60 branches. Its storage array is eight years old, and the maker has stopped supporting it. A new sealed array would cost about ₹1.2 crore. Instead, the IT team buys four standard servers with 40 TB of drives in all, and a software-defined storage licence. The bank's core banking database and the branches' shared folders move to the new pool. Two years later, the bank needs more space. The team adds a fifth server. The software spreads the existing data across all five servers while the branches keep working. When a server grows old, the team removes it in the same way, and the software first moves its data elsewhere.
Who chooses software-defined storage? Mostly companies that want to choose their own hardware, avoid being tied to one maker, or reuse servers they already own. The trade-off is testing. The company's own team must check that each server and drive works well with the software. With a sealed array, the maker has already done that testing.
The storage still needs a backup. A second copy goes to backup and recovery software. Redundancy protects against a failed drive, but not against a deleted file or an attack. Hyperconverged infrastructure takes the same idea one step further: the same servers also run the applications.