What is Software-Defined Storage (SDS)

1. http://www.webopedia.com/TERM/S/software-defined_storage_sds.html

Storage infrastructure that is managed and automated by intelligent software as opposed to by the storage hardware itself. In this way, the pooled storage infrastructure resources in a software-defined storage (SDS) environment can be automatically and efficiently allocated to match the application needs of an enterprise.

Separating the Storage Hardware from the Software

By separating the storage hardware from the software that manages the storage infrastructure, software-defined storage enables enterprises to purchase heterogeneous storage hardware without having to worry as much about issues such as interoperability, under- or over-utilization of specific storage resources, and manual oversight of storage resources.

The software that enables a software-defined storage environment can provide functionality such as deduplication, replication, thin provisioning, snapshots and other backup and restore capabilities across a wide range of server hardware components. The key benefits of software-defined storage over traditional storage are increased flexibility, automated management and cost efficiency.

Software-Defined Storage is Not Storage Virtualization

Software-defined storage is sometimes confused with the term storage virtualization, but as an article from CRN explains, while the latter term involves separating capacity from specific storage hardware resources (and thereby pooling storage devices), SDS involves separating the storage capabilities and services from the storage hardware.

Prominent examples of software-defined-storage include OpenStack, EMC ViPR, Nexenta and HP StoreVirtual.

===============================

http://techpageone.dell.com/technology/software-defined-storage-getting-started/

For a while, it seemed software-defined storage (SDS) might be more marketing hype than reality. But that’s changing as vendors invest in the technologies that enable the abstraction of storage services from the hardware.

SDS transcends storage virtualization by creating a unified pool of hardware resources and adding automation and monitoring tools. Most SDS configurations use commercial-off-the-shelf (COTS) server engines to run their storage software stacks on virtual machine instances.

The benefits of this approach lie in the ability to move functions out of the storage appliance and place them close to host data. SDS enables better load balancing as well as the on-demand spawning of new instances to handle surges in workload. Management tools orchestrate the instances automatically, reducing operational task load and improving responsiveness and flexibility. The storage farm appears to be a single pool of storage and can absorb legacy storage area network (SAN) and network attached storage (NAS) gear as well as new equipment.

Over time, SDS will make commodity storage boxes an alternative to large RAID arrays. Solid-state disk and bulk SATA hard drives are designed to operate in relatively low-tech enclosures, free from the concerns of rotational vibration seen with fast enterprise drives. Expect to see inexpensive JBOD for storage, with RAID functionality only in the SDS hosts if it at all.

With SDS, unified storage — in which the storage pool is virtualized and presented as a combination of object, NAS and block I/O storage — will become a storage system feature, along with compression, deduplication, tiering/caching, replication, backup, and archiving. Better SDS packages will extend connectivity to private or public clouds, and provide access control and data movement tools.

Because SDS is still evolving, don’t expect to see the entire story from any one vendor. Vendors will continue to work on virtual storage area network and virtual machine instance storage interoperability, especially in the box management area, with older gear being slow to interface with the management tools.

Another issue lies in performance and stability analysis. Especially with cloud deployments, storage bottlenecks and less-than-optimal data flows can slow down processing and lead to additional storage and transfer costs.

Rolling out SDS

Despite these caveats, SDS is clearly the future of storage. These tips will help you start experimenting. 

• First, try building a new storage pool (it doesn’t have to be large or mission critical) on low-cost hardware as a sandbox effort.
• Pick an SDS vendor with a solid product and a road map to cover any missing features. It helps to have a good idea of storage evolution in your data center, not only in terms of capacity, but also of any legacy gear that’s likely to be moved into the pool over time to ensure that sockets are available to connect the SDS code to them.
• Pay attention to external interfaces. SDS has to co-exist with your favorite virtualization system. It also has to work with a set of external constraints that might not sit within its direct management control. Examples include importing or exporting data to non-managed storage, interfacing a cloud storage complex, or managing efficient tape backup operation. Tape is a potential problem, because backup software may restore duplicates and uncompress files before compressing them again for writing to tape. This is a huge extra burden on backup windows, so ask SDS vendors if they offer a solution. 
• Keep it simple. Although it may be tempting to pick software that has controls for everything, remember that the SDS package should do most of the work so IT staff can focus on fine tuning and managing data flow.

A software-defined approach can effectively multiply the size of installed storage, and the savings from avoiding new storage purchases offset implementation costs. SDS offers a great value to complex cloud and campus deployments. Don’t be left behind.
===============================

Right now there's no standard definition, but it involves separating storage features from the storage hardware itself. IDC defines software-based storage as "any storage software stack that can be installed on any commodity resources (x86 hardware, hypervisors, or cloud) and/or off-the-shelf computing hardware, and used to offer a full suite of storage services and federation between the underlying persistent data placement resources to enable data mobility of its tenants between these resources."

So SDS enables companies to buy heterogeneous storage hardware—choosing whatever is best value (in terms of price, reliability or any other criterion) at any given time—and then buy and use the storage software that provides the features they need. "SDS means separating the hardware buying decision from the software buying decision," Peters explains.

Likely SDS Vendors

There is no shortage of hardware vendors, but the obvious question is where will the storage software come from? Laura DuBois, a storage expert at IDC, predicts it will come from three key camps:

1. Open source development—this includes Hadoop with HDFS, OpenZFS, Openstack Swift and storage-specific solutions like Ceph.
2. Storage ISVs - Companies like Inktank, Nexenta, Red Hat or Symantec
3. Existing storage vendors that have traditionally delivered storage systems and appliances. "Basically, any storage software stack that could run as independent software on industry standard hardware," says DuBois.

If you look at those three camps, you'll see that they are very similar to the groups that have appeared to provide software-defined networking solutions. There are open source projects like Nox and Floodlight that have produced software-defined networking controllers, and there are commercial vendors like Big Switch Networks and Nicira (now part of VMware.) But perhaps most significantly, the big networking hardware vendors like Cisco and Juniper are all getting involved in the software-defined networking space as well. They have to—if they don't, they may become irrelevant.