Software Defined Storage
Software Defined Storage (SDS) has developed in response to exponential growth in global data volumes, ever-changing business requirements, and constant cost pressures. It is an evolving concept, which refers to storage infrastructure solutions that are managed and automated by intelligent software rather than traditional underlying hardware. The disaggregation of hardware and software provides freedom from vendor lock-in, reduced infrastructure costs and more scalable solutions, all without sacrificing service or resilience. SDS provides a vendor agnostic, multi-protocol platform, meaning that it has the capability to manage different types of data from multiple applications, simultaneously.
The limitations of traditional ‘scale-up’ storage solutions
Traditional storage architecture is not equipped to provide flexible, cost effective solutions for today’s growing data volumes and usage. In this scenario, as a system reaches capacity the solution is to purchase additional drive shelves or a larger model, which can only be bought from the incumbent vendor. Considering on average, data is growing at a rate of 10x the budget for managing it, this solution is expensive and ultimately unsustainable.
One of the challenges facing the user is the requirement to accurately forecast data growth over the lifecycle of their storage device. The preferred system must be able to cope with the amount of current and expected data storage. This requires a large upfront investment and relies on an accurate forecast of data growth. Get this wrong and the user faces the expense and risk of migrating the data to a larger model or even acquiring another device. The addition of these larger models and systems creates ‘storage silos’ which create additional problems in terms of analytics and management.
What’s different about Software Defined Storage?
A Software Defined Storage platform consists of a number of storage “nodes” which grouped together form a cluster. Data is distributed across the nodes and is managed by intelligence sets within the host software application to determine the number of copies of data and the location of such copies. Each node contains compute and storage capabilities and is built using a standard X86 server.
Resilience is provided by creating multiple copies of each piece of data and locating them in separate nodes. Should a drive or even an entire node fail, the data is automatically rebuilt from the other copies held in the cluster, it is ‘self-healing’. The cluster does not have to reside in a single location; some applications will also support geographic distribution, which is ideal for DR and data sovereignty compliance.
Software Defined Storage is designed to ‘scale-out’. When additional capacity is required, another node is added to the cluster and the application will then redistribute data across all of the available nodes in accordance with the user defined rules of the system.
An elastic platform
As additional nodes are added the compute layer scales in line with capacity which means that the platform can grow without any deterioration in performance. In fact, the bigger the cluster, the better the performance as the overall compute resource is larger which results in faster rebuild times and greater resilience.
A note on performance
The performance of an SDS solution is delivered by aggregating the performance of all drives within the cluster into one single virtual pool of capacity. Additional performance can be delivered by tiering and the use of memory and flash media as a caching layer to receive, write and authenticate the movement of data onto cheaper, high capacity spindle based drives. This highly efficient use of SDS is far more cost effective than using an all flash array to provide performance.
Benefits of Software Defined Storage
In short SDS is cheaper and more flexible. The benefit is not only in the price tag of the application, it extends into the management of the solution, access to data and future proofing the storage strategy of your business.
- Scale seamlessly without capacity limitations reducing upfront CAPEX.
- Reduce risk of data loss through distributed replication.
- Manage multiple protocols and consolidate storage silo’s with a single platform.
- Eliminate storage downtime with a “self-healing” cluster.
- Provide effective DR solutions through geo distribution and replication.
- Break free from vendor lock-in and leverage low cost commodity X86 hardware, without sacrificing enterprise features such as data de-duplication and compression, snapshots and clones, QoS (Quality of Service) and encryption.
Choosing the right solution
Whilst the disaggregation of hardware and software provides transparency, which ultimately reduces cost, it does also remove an element of convenience for the end user. Vesper Technologies offers the assurance of a single point of purchase for both hardware and software without sacrificing any of the freedom associated with a software defined platform. We are partnered with a variety of Software Defined Storage vendors and are well placed to help customers evaluate the options and select the most appropriate solution for their specific environment.
We also understand the X86 platform and its importance as a successful SDS solution is not only about choosing the right application – it is imperative that the hardware is right. This means selecting the right server that will support the right type of CPU, memory, network connection and storage (flash or spindle) in the right ratios to deliver the required performance. Our detailed understanding of the commodity market enables us to find the best cost-performance ratios when architecting each specific solution.