Objective-Defined Storage is software technology that abstracts storage hardware from the data. It is a new market segment that combines the best of modern storage techniques to provide a single, ubiquitous, software-only platform for optimizing data mobility.
The entire focus of Objective-Defined Storage is to transition from managing storage to enabling storage. This is achieved by removing the complexity from storage management and allowing objectives to be created and rolled out across multiple applications. There are three broad categories for the objectives – they are: performance, capacity and resilience objectives for an application, and once they are set, the storage provisioning is fully automated.
Objective-Defined Storage discovers all your storage devices, profiles them for latency, capacity, and IOPS. It then creates a single storage pool / fabric. You set your objectives, and the artificial intelligence takes over to create and deploy policies. The policies adapt to achieve the objectives that have been set, based upon the user’s specific hardware. If a problem or failure is encountered, the Artificial Intelligence technology self-heals the storage fabric without user intervention.
Three Main Objectives
Performance Objective: The objective-defined storage product evaluates performance based on a number of qualifiers, including available IOPS, latency, and distance between nodes. Based on the objectives that are set, the product will move data to the most suitable storage devices to deliver that performance objective by leveraging the latest in micro-tiering technology.
As an example, if you set the latency objectives at say 150ms, the data may reside on devices that meet that criteria, e.g., the cloud. If the latency objective is changed to 25ms, the data is moved to devices (e.g. hard disk) that are equal to, or faster than, 25ms. If you set the latency to 10us, data is moved to devices faster than 10us, for example, Flash.
IOPS follows a similar process. If your storage is incapable of meeting the objectives that you have requested, you are informed so you can deploy more performance (e.g. Flash or DRAM), attach higher capacity (commodity) devices to reduce the load, or reduce your objectives. All devices can be shared across all applications, which creates maximum flexibility and scalability.
Further, real-time analysis is constantly occurring and looking at the ‘least recently used’ (LRU) list of data on each device. Data that is “hot or active” is moved to the lowest latency storage with highest IOPS to meet the objectives set. Warm data is moved slightly down – again, depending upon your objectives and available storage. Aged, stale, or cold data, which you probably will never access again, moves to the slowest (often also the cheapest) storage, for example, the cloud or commodity (cheap) high capacity disk drives. It is still accessible, available, and searchable, but doesn’t need to reside on expensive or fast disk – typically freeing up between 50% to 80% of expensive disk. If at any stage in the future that warm or aged data is required, it moves back up to the hot data tier, and moves back down over time depending upon the objectives set and your currently available storage devices.
Resilience Objective. Data resilience is critical and delivered through live data instances. Live data instances are resilient copies of your live or active data. As a default, you must have two instances, however, for critical applications, you can have up to four concurrent live data instances.
The Artificial Intelligence built into Objective-Defined Storage solutions automatically ensures that multiple instances of all data is on multiple devices and all are up to date. If a device fails, another device that holds that data, and that is best able to deliver the data according to the objectives that have been set, serves up that data to the application without downtime or failures. If the device is unable to achieve one or more of the objectives, for example latency, the data is automatically moved to a faster device, closer to the host to reduce the latency.
As this has now become the primary live data, another live data instance is automatically created to ensure that the number of live data instances is maintained to ensure that resilience objectives are being achieved. In old technology, this is comparable to clusters and can be local, or even in remote geographic locations. Of course, remote Geo’s will have greater latency than local live data.
This can be used in conjunction with RAID technologies, or as a replacement to RAID. If used in conjunction, it creates multiple extra layers of resilience. For example, if a number of disks in a RAID configuration fail, the RAID can fail resulting in extensive downtime, time consuming and expensive restore operations. If Objective-Defined Storage is used as well as RAID, in that exact same situation above, another device with the live data would instantly be available, serving up the data, eliminating downtime. The second live resilient instance can also be RAID, but it doesn’t need to be. Objective-Defined Storage is all about choice and decoupling the data from the storage hardware / vendor.
Capacity Objective. In many companies, typically 50% to 80% of all data on a SAN or NAS is aged or stale – it still needs to be accessible, and it doesn’t need to reside on expensive SAN, NAS, or SSD/Flash arrays. By using data-tiering as described above in the performance objective section, moving this aged, cold, or stale data to the least expensive, highest capacity commodity storage, or even the cloud, you are immediately able to free up substantial capacity on expensive SAN and NAS appliances. An ancillary benefit of this is that Objective-Defined Storage extends the life of your existing SAN or NAS products by freeing up expensive disk. All the data is still searchable and recoverable regardless of location.