Data Orchestration Makes Migrations Obsolete
Posted in tech
Typically an infrequent event, data migration happens because storage is at its end of life, current storage capacity or performance is inadequate, or because IT is moving datacenters or migrating apps to another system or location. In contrast, Smart Data Mobility orchestrates data placement, monitoring and moving it across all storage types to meet evolving business needs as defined by IT objectives. This is made by possible by data virtualization abstracting physical data locations within a global DataSphere. Let’s consider some of the differences between the two approaches.
Data Migrations are Manual and Extensively Planned; Data Orchestration is Automated in Real-Time
Data migrations require lengthy planning, purchasing and implementation cycles. Since applications reference a physical storage system, they often need to be reconfigured after data is moved so that data can be accessed from the new location. This often results in application downtime. This downtime can be critical for many organizations, such as banks, which can manage so many different applications that even locating where data is stored can be a challenge. Organizations like these plan for migrations up to a year in advance.
Data Orchestration requires the virtualization of data, abstracting the data’s physical location from the logical view presented to applications. This enables data to be moved automatically and in real-time without disrupting applications to proactively address evolving business needs.
Bottoms-Up Data Migration Versus Application-Aware Data Mobility & Orchestration
IT typically provisions entire LUNs or volumes of application data to static storage systems that IT has identified as the closest fit for an application’s most stringent requirements—in other words, the most expensive storage the application might need. Since data is difficult to move, IT and application owners commonly purchase double or more the expected lifetime capacity needs, which results in inefficient storage utilization. Delivering the many storage capabilities required by today’s many diverse enterprise applications results in a complex proliferation of storage systems, which can become very difficult to maintain as IT staff move on from the company. This inefficiency and complexity increases datacenter costs and reduces ROI.
Data Orchestration allows enterprise IT to see and manage data according to how applications are accessing it, at a granular level (app, directory, folder or file). IT sets objectives that dynamically place data across storage types based on performance (IOPS, bandwidth, latency), protection (availability, durability, security), and price needs. DataSphere automatically and non-disruptively moves data to the ideal storage resource to meet application needs, maximizing service levels and cost efficiency.
Idle Capacity on Expensive Storage Increases Costs; Mobility Extends Storage System Strengths to All Data
To avoid moving data, enterprise IT typically chooses a single storage system that can meet the needs of an application’s data over its projected lifetime. This creates storage silos, which become a big source of waste in the datacenter for a number of reasons:
· Capacity is overprovisioned for each app, as enterprises commonly purchase double or more than the projected capacity required so data will not have to be moved.
· Inefficient use of storage resources, as around 80% of each app’s data is better suited to more cost-efficient storage
· Increased complexity as IT must maintain multiple diverse and rapidly growing storage platforms separately.
The Primary Data DataSphere platform unites storage silos, giving IT granular control over data, so they no longer have to choose a single “best fit” storage system for an application. Instead, IT can assign objectives to data and allow intelligent Data Orchestration to automatically move it to the ideal storage for its needs. Temp and log files might reside on flash in the application server; active data files for a mission-critical app might reside on a flash array, while the cooler data for the same app might reside on a filer that can make data quickly available should it get hot; data that has been idle for a specified time period—say, 30 days—might be automatically moved to the cloud.
With the intelligent Orchestration that DataSphere delivers, enterprises can save significantly by:
· Extending existing infrastructure life by improving utilization to free as much as 66% of currently idle capacity
· Enabling scale out economics for apps that currently require scale up storage for performance or access to data services
· Moving data automatically and intelligently to the most cost-effective storage that meets applications’ business needs
· Reducing the work of managing data across multiple storage silos
The intelligent Orchestration DataSphere provides also improves service levels by making all storage systems’ strengths available to all data by:
· Non-disruptively optimizing data placement, in real-time, as application demands change as defined by IT objectives
· Enabling data to be managed centrally and shared globally across all storage resources, including in-server flash and the cloud
· Intelligently moving data across all storage resources, in response to actual end application performance, instead of blindly tiering data within a single storage array
The pain of data migration today is a symptom of storage silos. Data Orchestration unites storage silos within a global dataspace so all data can efficiently leverage the diversity of storage capabilities. This enables IT to proactively respond to evolving business needs, automating data placement by objectives to improve service levels, increasing storage utilization to reduce costs, and turning painful data migrations into a simple, non-disruptive ongoing activity performed by intelligent, data-aware software.
 According to Wikibon, 20% of data receives 80% of the access.Five Tier Storage Model, Wikibon, Jun 2012.