Yesterday’s storage picture
The evolution of server technology has led to fast servers and high utilization rates. Yet the evolution of storage has been stagnant. Yesterdays’ storage is in data centers today: simple LUNs with tuning capabilities limited to cache and interfaces. These limitations create a load imbalance of application I/O requirements, resulting in average utilization rates of around 40%. Add server virtualization to the picture, and you have servers running multiple applications with partitioned resources pumping more I/O to the storage systems--causing much more strain on a storage system than the same amount of non-virtualized servers.
For storage systems to keep pace with their server counterparts, they must be able to differentiate services based upon application workload as defined by best practices and administrative settings.
The storage picture of today -- and tomorrow
Pillar Data Systems is the first vendor in the storage industry to provide an application-aware storage system. With the Pillar Axiom, performance is dynamically adjusted for those applications which require it, and your storage utilization is greatly increased due to the ability to more reliably predict performance. Now administrators can predict their utilization and be comfortable that the storage will not be over provisioned.
Even with this added functionality, the Pillar Axiom 500 application-aware storage system drives down acquisition and operating costs by allowing you to:
- Buy less storage. If you could use 40 to 80% more of your storage system, then that’s fewer systems that you have to buy. The Axiom allows you to get differentiated services from multiple LUNs striped across high-capacity spindles, which means the best of both worlds: high performance and high utilization. And when your company growth means you need more storage -- you can scale the Axiom to over 800TB of raw capacity.
- Provision your storage like you provision your servers. Management and maintenance tasks are much easier when you can provision LUNs by application, like you provision your virtualized servers. No more complex forecasting. No more array read/write cache rebalancing to match changes in application workload. And with thin provisioning you will get even greater benefits!
- Rely on dynamic provisioning of resources. Storage silos are a thing of the past. Storage arrays shouldn’t be disposable devices bought for a project and then scrapped at completion or, worse yet, left serving data which is essentially no longer accessed or needed. Once a project is completed, the Axiom dynamically reassigns priority and a new project can be started on the same array. You can even set up a policy to automatically increase the priority of specific applications temporarily as it is needed.
- Dump the dinosaurs and consolidate. How many old-style storage systems are using up power, space, and management resources in your data center? The Axiom can handle multiple disk types and protocols and can provide varying performance levels to your data and applications. With utilization rates as high as 80%, you can consolidate your existing storage on an Axiom.
How do you deliver performance while cutting costs?
Most data centers have multiple tiers of data that need to be accessible at any time, from high-priority Tier 1 data to archival (Tier 4) data. Obviously Tier 1 data needs to be accessed at the highest speeds, while archival data can be accessed at a slower pace. The Quality of Service (QoS) level you set for each tier of data will determine how much you can store on each disk (Disk Utilization), since data stored at the outside edge of the drive is the fastest to access.
To increase performance by reducing seek times, most manufacturers use a technology called short stroking, which writes data only to the outer bands of the disk, where speeds are the highest. While this reduces seek times, it also reduces the amount of data you can store on a disk, since only the outer bands are used.
Pillar uses a new methodology that places the data requiring the highest performance on the outer rings, and the data most tolerant of longer seek times (such as archival data) on the inner rings. This makes better use of each disk, while still keeping performance high. Pillar QoS manages the I/O queues to appropriately deliver the right level of performance to each type of data, simplifying life for IT managers.
The more you can store on each disk, the more you can consolidate your drives (Drive Consolidation), further saving on space while reducing power and cooling costs.
How do you improve capacity without sacrificing performance?
The easiest way to increase data storage performance is to reduce seek times. Most of the industry achieves this by "short stroking" - writing data only to the outer bands of the disk, leaving the inner bands (60 to 70 percent of the disk) empty.
Using proprietary Pillar QoS technology, Pillar segregates data on each disk platter based on accepted seek times. So your primary database files and other high-priority data (your Tier 1 data) is stored on the outer rings where performance is highest. Tier 2 data (records that are between 90 and 180 days old) is stored on the disk closer to the center of the platter. Tier 3 data (records between 180 and 365 days old) is stored further inside the disk, with archival (or Tier 4) data stored at the very center of the drive.
This allows you to reduce the amount of data storage capacity you need by approximately 63 percent. Which also reduces your power, cooling, and space needs.
By utilizing more of each disk, you can improve your Quality of Service and your ability to consolidate your drives (Drive Consolidation).
How do you get more out of a single floor tile?
Most IT professionals put simplification at the top of their key issues. One way to reduce complexity in the data center is to deploy storage platforms that are designed to host multiple services. This allows you to move off of a DAS (Direct-Attached Storage) system of silos, and onto a network-based storage solution that offers increased capacity, scalability, and cost savings.
With Pillar, you can host SAN (Storage-Area Network), NAS (Network-Attached Storage), and WORM (Write-Once, Read-Many) functionality on one Pillar Axiom system. With one user interface, one licensing scheme, one training class, and all in one single floor tile.
For example, you can consolidate your file shares off multiple servers into one NAS Slammer solution. Separate Pillar Bricks within the unit can support SQL and Exchange server files. Backups can be handled on the inner bands of disks, eliminating much of the need for tape backup. And images can be hosted on Axiom drives, eliminating the need for an optical library.
By consolidating your storage onto a single, multifunctional system, you space, power, cooling, and time, which dramatically improves efficiency. And, since IDC calculates that administrators can manage more than 20 TBs more data in a networked environment, this approach will allow your IT department to focus resources on other activities that can deliver more value to the business.
The storage-of-choice in your data center
Its unique application-aware features, low acquisition and operating costs, unparalleled high utilization rate, and ability to scale to over 800TB of raw capacity all combine to make Axiom much, much more than a mere LUN. Whether as the consolidated pooled storage array in your virtualized data center, your main storage in your mission-critical system, or the trusted storage as a key to your business continuance processes -- the Pillar’s application-aware Axiom 500 and 600 is fast becoming the storage of choice in today’s data center.