A single, more flexible approach to storage generally better addresses business needs in the move from traditional IT architecture to a design-for-services paradigm. So says Mark Edwards, director of product and services at Intuate Group, a NetApp certified reseller, who maintains that multiple storage tiering can make it more difficult to achieve the economies of scale so critical to IT success.
“The beauty of tierless storage is that it is able to deliver multiple classes of service (COS) from a single storage infrastructure, a move that cuts the management overhead through a single management environment while also providing greater flexibility in meeting the needs of the design-for-service IT model.”
Significant advantages for the architecture of tierless storage solutions are achieved with the NetApp Data ONTAP operating environment, including a truly unified storage architecture capable of serving all storage protocols while offering unique features for secure multi-tenancy, service automation, data mobility, storage efficiency and all-important data protection.
“Today’s technology has advanced to the point where IT as a service (ITaaS) can be delivered. Whereas IT architects used to design to meet specific requirements, IT infrastructure today is designed to deliver standard services. The advantage of the design-for-service approach is that it is cheaper to use than the old method,” adds Edwards.
“Designing for service uses shared infrastructure and consolidated workloads that increase infrastructure utilisation while realising much greater economies of scale. Instead of having a data centre full of application-based silos, every application uses the same servers, networks and storage, which translates into reduced capital and operating costs.”
Traditional storage tiering no longer makes sense in the design for services paradigm as it stands in the way of achieving the economies of scale that are key to IT success and fails to meet today’s key business requirements.
Edwards maintains that a single tier of storage, offering different classes of service, meets modern needs much more effectively.
“The most obvious limitation of tiered storage is the difficulty of categorising data to establish in which tier it belongs. The market has no consensus on this and no foolproof computer algorithms have emerged for its automation, with most software tools relying on a few attributes, such as time of last access and file ownership. In short, to establish which data belongs in which tier is both laborious and error-prone while being dependent upon end-user assessments of data needs.”
A Gartner study established that a third of storage implementations were less than 50TB, another third were greater than 350TB and the rest somewhere in between. However, Edwards points out that the real surprise finding was that the total capital and operating cost per TB was three times higher for small storage installations compared to large installations.
“The conclusion is that by breaking storage up into multiple tiers, companies may incur greater cost per TB if, for example, they operate four small to medium installations as opposed to a single large one. Another interesting statistic is that labour is frequently the biggest component of the storage budget, often greater than the storage hardware and software combined.”
Edwards says this indicates that the way to control storage costs is to reduce the labour costs in the overall storage budget, through streamlining storage management. An obvious way to achieve that is by simplifying or eliminating storage tiers.
“Storage system availability is also important as sometimes the cost of business losses when IT infrastructure is not available includes lost productivity and missed opportunities, such as an order fulfilment system going down. These impact on the bottom line. It is important to consider that because tiered storage is designed to deliver different service levels in different tiers, its availability and resilience as a whole is reduced.
“On the other hand, tierless storage delivering multiple classes of service from a single infrastructure reduces the management overhead through a single management environment while at the same time providing greater flexibility in order to meet the needs of the design-for-service IT model.”
He points out that NetApp storage is well-suited for the architecture of tierless storage solutions that deliver more value using less physical storage, as tierless storage with NetApp can be adjusted to deliver different classes of service: from a single media type to cater for priority of service, intelligent caching and aggregate size, which places an upper boundary on the I/O performance of any given volume within the aggregate. Volumes contained in an aggregate with four disks will offer relatively low performance while volumes in an aggregate with 50 disks will offer high performance.
Edwards adds that most service providers and IT organisations require only three or four performance levels. “While it may be tempting to describe dozens of virtual tiers, in practice tierless storage allows for infrastructure usage to meet the needs of most workloads on demand. Delivery of this level of flexibility is the whole point of a tierless approach.”
Other advantages include determining scale performance and capacity on demand from a single platform and secure multi-tenancy for cost-effective and secure partition of a single system in support of multiple tenants, including applications, customers, workgroups and security zones.
NetApp also allows more gigabytes to be put in service with less physical storage with a specific set of storage efficiency technologies, thereby delivering more with less.