For the Cloud Service Provider, illness hospital storage decisions are not merely technical: they are ultimately economic, affecting the cloud service providers’ revenue, cost structure, and profitability.
The transition from inside-the-firewall services to cloud-based deployments caught the entire storage industry off-guard, and most commercially-available storage options, for the Cloud Service Provider (CSP), are neither designed for cloud economics, nor suitable for the range of value-added higher-margin CSP applications, from high-performance application acceleration to high-throughput mixed-mode virtualisation support.
Any storage vendor will tell you they’re cloud-capable, cloud-ready, or built-for-the-cloud. Given the wide variety of architectural approaches and even wider variety of technical choices made by these vendors, that begs the question: what does cloud-enabling storage really look like?
At X-IO, we think there are three primary dimensions to the cloud storage decision:
- Price / Performance
- Configuration and Change Management
- Total Cost of Ownership
Price / Performance is usually disguised, by storage vendors, as the cost of a single technical dimension: IOPS/$, capacity/$, throughput/$, or latency/$.
However, any well-run CSP defines Price / Performance based on the market the CSP serves, and the services it providers. Any CSP focused on collaborative work, or on multi-tenant virtualized (or dedicated, or tiered) IaaS, PaaS or SaaS, will focus on a balance of capacity/$, IOPS/$ and throughput/$, because that’s the tricky balance those kinds of CSPs have to strike to compete effectively in their markets.
Configuration and Change Management
Configuration and Change Management (CCM) models implicit in storage vendors’ architectures are also something well-run CSPs spend a lot of time considering when they make architectural storage decisions.
CSPs are well aware that frame-based, controller-oriented storage architectures from old-guard enterprise storage companies are fundamental incompatible with the software-centric, software-defined architectural model of the cloud: a model in which the core elements of the software stack (hypervisors, operating systems and DBMSs) control their storage. And, in the world of the CSP, there are at least two other dimensions to CCM:
- flow-through provisioning
- autonomic contro
In most CSPs, self-provisioning (as a mechanism for reducing operating costs) means flow-through provisioning. Changes instigated by a client systems administrator have to ripple through the entirety of the CSP’s architecture and infrastructure, resulting in changes all the way down the stack, from applications down to the system hardware and network infrastructure.
This is the fundamental idea behind cloud software architectures like OpenStack: standard and highly automated ways of provisioning, reprovisioning, monitoring and controlling the complex technology stack are essential to CSP profitability. Armies of network operations center personnel and system administrators, nursing complex, brittle stacks are inimical to profitability. This flow-through complexity is exacerbated by the customers’ increasing demand for performance-oriented service level agreements (SLAs).
Under performance-based SLAs, not only do CSPs have to provision, and reprovision every component of their architecture in a seamless fashion, but they also have to monitor every component in real-time, and be able to relate the performance of every component to customer SLAs.
Total Cost of Ownership
Ultimately, Total Cost of Ownership (TCO), it turns out, is the most important dimension of storage selection, as far as a CSP’s cost structure, profitability and EBITDA are concerned.
Storage is the single most expensive part of a typical CSP’s technology infrastructure, and a three- to six-year depreciation schedule for CSP storage means that 40% to 70% of the real cost of a storage architecture decision hits the books as second-order operating expenses for power, cooling, space and personnel. While some storage options may appear attractive on the basis of their first-order acquisition costs, most traditional storage offerings are power-hungry heat engines that consume acres of rack space, at scale, and require a small army of expensive administrators to nurse controllers, update firmware, and swap and shred drives — one quarter of which can be expected to fail during a six-year duty cycle.
X-IO’s Intelligent Storage Elements are not “designed for the cloud” they are “ideal for the cloud.”
Based on a $100M Seagate investment that began in the early 1990s, with more than 10 years of deep and systematic R&D by the world’s leading drive manufacturer, the Intelligent Storage Element has been available in the enterprise storage market for more than five years, with thousand of units in operation. Read Case Study.
Our ISE product line, which scales from less than 20 TB to hundreds of petabytes, one 3U unit at a time, with between 50,000 and several million IOPS, and up to 2 GB/second of throughput per ISE, is expressly designed for use in high-performance, variable-demand, virtualization and DBMS environments, where balanced performance and scalability are crucial.
Using our patented Continuous Adaptive Data Placement (CADP) algorithm, our HyperISE products seamlessly meld solid-state drives and conventional hard disk drives into high performance storage blocks that:
- obviate the need for manual or auto-tiering technologies through traffic-based self-tuning
- scale linearly to multiple petabytes and are designed to be managed directly by the storage management features of hypervisors, operating systems and DBMS.
Inside each ISE, our managed reliability controller (MRC) boards that operate in tandem to control and monitor, continually and in real-time, the performance and health of every drive in each array, ensuring that our arrays provide the same or better performance when full as they do when initially provisioned, something no other storage vendor in the industry can offer. This capability alone makes ISEs a superior alternative for CSPs, when compared to conventional enterprise arrays and hybrid storage vendors, which experience performance degradation as storage is consumed — often making 25% to 40% of purchased storage unusable. Drives do not need be replaced, ever, by anyone.
Our patented remanufacturing capability means that ISEs fix drives, rather than failing them. Our MRCs identify, test, remanufacture and return to service any drive that exhibits reliability problems — with no human intervention.
Every ISE exposes, through a set of RESTful APIs, every piece of its operating telemetry, and every CCM function of that ISE, to any RESTful management backplane or stack-oriented cloud architecture, including CloudStack and OpenStack. And the ISEs are self-aware, and autonomically aggregated — to discover every operational ISE in a given multi-petabyte storage pool, a third-party REST-based application need only collaborate with one ISE, which can in turn make that REST-based application aware of every other ISE in the pool.
ISEs are the most energy-efficient, cool-running, rack-dense storage arrays in the market.
Our dual redundant power supplies draw a mere 700 watts, and power up to 48 TB of raw storage within a 3U enclosure that displays a front-to-back temperature differential of less than 15 degrees Centigrade, with very low ambient heat dissipation.
That ultra-green energy profile, combined with our no-touch in-situ remanufacturing and our five-year-no-cost warrantee, means that the ISE’s five-year total cost of ownership is the best in the large-scale storage industry, period.
Our customers typically enjoy a 30-50% reduction in their TCO, as compared to conventional enterprise storage, on top of the 40% improvement in storage utilization. In a cloud service provision environment, that differential can amount to millions of dollars of cost savings and EBIT improvement, each and every year of operation. That kind of value proposition — superior price/performance, superior TCO, and a superior operational profile for cloud deployment — raises a lot of eyebrows when we talk about it.
However, don’t just take our word for any of this. If you’re a cloud service provider, a managed service provider, or a colocation facility looking to transform the economics of your business, improve your customers’ operating experience, and reduce your operating cost basis, we’ll provision you with ISEs, and you can pay-as-you-go, using ISEs in your production environment at no cost to you until you’ve sold, and consumed, 50% of their capacity.