Intelligent Hardware

The ISE:  Hardware overview

ISEs, as mentioned in the previous section, provide a carrier-grade scale-out high-performance storage infrastructure, with a five- to seven-year operating lifespan, and provide 2 to 10 times the performance of alternatives, at 30-70% of the total cost of ownership of those alternatives.

The ISE achieves that kind of scale and performance by focusing on the effective interoperation of hardware and software, in concert with one another. The basis of the ISE performance and reliability lies in its proprietary, patented hardware design.

The ISE 3 rack-unit (3U) chassis is designed to shield the storage media from environmental anomalies, and to eliminate adverse environmental factors that are introduced by the storage media themselves. Sensors monitor all aspects of the chassis’ environment, and custom-engineered cooling systems pull (rather than push) air from the cold aisle through the unit. This produces a unique operating profile: the ISE is whisper-quiet (relative to conventionally-integrated storage arrays) and the temperature differential, front to back, is typically less than 8 degrees Celsius: far cooler than conventional storage arrays.

Managed reliability controllers

The Managed Reliability Controllers (MRCs) are based on current-generation, multicore, Intel low-power processors, and include hardware RAID cores. They run in precisely-synchronized tandem, as a truly active-active pair, during ISE operation. All data within the ISE is available to either controller at any time for reading or writing. This synchronized tandem operation means that the loss of one MRC during operation will have no appreciable effect on the performance or capacity of the ISE. It enables MRCs to be hot-upgraded, in operation, without taking the ISE offline. Each MRC:

  • Provides the storage fabric connectivity for the ISE—up to four 8 Gbps Fibre Channel connections per MRC (for a maximum of 64 Gbps of connectivity per ISE)—and two 10/40 Gbps Ethernet connections per MRC for iSCSI (for a maximum of 160 Gbps per ISE).
  • Manages the ISE DataPacs and their constituent media, while using a proprietary on-controller software stack (discussed in detail in the next section) that (a) joins the DataPacs together into a single logical pool, (b) spreads data across the entirety of the constituent media in the DataPacs for optimal performance and data protection, while (c) retaining a portion of each DataPac and each disk as online spare capacity. MRCs also manage the staging and online installation of media-level firmware updates.
  • Monitors the operation of each storage medium within each DataPac, as well as the telemetry from the chassis’ sensor arrays, at the sub-second level. They also collect operational telemetry for reporting and detect environmental conditions, soft and hard faults, and operational anomalies that indicate failing media, requiring the attention of the ISE in-situ remanufacturing services.
  • When required, takes portions of single media offline (without disruption or loss of data access) for in-situ remanufacturing and restoration (discussed in detail in the next section), without disrupting operation, and while preserving the rated capacity and performance of the ISE.
  • Hosts the ISE CorteX REST web services, integrating each ISE into the CorteX service pool and allowing all ISEs to be interrogated via a single connection to any one ISE within the service pool.
  • Transmits summary and detailed (near-real-time) operating telemetry to X-IO for customer service purposes.
Download The Intelligent Storage Element - An Introduction

The DataPacs

The storage media themselves—whether all enterprise-grade SAS hard disk drives (HDDs) or a mix of hard disk drives and enterprise-grade solid-state drives (SSDs)—are fitted with custom X-IO firmware that provides the ISE with low-level control over the internal components of each media device.

These media are mounted in custom, sensor-laden carriers called DataPacs, which:

  • Eliminate the effects of vibration introduced by the media’s electromechanical operations
  • Pack drives closely together for density improvements
  • Organize the drives for optimal airflow and cooling
  • Enable low-level access to each media device’s power and data lines, for low-level internal control of the media device.
Download The Intelligent Storage Element – An Introduction

DataPacs alone improve the reliability of storage media by 2 to 3 orders of magnitude.

DataPacs act, within the ISE, as multi-spindle “superdrives” that the ISE can treat as a single storage unit, while relying on the DataPacs to manage the components of that storage unit—the individual media—as granular units of storage within the superdrive. ISEs use 100% of the performance and capacity of each storage medium within the DataPac in unique and sophisticated ways, to contribute to the performance and capacity of the superdrive, as required by the performance and capacity rating of the ISE itself. Currently, ISE DataPacs are configured either in all-HDD models or in hybrid HDD/SSD models, and they provide between 6 to 48 TB of raw capacity for the ISE.

The DataPacs and their constituent media are managed by the dual, active-active managed reliability controllers (MRCs), located in the rear center of each ISE.

Power and cooling

Redundant power and cooling units, based upon a proprietary design featuring aeronautic-grade blower designs, sit in the rear of each ISE, consuming only 700 watts of power at maximum load, and shedding a similarly small heat load, while powering and cooling the chassis and all of its components.

Download The Intelligent Storage Element - An Introduction

Supercapacitors

Protecting data that is taken into cache but not yet written to the DataPacs is an essential feature in any well-designed storage array. In traditional storage arrays, lead-acid batteries are used to supply enough power to move cached data to disk. Over time, these batteries lose their capacity to hold a charge and must be replaced, which often leads to array downtime.

Instead of using batteries, the ISE uses supercapacitor arrays, located in the front of the unit. These supercapacitor arrays store sufficient power to guarantee orderly shutdown of the unit in the event of a power loss—with no cache corruption or data loss. Supercapacitors do not harm the environment, and are not susceptible to degradation like lead-acid batteries, and will last for the life of the ISE without the need for replacement.

Download The Intelligent Storage Element - An Introduction
© 2016 Xiotech Corporation. All right reserved. | X-IO Headquarters | 2375 Telstar Suite 150 | Colorado Springs, CO 80920 | Tel: 719.388.5500
Back to Top