The AmpliStor system comprehensively reduces the Total Cost of Ownership (TCO) of Big Data storage. To do so, the system has been designed to reduce both the Capital and Operational aspects of TCO. The major enablers of this are as follows:

Reduced Capex:

  • Commodity Platform: The AmpliStor Object Storage system is built on a scale-out commodity-based platform. The system utilizes commodity servers, disk drives and network fabric to ride commodity cost curves for platform components.
  • Reduced Storage Overhead: The BitSpread algorithm reduces the raw storage overhead required for high levels of storage durability by 50% (compared to mirrored RAID schemes), and by 70% (compared to triple mirrored cloud storage). This can result in Petabytes of eliminated disk drives and storage capacity, which directly translates into reduced cost.

Reduced Opex:

  • Highly-Available: the system has been designed with no single points of failure, redundancy at all levels, and the highest levels of data durability and availability. Creating a highly available system at the lowest possible cost, directly impacts the cost of managing large-scale big data.
  • Single System View: The system scales as a single pool, to reduce the cost of managing 10’s to 100’s of individual systems. Instead, the administrator can manage 10’s to 100’s of Petabytes as a single system, from a console that manages the entire pool from a “single pane of glass”. This makes it possible for a single storage administrator to manage very large storage configurations.
  • Automation: AmpliStor automates common, time-consuming storage administration tasks. The system virtualizes all element level management of individual components such as disks, and instead enables management of higher level storage “pools”. There is no need for Volume, or RAID group management. Capacity upgrades can be introduced at the click of a button. Software upgrades are performed online. The system performs “self-healing” in the event of a component level failure, with no human intervention required.
  • Fail-in-place model: as disk drives fail, there is no immediate need to perform emergency “disk swaps”. Since the system provides failure protection against very high numbers of simultaneous component failures, and responds to any failure by automatically self-healing, administrators no longer have to be alerted in the middle of the night to respond to simple component level alerts. In addition, this reduces the likelihood of human induced errors related to manual disk or component swapping.
  • Remote monitoring: the Amplidata Global Support Center can remotely monitor customer systems and proactively respond to any conditions that require support, thereby further offloading administrators from this burden.
  • High-Density Storage: with up to 30TB per 1U storage module, AmpliStor can dramatically reduce the required floor space and rack space for large-scale storage.
  • Low-power: with low-power disk drives and processors, AmpliStor drives power consumption down to 65 Watts per Storage module, or as low as 2.1 Watts per TB. This can reduce power consumption by 50-70% compared to traditional systems. 
    Through this comprehensive approach to reducing storage costs, AmpliStor can effectively provide a highly durable storage solution that combines the durability of multiple data copies on disk, or even on low-cost tape, and provide a very compelling cost of ownership for Big Data.