- AmpliStor System
The AmpliStor Object Storage system provides an extensive list of features that provide powerful benefits for applications that require large-scale data storage. These features provide users with a powerful new storage capability that will reduce their capital and operational costs by significant factors, to truly enable cost-effective Petabyte scale storage.
Scalability to Petabytes and Beyond
The system scales-out capacity and performance to support scale from 100’s of Terabytes to hundreds of Petabytes of capacity and to 10′s of Gigabytes per second of throughput. There are no limits on object sizes, or on numbers of objects that can be stored and managed – making it possible to efficiently store Billions of objects. Scale-out AmpliStor by adding AmpliStor Controllers or Storage Nodes “on-the-fly”. No user reconfiguration is required – the system automatically utilizes the added capacity in the storage pool for new objects or as global spare capacity for rebuilds.
Policy-Based Storage Durability
AmpliStor provides a policy-based interface to storage durability. Administrators can specify a unique storage policy per Namespace, that instructs the system how many failures to tolerate. This enables the system to be optimized for a variety of application priorities, and to trade durability versus storage overhead. The “16/4” policy already provides tolerance against any 4 disk or node failures, and provides an incredible “Ten 9’s” of storage durability, and policies that tolerate 6, 8 or more simultaneous failures can also be implemented.
AmpliStor’s BitSpread can also geographically spread a single instance of data across multiple sites, to provide additional site failure protection, without the overhead of storing multiple copies of the data. The system can then ensure that data remains available and fully protected, even in the event of a site failure. Policies can be created that tolerate site failures, and even have sufficient durability to still protect against additional disk or Storage Node failures.
The system requires 50-70% less raw disk capacity while still providing the highest-levels of reliability. This reduction in raw capacity provides dramatically lower power and cooling requirements, and in conjunction with the AmpliStor storage nodes, which require only 2-5 Watts per Terabyte (AS30 Storage Nodes), this creates one of the most power efficient storage solutions on the market.
AmpliStor simplifies systems management by eliminating many low-level and time-consuming disk and volume management tasks, and instead provides simplified management. The administrator may create one or more Namespaces, associate a reliability policy with the Namespace, and then store objects within the Namespace. Note that each Namespace still fully utilizes the back-end storage pool, as described earlier.
The system provides a central management capability through an intuitive web based GUI for systems management and monitoring. The GUI provides simple installation and configuration of new Controllers and Storage Nodes, reporting and trending (disk capacity, CPU, rebuilds, disk failures), events and alerts, and SNMP MIBs and Traps. The system also provides a call-home capability to the Amplidata support centers. With AmpliStor, system administrators can manage an order of magnitude more storage.
Continuous Integrity Verification
The assurance of data integrity is a key aspect of the system as a whole, and is managed in each Storage Node by AmpliStor’s BitDynamics maintenance agents, which perform out-of-band integrity verification and assurance sweeping of all stored objects. The system also performs verification of integrity upon read of an object, as well as providing an administrator command for manual object verification. This enables administrators to report on the protection level of any object in the system, to understand its current status whether 100% protected or in a rebuild status.
Given the level of self-healing and multi-way protection enabled by AmpliStor, it is no longer necessary to urgently replace disks in the case of a failure: the entire system is permanently monitored and it reconfigures itself as needed. In this model, the Storage Node as a FRU (field replaceable unit) becomes a more manageable entity. Over time, the user may decide to swap Storage Nodes that have several failed drives for a new one. This is enabled through AmpliStor functions for decommissioning a node, which ensures that data is cleanly transferred off of the node before it is powered down.
The AmpliStor Object Storage system provides a variety of application interfaces to facilitate access to AmpliStor services from a wide variety of applications. Through this choice of interfaces, AmpliStor reduces the barriers for users to access AmpliStor services from web-based (HTML) applications, compiled/embedded applications, or from management scripts.
AmpliStor provides simple, native http/REST object-based API access to enable web based applications to store, access and manage objects directly. This interface method is easily embedded in web-based applications, in an http POST (PUT), GET, DELETE style interface that follows RESTful web API guidelines. For many online applications that manage media data (images, video, audio), this is the most natural method of interfacing with the underlying storage system. The popular libcurl library available on most Linux platforms make it easy to call http/REST calls directly from scripts and applications.
Microsoft. Net SDK
Enables native access from .Net style applications on the Microsoft Windows platform.
The AmpliStor REST API can be natively accessed from Java code via existing Java httplibrary functions.
The system may be accessed over a variety of WebDAV compliant tools, such as CyberDuck, for simple drag-and-drop style file access via a convenient Web Folders style interface.
AmpliStor provides a scriptable, Python-based Command Line Interface that allows simple interactive management of AmpliStor systems, or embedding of frequently used commands into a simple to use management scripts.
AmpliStor also provides a wide variety of partner based interface methods, tools and solutions for accessing the system from file interfaces, archiving tools, and cloud storage solutions. See the full Partner Solutions here.