IBM Elastic Storage Server (ESS)

2017: IBM Elastic Storage Server

IBM Elastic Storage Server is technology for ensuring scalability and performance when processing large volumes of data.

Specific Features

• Performance improvement thanks to a possibility of sharing of the storage resources given by all workloads using a uniform pool for files, objects and workloads of Hadoop
• Scaling of capacity and/or performance using modular assemblies with automatic balancing of loading
• Developed in IBM technology of the dispersed RAID on the basis of noiseproof coding allows to recover faulty disks
• Placement of several user systems, setup of selection of resources and scaling in process of measurement of requirements
• Use of the integrated system created on the basis of the IBM Power servers and the software of IBM Spectrum Scale
• Reduction of time of backup and recovery by providing a direct system for backup
• Elimination of growth of number of fayler and bottlenecks of performance of the storage systems (NAS) connected to network.

2014: Technology of a software-defined storage for processing of Big Data

The IBM company in the spring of 2014 expands a portfolio of a software-defined storage, giving the chance to the organizations to reduce costs and to process data in any quantity, a type and using the different storage devices located worldwide.

One of new technologies – Elastic Storage – offers performance and unlimited scaling. It automatically moves data on the most economic storage device, thereby reducing storage costs for 90 percent.

This patented technology was created in research laboratory IBM Research. Elastic Storage allows the organizations not just to manage, and with benefit for itself to use the growing amounts of data generated by numerous devices, sensors, different business processes and social networks. The new solution for storage is ideal for the applications requiring processing of a large number of data and high-speed access to information arrays – from processing of seismic data, risk management and financial analytics, drawing up weather forecasts and scientific research before actions planning in real time.

Software-defined storage is a set of means of software which automatically manage data at the local and global level, providing unprecedented access rate, easy management and a scaling option of infrastructure in process of growth of information volumes. In addition, these technologies can work on any equipment and provide the automated and virtualized storage.

The technologies which are been the basis for Elastic Storage were also used in the IBM Watson supercomputer for its victory over two champions of the television game Jeopardy! (analog of the Russian "Game"). Using Elastic Storage about five terabytes of "knowledge" were loaded into memory of the supercomputer in only several minutes. Thus, during the television match IBM Watson had an access to 200 million pages of the structured and unstructured data, including the complete text of Wikipedia.

One of the main reasons that Elastic Storage was used for IBM Watson was the fact that it can be scaled to thousands of yottabytes. The yottabyte is one billion the petabyte requiring a data processing center of one million city quarters in size.

The research laboratory IBM Research showed that Elastic Storage can successfully scan 10 trillions files on one memory block in 43 minutes that guarantees the unsurpassed performance of the similar systems in the analysis of huge data arrays.

The global file software of IBM which provides management of storage online, the scalable access and the integrated tools capable to manage huge data arrays and billions of files is the cornerstone of Elastic Storage technology. For example, Elastic Storage uses server a flash memory for performance improvement more than six times in comparison with a SAS disks. If the server has a flash memory, it is automatically used as a cache memory for performance improvement.

Elastic Storage virtualizes storage systems, allowing several systems and applications to use memory shares. It also provides transparent global data access without need to modify applications or to add additional management systems for storage. As Elastic Storage does not rely upon the central system for determination of recording location of the file, clients have data access even in case of failures of work of software or the equipment.

For Laboratory of computing and information services (Computational and Information Services Laboratory, CISL) which is in the structure of the National center of the USA for a research of a terrestrial atmosphere (National Center for Atmospheric Research), the growing amounts of data – an integral part of its structure. The organization manages more than 50 petabytes of information which is stored in the centers in Wyoming and Colorado, and uses Elastic Storage with the purpose to provide to researchers quick access to various data.

The key Elastic Storage component is its opportunity to automatically move data to strategically preferable and most cost-efficient storage system. Thanks to the functions focused on relevant politicians and data analysis opportunities in real time Elastic Storage can automatically redirect seldom used information on less expensive tape drives. At the same time often used data can be stored in high-speed a flash systems that guarantees to users online access to information.

Moreover, functions of native enciphering and safe removal of data allow to erase irrevocably information that conforms to the regulatory requirements HIPAA and Sarbanes-Oxley.

Thanks to OpenStack software maintenance for management of cloud computing, Elastic Storage also gives to customers opportunities of storage, data management and joint work in private, public and hybrid cloud environments. In addition to compatibility with OpenStack Cinder and Swift services, Elastic Storage also supports other open API, such as POSIX and Hadoop.

While traditional DWH moves information between the designated systems for follow-up and analysis of transaction, Elastic Storage can automatically configure balance of resources for support of all types of loading of the application, including analytics tools on the basis of Hadoop. It allows to accelerate considerably process of information analysis and saves from expensive need of creation of numerous copies.

Elastic Storage will also be available as a cloud service based on IBM SoftLayer in 2014.