StateServer 5.0 Beta From ScaleOut SoftwareArchitecture to run in cloud infrastructures and scale to hundreds of servers
This is a Press Release edited by StorageNewsletter.com on 2012.02.23
IMDGs provide distributed, in-memory storage which scales application performance and enables query and analysis of stored data. Version 5.0 introduces a new architecture designed to run in cloud infrastructures and scale to hundreds of servers. This version also adds the ability to transparently combine multiple IMDGs into a single virtual data grid and other key innovations.
"Distributed, in-memory data grids are being rapidly adopted for both on-premise and cloud use," said Dr. William L. Bain, founder and CEO of ScaleOut Software. "Enterprise applications increasingly require immediate access to data globally from a variety of sources, and developers are faced with the challenge of how to share information across logical and geographical boundaries. Addressing these challenges is one of the focus areas for Version 5.0."
With its ScaleOut GeoServer option, Version 5.0 introduces features for multi-site integration. In addition to asynchronous data replication for disaster recovery, GeoServer now lets users combine in-memory data grids at multiple sites into a single, logically coherent, virtual grid, enabling data access across sites without the need to explicitly track which site is hosting requested data. This capability saves both development time and resources at a time when IT head count and budgets continue to be under scrutiny. This helps increase efficiency and productivity across the enterprise.
With the enhanced feature set in ScaleOut AnalyticsServer (formerly Grid Computing Edition), Version 5.0 enables in-memory data grids to offer capabilities and performance in parallel query and data analysis. AnalyticsServer now supports optimized parallel query of stored data based on Java or C# object properties, and it supports the Microsoft Language Integrated Query (LINQ) standard. These capabilities have been integrated into its map/reduce engine to provide a platform for memory-based data analysis. Its object-oriented map/reduce model saves development time, and its fast, memory-based analysis lowers latency in comparison to other map/reduce platforms.
"Big data, rapid data churn and the need to minimize latency present huge obstacles for enterprises performing data analysis," said Dr. Bain. "Popular analytical techniques like Hadoop's map/reduce work well for petabyte-sized datasets but are complex to use for many problems, requiring significant effort to develop and tune applications."
Continued Dr. Bain: "ScaleOut has integrated the industry's powerful map/reduce analysis model into its scalable IMDG architecture to create an intuitive and extremely high performance platform for analyzing memory-based datasets. We also expect that hosting ScaleOut AnalyticsServer in a cloud environment will create a breakthrough opportunity by enabling users to easily take advantage of a very powerful and elastic computing platform for their data analysis."
"2012 holds an exciting new opportunity to leverage the cloud, as many datasets now can be held entirely in memory," said David Brinker, COO of ScaleOut Software. "Cloud computing makes it practical to harness hundreds of servers to perform data analyses. As global enterprises begin to adopt private and eventually, public clouds, a new generation of IMDGs pioneered by ScaleOut StateServer Version 5.0 will revolutionize data analysis."
With version 5.0, the GeoServer option now offers the flexibility to support multiple usage models in sharing stored data across multiple IMDGs. Customers needing fast data retrieval worldwide for data principally stored and updated at a central site now can access these data and receive timely updates over WANss. For example, financial institutions can make slowly changing portfolio information available to their satellite offices through a network of IMDGs. In a separate usage model, applications can transfer ownership of shared data among sites as needed for synchronized updating. For example, Web sites can expand into a cloud-host Web farm to handle high traffic loads and transparently migrate shopping carts to requesting Web servers as needed.
Version 5.0 of AnalyticsServer integrates parallel query with an map/reduce development model to sidestep many of the complexities found in popular map/reduce platforms, such as Hadoop. Called parallel method invocation (PMI), ScaleOut's approach identifies objects to be analyzed using a parallel query on relevant object properties; queries can be formulated using either Java filters or C# LINQ expressions. Also, PMI offers a self-tuning parallel architecture that extracts full parallelism in both the map and reduce phases and takes advantage of all IMDG servers, processors, and cores. In addition to PMI, it adds a columnar computation model called single method invocation for fast analysis and updating of specific objects in the IMDG.