Sie sind auf Seite 1von 2

Microsoft Codename Cloud Numerics" Simplify working with large data sets by using a rich numerical library that

makes it easy to write algorithms that scale on Azure


We live in the big data era. Sources of data are ubiquitous, storing and collecting data is cheap and our thirst for extracting insights from data to drive our business decisions, research, or scientific discovery is insatiable. However, working with distributed storage and the distributed compute infrastructure required for storing and analyzing big data calls for expertise which is typically not part of the Data Scientists skill set. Data Scientists are trained in mathematics, statistics, and machine learning. Enabling Data Scientists to easily interact and work with distributed big data platforms allows businesses to be more agile in their decision making.

Who should use Cloud Numerics? Data Scientists: Take full advantage of the power and simplicity of Cloud Numerics to extract insights from your data at scale Developers: Integrate Cloud Numerics models developed by your Data Scientist into your production environment taking full advantage of the .NET platform

Cloud Numerics, currently being offered as an early concept via SQL Azure Labs, is a numerical and data analytics library for data scientists, quantitative analysts and others who write computationally intensive C#/F# applications which can now be scaled out, deployed, and run on Windows Azure. Cloud Numerics complements other big data offerings like Microsofts Hadoop on Azure (https://www.hadooponazure.com/), by enabling users to write numerical algorithms that scale-out to the cloud. Cloud Numerics is focused on high performance in memory and numerically intensive computation.

With Cloud Numerics you can:


Write algorithms that scale out using an easy-to-use programming model Have access to an extensive .NET library of numerical algorithms ranging from basic mathematics to advanced statistics to linear algebra Deploy your application to Windows Azure and take advantage of the immense compute power in the cloud

SQL Azure Labs tests concept ideas and prototypes. The goal is to share our current thinking and ideas in specific areas and solicit your feedback so we can shape our future direction.
Learn more about SQL Azure Labs at http://www.microsoft.com /en-us/sqlazurelabs

There are two main concepts in Cloud Numerics: the first is the distributed array object, which is capable of holding large data sets by partitioning memory across several computers. The second is an extensive library of numerical functions that operate on a distributed array. The result is that the user can concentrate on the modeling and the math instead of the complexities of distributed computation. The example below reads a large dataset from disk and applies a Singular Value Decomposition, an operation commonly used in classification and compression methods.
static void Main() { ... // Read data from CSV file var data = Loader.LoadData( CsvLoader("data.csv",1,0.,true)); // Perform a distributed Singular Value Decomposition. var svdResult = Decompositions.Svd(data); ... }

Cloud Numerics enables Data Scientists and others to work with data at scale and tap into the full potential of the information and insight contained in these data sets: Easy to use programming model: Focus on modeling and analytics instead of the complexities of distributed systems Extensive numerical library: Functionality ranging from Mathematics, Statistics, Linear Algebra, Time Series, to Signal Analysis Integration with the Microsoft cloud platform: Read data from Windows Azure storage or from HDFS (supported in the upcoming release) into Cloud Numerics in parallel

Sign up at http://www.microsoft.com/en-us/sqlazurelabs/numerics
The projects in labs are experiments with no current plans to be included in a product and are not production quality.

Das könnte Ihnen auch gefallen