On December 4, Microsoft made a number of announcements on various products and services at the event sponsored by the company, Microsoft Connect ();.
What I particularly focus on was Azure data, analytics, and AI items.In this announcement, there was an important announcement for the two attractive services of Azure.
Let's start with Microsoft's "COSMOS DB" topic.Azure Cosmos DB, a global distributed database service that supports NOSQL natively, has long been technically attractive, but has been a major barrier for many customers.
In this price revision, the minimum RU (request unit) of the container provisioned throughput was reduced from 1000RU/second to 400RU/second.This means that the minimum fee has been reduced from $ 60 per month to $ 24 per month.
Microsoft has introduced a mechanism to share a throughput for customers who divide data into multiple containers, and this is a database -units of throughput.In this revision, the minimum RU in the database unit was changed to 400Ru/second.Here, 10,000 RU/second was the minimum unit.
In other words, the minimum cost of the Cosmos DB database consisting of multiple containers has been reduced from $ 600 per month to $ 24 per month.In addition, the scal -up/down granularity of the provisioned throughput was also reduced to 100 ru/second ($ 6 per month) (regardless of container or database).Considering that the minimum granularity was 1000RU/second, it was literally smaller.
On the other hand, in AI -related, the GA has been announced (GA) for the "Azure Machine Learning Service".
Azure's first machine learning service (currently called "Azure Machine Learning Studio") was a kind of visual development environment for machine learning.This solution has a very high technical difficulty, is a unique technology of Microsoft, and is not compatible with code, so in order to make the most of it, data scientists are needed anyway.It was hard for data scientists to be attractive.
However, the Azure Machine Learning (Azure ML) service, which has been provided this time, supports almost every Python development environment.Python is a particularly popular language for data scientists.Data scientists include the command line, "Pycharm", and even from "Jupyter Notebook", even from the "Azure Databricks" notebooks, Azure ML experiments, data preparation, training, deployment, model management, equipment.In addition to monitoring, you can use various algorithm frameworks (including Pytorch, TensorFlow, Scikit-Learn) for machine learning and deep learning.
The Azure ML service also supports access from Visual Studio Code.In fact, if the Python extension is installed in Visual Studio Code, it can also support Jupyter Notebook, and the editor also uses the Matplotlib visualization function, which has improved convenience.
This article edited an article from overseas CBS Interactive by Asahi Interactive for Japan.