Is there a Java SDK for Azure Machine Learning service? If not, is there a way to create Azure ML pipelines, experiments etc from Java codebase?
No, today there is only a SDK for Python. Alternatively you could use the Azure CLI extension for ML which you could call from your Java code: https://learn.microsoft.com/en-us/azure/machine-learning/service/reference-azure-machine-learning-cli
The Azure Machine Learning Studio directly supports only R & Python.
There is no way to handle Azure ML pipelines with java code at the moment.
However, you can consume Azure Cognitive Services (which provides pre-trained models) through the exposed Java API.
Refer this for more details.
Related
So I am using Azure Functions at work and thought I would have a play and install them on my own server. I have successfully installed Azure Functions Runtime 2 (preview).
I have then followed the Java tutorial to create an Azure Function :
https://learn.microsoft.com/en-us/azure/azure-functions/functions-create-first-java-maven
How do I then deploy this function to my own Azure functions server?
In the guide it says about using :
az login
mvn azure-functions:deploy
Unfortunately, mvn azure-functions:deploy is to deploy functions to Azure site as az login is required before deployment, which doesn't support deployment to on-premises Runtime portal.
But the key point is, the on-premises Runtime is obsolete(one year behind the latest bits), new project probably can't work with it even if we find how to publish.
Since v2 becomes GA, it is recommended to leverage custom image for usages out of Azure box.
Update
Missed one point. As AF Team answered in the issue you post, no Java Image for now so the usage of Azure Java Function outside Azure is blocked unless we could figure out creating the image on our own.
As opposed to writing an Azure Function in C# or JS, I'd like to write a Azure Function that runs azure cli commands.
How would I go about doing that?
As Ashok mentioned, Azure function does not support Azure CLI now.At present, there are several Generally available and experimental languages in Azure Function V1, and preview languages in V2.
You could see the Azure roadmap fo find information about planned changes to language support. If you want Azure Function supports CLI command, This is the feedback.
C# and Javascript are the only non-experimental languages at the moment. If you are looking to automate Azure operations, have you looked at Azure Automation? https://learn.microsoft.com/en-us/azure/automation/automation-intro
This article discusses that an Azure function can easily be deployed in local datacenters without Azure.
Open source runtime that runs anywhere
The Functions runtime, templates, UI and underlying WebJobs SDK are all open source projects being developed publically. The Azure Function runtime will be portable so you can run Functions anywhere - on Azure, in your datacenter or other clouds.
Question:
How can I create an installer package that I can give to customers?
What are the requirements of this installer?
Can the target environment be ASP.NET core or run on Linux/Mac?
Azure Functions Runtime (currently in preview) enables an on premise deployment of Azure Functions, which includes portal, functions execution runtime and other components, all in an installer you can use. This also removes some of the cloud dependencies like Azure Storage, allowing you to run all within your internal network.
Currently, .NET Core is not supported, so you'll need to deploy this on a Windows environment, but we're actively working on migrating the core Azure Functions runtime bits and the on premise Azure Functions Runtime would follow.
You can find the requirements on the document I've linked to.
Another option is do download the functions releases directly from the repository here, but that does not have an installer that enables an end-to-end deployment experience (or components like the portal), so you'd need to provide that.
I am currently looking into using ARM to deploy new environments of our Azure Components. We have a lot of web apis that are deployed to Cloud Services, since we need access to the underlying OS to get CPU usage for App Dynamics metrics. When reading through ARM, it seems as if they are deprecating Cloud Services, but I have some confusion about what is replacing it. I see that you can create Web Apps using this and use Publish from Visual Studio to deploy the app it, but I see no options for Cloud Services. So what is the preferred method to do this? Creating a VM and deploying using Web Deploy? What about VM scaling. Any help would be greatly appreciated.
My company is in the process of moving a legacy app to Azure using Cloud Services and we were concerned about the future of Cloud Services. Since we are somewhat early into the project and would like to use the ARM model we thought it would be easier to make a move now if we knew the future. After a few conversations with project leads at Microsoft on this topic we were only told that we should continue working with Cloud Services as they would continue to be supported.
It's quite clear that the move to ARM was not well thought out in terms of direction and consequences of existing services that many people are already using. Reading between the lines I would say that there is no plan to convert Cloud Services from the ASM to ARM model.
If application insights (https://azure.microsoft.com/en-us/services/application-insights/) are sufficient for your monitoring needs, then you can use web apps. These can be deployed via ARM, and can automatically deploy from a git repo or web deploy package. See this example:
https://github.com/Azure/azure-quickstart-templates/tree/master/201-web-app-github-deploy
The issue here that you may be trying to do operations which are restricted by the web app sandbox.
Failing that, you can use an ARM template to set up a VM and then invoke a custom site extension which will run powershell code to do further provisioning. This powershell code can fetch any package that you may want to install. See this ARM template for example: https://github.com/Azure/azure-quickstart-templates/tree/9ad72f1f5f0008c14311be79eee036b871712394/201-list-storage-keys-windows-vm
Once the VM is created you would be able to modify it and scale as needed.
I'm new to cloud foundry and would like to do a detailed comparison between Windows Azure and Cloud Foundry. I've searched around a lot but haven't been able to find anything useful. Is there a good post or some material which does a detailed feature wise comparison of the two?
Regards,
Vikram
You're not exactly comparing like for like here. Azure has IaaS type capability as well as PaaS, not only can you push applications to it but you can also deploy VM images too, including Linux.
However, as Cloud Foundry is open source the number of runtimes and frameworks if supports evolves quickly as VMWare openly encourage contributions from the OSS community. Correct me if I am wrong, but in a lot of cases, with Azure, you have to provide the runtime you wish to use where as Cloud Foundry supports them "natively", if that's the correct word?!
Right now, Cloud Foundry supports the following runtimes and frameworks;
Runtimes
java - 1.6.0_24
java7 - 1.7.0_04
node - 0.4.12
node06 - 0.6.8
node08 - 0.8.2
ruby18 - 1.8.7p357
ruby19 - 1.9.2p180
Frameworks
grails
java_web
lift
node
play
rack
rails3
sinatra
spring
standalone
They also provide all the major storage services too, including MySQL, Postgres, RabbitMQ and Redis.
The actual open source project supports a whole lot more too!
I don't know much about Azure. But, I've used Cloudfoundry. Its great for Java deployments. I use VMC from Ruby gem for deployment and its so 3 - 4 step easy way to push your WAR to cloudfoundry server. They also provide neat documentation for configuration and setup. Oh and adding services (like Mongo DB, MySql) are also very simple. Though,sometimes, debugging server related issues are annoying with it. But, overall, its good for me :)
http://docs.cloudfoundry.com/tools/deploying-apps.html
Likewise, I don't know much about Cloud Foundry but I'm using Windows Azure for a couple of client projects and I have to say that I'm now very impressed with the development environment. I'm using the Websites Preview feature with continuous Git deployment via BitBucket. Setting this up is a breeze and allows me to push my changes to BitBucket and have Windows Azure deploy them automatically for me. There's currently no ability to run unit tests as part of the deployment cycle as per other cloud platforms (e.g. AppHarbor) but the feature set for getting up and running with a .NET application and SQL Azure database is now pretty slick. Here's a couple of links:
Deploying an ASP.NET Web Application to a Windows Azure Web Site and SQL Database
Continuous Deployment with Windows Azure Websites and Bitbucket