Slashdot Mirror

← Back to Stories (view on slashdot.org)

Microsoft Releases CentOS-Based 'Linux Data Science Virtual Machine' For Azure (betanews.com)

Posted by msmash on from the microsoft-linux-affair dept.

An anonymous reader writes: Microsoft has announced a CentOS-based VM image for Azure called 'Linux Data Science Virtual Machine'. The VM has pre-installed tools such as Anaconda Python Distribution, Computational Network Toolkit, and Microsoft R Open. It focuses on machine learning and analytics, making it a great choice for data scientists. "Thanks to Azure's worldwide cloud infrastructure, customers now have on-demand access to a Linux environment to perform a wide range of data science tasks. The VM saves customers the time and effort of having to discover, install, configure and manage these tools individually. Hosting the data science VM on Azure ensures high availability, elastic capacity and a consistent set of tools to foster collaboration across your team", says Gopi Kumar, Senior Program Manager, Microsoft Data Group.

23 comments

  1. Don't worry... It's Microsoft. by Anonymous Coward · · Score: 0

    I'm sure they'll find a way to fuck it up and ruin it for everyone.

    1. Re:Don't worry... It's Microsoft. by Anonymous Coward · · Score: 0

      Perhaps they will spy on the data and passwords to improve user experience on this OS too?

    2. Re: Don't worry... It's Microsoft. by Anonymous Coward · · Score: 1

      Too late, SystemD already did that.

    3. Re: Don't worry... It's Microsoft. by GioMac · · Score: 1

      another debian freak :))

      --
      "It feels like I'm at the Zoo when reading this thread - I'm frightened, but it's interesting" (c)
  2. Yes, but... by Anonymous Coward · · Score: 3, Funny

    .. does it run ... oh, I see what you did there.

  3. This just in... by Anonymous Coward · · Score: 0

    Pigs have been spotted flying over a frozen Hell.

  4. It's a nice toolset by Anonymous Coward · · Score: 2, Insightful

    As someone who does a fair bit of (what the world now calls) data science --300GB/day of new stuff coming in -- I have to compliment Microsoft on the toolset they have chosen for this image. I know firsthand that it's a pain to set all that stuff up. Just getting that damned R kernel working for Jupyter has cost me many countless hours (albeit mainly on OSX).

    There have been AWS scientific computing images available here and there over the last few years, but none have been all that great.

    1. Re:It's a nice toolset by Anonymous Coward · · Score: 0

      I guess there was more to it for OSX than:

      Install R
      Install Anaconda
      Install Jupyter (via Anaconda)
      Install IRkernel (via one-line R command)

      All of these are available as binaries (for Windows and OSX, IRkernel needs to be compiled in Linux) so the installations are easy. My guess is you didn't use IRkernel or equivalent to get the R kernel working in Jupyter.

      The nice thing is the majority of the Big Data and Data Science stacks are free and open source which makes them more accessible to play around with. Also, if a stack is a pain to set up you can create your own virtual machine and snapshot it after everything is set up (for your use or distribution). That Microsoft is providing good quality, preset images is of course good as well.

    2. Re:It's a nice toolset by GlobalEcho · · Score: 1

      IRKernel is quite tricky to get working if Anaconda is not how you want to get your Python (and some associated stuff). Your list probably works just fine for users without special requirements like mine, and I would recommend anyone reading this to follow it.

        -- The GP poster

  5. Microsoft R Open? by chipschap · · Score: 1

    What in blazes is Microsoft R Open? A Microsoft rebranding of R?

    1. Re:Microsoft R Open? by tlhIngan · · Score: 1

      What in blazes is Microsoft R Open? A Microsoft rebranding of R?

      Basically, yes. There is a Visual Studio R thing that Microsoft added quite recently, so I'm pretty sure they built a R runtime for that so you can develop and deploy using Visual Studio to the image.

    2. Re:Microsoft R Open? by Anonymous Coward · · Score: 0

      What in blazes is Microsoft R Open? A Microsoft rebranding of R?

      Microsoft acquired Revolution Analytics whose R-based products are the most widely used in Enterprise.

  6. Step 1: by cybersquid · · Score: 1

    Embrace.

    1. Re: Step 1: by Anonymous Coward · · Score: 0

      Extend

    2. Re: Step 1: by Anonymous Coward · · Score: 0

      Turns out, luckily, that people have to actually use their products for them to Extingush the competitors.
      I've not, so far, seen a single HPC use case for Azure that isn't completely undermined by its 'features', and researchers might like flashy interfaces but they also like being able to actually fit the DBMS inside their research budget.

  7. I'm afraid by Anonymous Coward · · Score: 0

    they try to make money from our hard community work, I cant believe and support that!

    1. Re:I'm afraid by Anonymous Coward · · Score: 0

      And what, pray tell, did *you* contribute, other than well-wishing?

  8. Wow, they saved me running 3 whole yum commands by daboochmeister · · Score: 1

    THAT'S value-add, by-gummity.

    I think maybe they don't get how easy it is to install/maintain software on Linux. Makes sense, since they've always made it so much harder than it has to be.

    --
    "Ahh! I see you're in that indeterminate Schrodinger state where - oh, uh ... never mind." Dave Bucci
  9. Here's your MSLinux. by Anonymous Coward · · Score: 0

    You've been asking for it for the last 15 years or so.

    Now I want my MSBSD please.

    1. Re: Here's your MSLinux. by nura78 · · Score: 1

      Wouldn't that be Xenix?

  10. Maybe Microsoft should look at the basics first... by Ash-Fox · · Score: 1

    Perhaps Microsoft could work on the basics for their VM cloud services?

    I don't know, like maybe... One click snapshot images?

    Simple VM backup management?

    Instead right now on Azure, they're requiring people to learn and write hundreds of lines in PowerShell scripts to do it. There aren't even working practical examples of how to do this on Azure RM and their own support doesn't know how to write the scripts themselves. The performance is surprisingly poor at the lowest system configurations, particularly because similar configurations else where are more performant. It's a travesty.

    If this was a provider sitting on it's own, without the Microsoft name, nobody would use it. The API is terrible, the web interface is terrible, the PowerShell cmdlets are terrible.

    You would be hard pressed to find anyone that would use this over other providers. The slightly cheaper pricing to other providers like Amazon in certain circumstances just isn't worth all the loss of functionality, ease of use and time for Azure's VM offerings.

    --
    Change is certain; progress is not obligatory.
  11. Re:Microsoft R Open? ... umm, your Google broken? by fygment · · Score: 1

    why would you ask that here? You're lucky you got a straight answer.

    --
    "Consensus" in science is _always_ a political construct.