May 12, 201511 yr Hello unraid & virtualization community, I'm admittedly a virtualization n00b, so some guidance here would be greatly appreciated. I'm wondering, should I setup a docker container or a KVM vitalization image. I'm looking to be able to run some scientific computing work on my unraid machine. Specifically, I want to be able to run a project Jupyter/IPython notebook, and perhaps a R-Studio Server install. With the ipython notebooks, you can utilize multiple cores (if you write your code to benefit from that). I will need to be able to compile C/C++ code as well; however I'm not sure if I would be trying to run any CUDA code (which would require GPU pass-through). It should be pointed out that all these services I will want to be able to access remotely as well. Any input/thoughts would be greatly appreciated.
May 12, 201511 yr Hello unraid & virtualization community, I'm admittedly a virtualization n00b, so some guidance here would be greatly appreciated. I'm wondering, should I setup a docker container or a KVM vitalization image. I'm looking to be able to run some scientific computing work on my unraid machine. Specifically, I want to be able to run a project Jupyter/IPython notebook, and perhaps a R-Studio Server install. With the ipython notebooks, you can utilize multiple cores (if you write your code to benefit from that). I will need to be able to compile C/C++ code as well; however I'm not sure if I would be trying to run any CUDA code (which would require GPU pass-through). It should be pointed out that all these services I will want to be able to access remotely as well. Any input/thoughts would be greatly appreciated. A VM is the route to go. GPU pass through does not work without KVM. Everything else is a moot point then since this is one of your requirements.
May 12, 201511 yr Author Hello unraid & virtualization community, I'm admittedly a virtualization n00b, so some guidance here would be greatly appreciated. I'm wondering, should I setup a docker container or a KVM vitalization image. I'm looking to be able to run some scientific computing work on my unraid machine. Specifically, I want to be able to run a project Jupyter/IPython notebook, and perhaps a R-Studio Server install. With the ipython notebooks, you can utilize multiple cores (if you write your code to benefit from that). I will need to be able to compile C/C++ code as well; however I'm not sure if I would be trying to run any CUDA code (which would require GPU pass-through). It should be pointed out that all these services I will want to be able to access remotely as well. Any input/thoughts would be greatly appreciated. A VM is the route to go. GPU pass through does not work without KVM. Everything else is a moot point then since this is one of your requirements. Hi John, Thanks for the reply. I stated that I was not sure if I would be running CUDA code (I knew would require a KVM image), but I guess the better question might be, in abscence of a GPU requirement, how would a docker container perform vs. a KVM image for multicore computations?
May 12, 201511 yr Hello unraid & virtualization community, I'm admittedly a virtualization n00b, so some guidance here would be greatly appreciated. I'm wondering, should I setup a docker container or a KVM vitalization image. I'm looking to be able to run some scientific computing work on my unraid machine. Specifically, I want to be able to run a project Jupyter/IPython notebook, and perhaps a R-Studio Server install. With the ipython notebooks, you can utilize multiple cores (if you write your code to benefit from that). I will need to be able to compile C/C++ code as well; however I'm not sure if I would be trying to run any CUDA code (which would require GPU pass-through). It should be pointed out that all these services I will want to be able to access remotely as well. Any input/thoughts would be greatly appreciated. A VM is the route to go. GPU pass through does not work without KVM. Everything else is a moot point then since this is one of your requirements. Hi John, Thanks for the reply. I stated that I was not sure if I would be running CUDA code (I knew would require a KVM image), but I guess the better question might be, in abscence of a GPU requirement, how would a docker container perform vs. a KVM image for multicore computations? For your specific use-case, it'd be hard to say, but given your needs, I would highly advise a VM. VMs allow for more customization and tuning of the operating system and software packages, which makes them ideal for self-managed operating environments. To tweak the environment for a container, much more work must be done. For custom use cases like yours, VMs provide a familiar working environment with less barrier to entry in terms of learning / training. Containers are ideal for hosting single applications whereas VMs can be tuned operating environments for multiple applications. From a performance standpoint, KVM will have slightly more overhead than Docker due to hardware emulation, but a lot of that is overcome by using hardware-assisted virtualization. In short, Containers are great, but for custom environments like this, VMs are just far more flexible.
May 12, 201511 yr Author Hello unraid & virtualization community, I'm admittedly a virtualization n00b, so some guidance here would be greatly appreciated. I'm wondering, should I setup a docker container or a KVM vitalization image. I'm looking to be able to run some scientific computing work on my unraid machine. Specifically, I want to be able to run a project Jupyter/IPython notebook, and perhaps a R-Studio Server install. With the ipython notebooks, you can utilize multiple cores (if you write your code to benefit from that). I will need to be able to compile C/C++ code as well; however I'm not sure if I would be trying to run any CUDA code (which would require GPU pass-through). It should be pointed out that all these services I will want to be able to access remotely as well. Any input/thoughts would be greatly appreciated. A VM is the route to go. GPU pass through does not work without KVM. Everything else is a moot point then since this is one of your requirements. Hi John, Thanks for the reply. I stated that I was not sure if I would be running CUDA code (I knew would require a KVM image), but I guess the better question might be, in abscence of a GPU requirement, how would a docker container perform vs. a KVM image for multicore computations? For your specific use-case, it'd be hard to say, but given your needs, I would highly advise a VM. VMs allow for more customization and tuning of the operating system and software packages, which makes them ideal for self-managed operating environments. To tweak the environment for a container, much more work must be done. For custom use cases like yours, VMs provide a familiar working environment with less barrier to entry in terms of learning / training. Containers are ideal for hosting single applications whereas VMs can be tuned operating environments for multiple applications. From a performance standpoint, KVM will have slightly more overhead than Docker due to hardware emulation, but a lot of that is overcome by using hardware-assisted virtualization. In short, Containers are great, but for custom environments like this, VMs are just far more flexible. Thanks Jon, this is great information
May 12, 201511 yr Hello unraid & virtualization community, I'm admittedly a virtualization n00b, so some guidance here would be greatly appreciated. I'm wondering, should I setup a docker container or a KVM vitalization image. I'm looking to be able to run some scientific computing work on my unraid machine. Specifically, I want to be able to run a project Jupyter/IPython notebook, and perhaps a R-Studio Server install. With the ipython notebooks, you can utilize multiple cores (if you write your code to benefit from that). I will need to be able to compile C/C++ code as well; however I'm not sure if I would be trying to run any CUDA code (which would require GPU pass-through). It should be pointed out that all these services I will want to be able to access remotely as well. Any input/thoughts would be greatly appreciated. A VM is the route to go. GPU pass through does not work without KVM. Everything else is a moot point then since this is one of your requirements. Hi John, Thanks for the reply. I stated that I was not sure if I would be running CUDA code (I knew would require a KVM image), but I guess the better question might be, in abscence of a GPU requirement, how would a docker container perform vs. a KVM image for multicore computations? For your specific use-case, it'd be hard to say, but given your needs, I would highly advise a VM. VMs allow for more customization and tuning of the operating system and software packages, which makes them ideal for self-managed operating environments. To tweak the environment for a container, much more work must be done. For custom use cases like yours, VMs provide a familiar working environment with less barrier to entry in terms of learning / training. Containers are ideal for hosting single applications whereas VMs can be tuned operating environments for multiple applications. From a performance standpoint, KVM will have slightly more overhead than Docker due to hardware emulation, but a lot of that is overcome by using hardware-assisted virtualization. In short, Containers are great, but for custom environments like this, VMs are just far more flexible. Thanks Jon, this is great information You are most welcome! There is a lot of information in the unRAID 6 manual (see link in my signature) on how to get started with VMs. There's also some good info in the KVM forum. Be sure to post again over there if you have questions or difficulties getting started.
May 12, 201511 yr @ogi: from a practical standpoint, are there any ready-to-use docker, kvm (or even xen and vmware) image for those apps that you need to run? Because if there are, then you can get started very quickly and evaluate them right away.
May 12, 201511 yr Author @ogi: from a practical standpoint, are there any ready-to-use docker, kvm (or even xen and vmware) image for those apps that you need to run? Because if there are, then you can get started very quickly and evaluate them right away. There is, sort of. https://github.com/ContinuumIO/docker-images/tree/master/miniconda3 This guy contains the the conda python development environment, which allows for the installation of other python modules like cython, numpy, numba, irkernel, and so on... > conda install numpy > conda install cython > conda install ipython-notebook etc etc... Cython requires a GCC compiler (which I'm not sure is present on the docker image). I actually have it running on boot2docker on my windows desktop, I haven't messed with it too much (just got it up and running a day ago) but I suppose what would make most sense here is to actually benchmark some intensive computations. After some googling, there is a R-Studio Server docker image too: https://github.com/rocker-org/rocker/wiki/Using-the-RStudio-image I have no feeling for how multicore computations will scale with Docker vs. KVM, but if they can both handle multi-core computations, and intensive memory usage (big data is just data that doesn't fit into memory after-all ) then I suppose pursuing the 'easiest' method would be what makes sense here.
May 12, 201511 yr @ogi: from a practical standpoint, are there any ready-to-use docker, kvm (or even xen and vmware) image for those apps that you need to run? Because if there are, then you can get started very quickly and evaluate them right away. There is, sort of. https://github.com/ContinuumIO/docker-images/tree/master/miniconda3 This guy contains the the conda python development environment, which allows for the installation of other python modules like cython, numpy, numba, irkernel, and so on... > conda install numpy > conda install cython > conda install ipython-notebook etc etc... Cython requires a GCC compiler (which I'm not sure is present on the docker image). I actually have it running on boot2docker on my windows desktop, I haven't messed with it too much (just got it up and running a day ago) but I suppose what would make most sense here is to actually benchmark some intensive computations. After some googling, there is a R-Studio Server docker image too: https://github.com/rocker-org/rocker/wiki/Using-the-RStudio-image I have no feeling for how multicore computations will scale with Docker vs. KVM, but if they can both handle multi-core computations, and intensive memory usage (big data is just data that doesn't fit into memory after-all ) then I suppose pursuing the 'easiest' method would be what makes sense here. Are you worried about the overhead that Docker and KVM consume, compared to installing those applications on the bare metal? I think Docker will have very little overhead as it runs on the same kernel as unraid's, but even so KVM (and Xen and VMWare) also uses very low overhead on modern system, if I'm not mistaken they're all in the single digit percentage (forgot what benchmarks were used). Also important, how many instances of this system will you need to run on the server?
May 12, 201511 yr Author @ogi: from a practical standpoint, are there any ready-to-use docker, kvm (or even xen and vmware) image for those apps that you need to run? Because if there are, then you can get started very quickly and evaluate them right away. There is, sort of. https://github.com/ContinuumIO/docker-images/tree/master/miniconda3 This guy contains the the conda python development environment, which allows for the installation of other python modules like cython, numpy, numba, irkernel, and so on... > conda install numpy > conda install cython > conda install ipython-notebook etc etc... Cython requires a GCC compiler (which I'm not sure is present on the docker image). I actually have it running on boot2docker on my windows desktop, I haven't messed with it too much (just got it up and running a day ago) but I suppose what would make most sense here is to actually benchmark some intensive computations. After some googling, there is a R-Studio Server docker image too: https://github.com/rocker-org/rocker/wiki/Using-the-RStudio-image I have no feeling for how multicore computations will scale with Docker vs. KVM, but if they can both handle multi-core computations, and intensive memory usage (big data is just data that doesn't fit into memory after-all ) then I suppose pursuing the 'easiest' method would be what makes sense here. Are you worried about the overhead that Docker and KVM consume, compared to installing those applications on the bare metal? I think Docker will have very little overhead as it runs on the same kernel as unraid's, but even so KVM (and Xen and VMWare) also uses very low overhead on modern system, if I'm not mistaken they're all in the single digit percentage (forgot what benchmarks were used). Also important, how many instances of this system will you need to run on the server? I will be running 1 instance only. I'm not so worried about performance losses through virtualization, as I am about simply choosing the easiest tool to work with in this case. I've done only some really elementary virtualization with virtualbox before, so I guess you can say I'm in the camp of being too dumb to know the right questions to ask. In general I've always thought of VMs as scaling by adding more instances, but in this case I'm looking to just have 1 instance be capable of utilizing multiple threads of a CPU as well as many GBs worth of memory. Thanks for your input!
May 12, 201511 yr I will be running 1 instance only. I'm not so worried about performance losses through virtualization, as I am about simply choosing the easiest tool to work with in this case. I've done only some really elementary virtualization with virtualbox before, so I guess you can say I'm in the camp of being too dumb to know the right questions to ask. In general I've always thought of VMs as scaling by adding more instances, but in this case I'm looking to just have 1 instance be capable of utilizing multiple threads of a CPU as well as many GBs worth of memory. Thanks for your input! AFAIK, Virtualization is primarily usually used to partition a machine so it can be used by multiple smaller instances, or to simulate multiple machines (mostly in lab scenario), and/or to run different OSes on a single machine. So running a single vm on a single machine is kind of pointless, unless you have plans to clone said vm to be run on multiple machines or you want it to be portable to be moved to other servers in the future. Or are you looking to run VM in this case because you have a more powerful unraid machine than your other machine that runs virtualbox on? IMHO, the simplest way in your case is to duplicate the environment that your instructor/class/sources use (the distro, etc), so that you don't have to deal with discrepancies in OS/distro level, unless that is of your interest as well.
May 12, 201511 yr Author I will be running 1 instance only. I'm not so worried about performance losses through virtualization, as I am about simply choosing the easiest tool to work with in this case. I've done only some really elementary virtualization with virtualbox before, so I guess you can say I'm in the camp of being too dumb to know the right questions to ask. In general I've always thought of VMs as scaling by adding more instances, but in this case I'm looking to just have 1 instance be capable of utilizing multiple threads of a CPU as well as many GBs worth of memory. Thanks for your input! AFAIK, Virtualization is primarily usually used to partition a machine so it can be used by multiple smaller instances, or to simulate multiple machines (mostly in lab scenario), and/or to run different OSes on a single machine. So running a single vm on a single machine is kind of pointless, unless you have plans to clone said vm to be run on multiple machines or you want it to be portable to be moved to other servers in the future. Or are you looking to run VM in this case because you have a more powerful unraid machine than your other machine that runs virtualbox on? IMHO, the simplest way in your case is to duplicate the environment that your instructor/class/sources use (the distro, etc), so that you don't have to deal with discrepancies in OS/distro level, unless that is of your interest as well. >Or are you looking to run VM in this case because you have a more powerful unraid machine than your other machine that runs virtualbox on? Bingo! This is the bulk of it, but there are some other benefits. 1) I can have identical development environments whether I'm on my laptop or desktop, as I am interacting with the same kernels 2) I can work remotely (this will involve setting up some network configuration but I'll worry about that when I get there) 3) As you said, the NAS has far more computational power/memory than my laptop (or desktop for that matter). I am unsure if I would be able to run the python/R/julia development environments on the bare metal unraid OS (if I was a linux guru I could probably figure it out, but I'm not), but the state of python development (and R) is pretty good for Linux, where-as compiling python packages using Cython in windows is like pulling teeth.
May 12, 201511 yr I will be running 1 instance only. I'm not so worried about performance losses through virtualization, as I am about simply choosing the easiest tool to work with in this case. I've done only some really elementary virtualization with virtualbox before, so I guess you can say I'm in the camp of being too dumb to know the right questions to ask. In general I've always thought of VMs as scaling by adding more instances, but in this case I'm looking to just have 1 instance be capable of utilizing multiple threads of a CPU as well as many GBs worth of memory. Thanks for your input! AFAIK, Virtualization is primarily usually used to partition a machine so it can be used by multiple smaller instances, or to simulate multiple machines (mostly in lab scenario), and/or to run different OSes on a single machine. So running a single vm on a single machine is kind of pointless, unless you have plans to clone said vm to be run on multiple machines or you want it to be portable to be moved to other servers in the future. Or are you looking to run VM in this case because you have a more powerful unraid machine than your other machine that runs virtualbox on? IMHO, the simplest way in your case is to duplicate the environment that your instructor/class/sources use (the distro, etc), so that you don't have to deal with discrepancies in OS/distro level, unless that is of your interest as well. >Or are you looking to run VM in this case because you have a more powerful unraid machine than your other machine that runs virtualbox on? Bingo! This is the bulk of it, but there are some other benefits. 1) I can have identical development environments whether I'm on my laptop or desktop, as I am interacting with the same kernels 2) I can work remotely (this will involve setting up some network configuration but I'll worry about that when I get there) 3) As you said, the NAS has far more computational power/memory than my laptop (or desktop for that matter). I am unsure if I would be able to run the python/R/julia development environments on the bare metal unraid OS (if I was a linux guru I could probably figure it out, but I'm not), but the state of python development (and R) is pretty good for Linux, where-as compiling python packages using Cython in windows is like pulling teeth. Yeah, in that case I think the quickest way to replicate what you have on virtualbox is by installing the same distro on a KVM instance. I think there's a way to copy and convert your virtualbox disk images to KVM too, so you don't even need to reinstall things; and would make a good bnechmark against your notebook since both environment are practically identical.
May 12, 201511 yr Author I will be running 1 instance only. I'm not so worried about performance losses through virtualization, as I am about simply choosing the easiest tool to work with in this case. I've done only some really elementary virtualization with virtualbox before, so I guess you can say I'm in the camp of being too dumb to know the right questions to ask. In general I've always thought of VMs as scaling by adding more instances, but in this case I'm looking to just have 1 instance be capable of utilizing multiple threads of a CPU as well as many GBs worth of memory. Thanks for your input! AFAIK, Virtualization is primarily usually used to partition a machine so it can be used by multiple smaller instances, or to simulate multiple machines (mostly in lab scenario), and/or to run different OSes on a single machine. So running a single vm on a single machine is kind of pointless, unless you have plans to clone said vm to be run on multiple machines or you want it to be portable to be moved to other servers in the future. Or are you looking to run VM in this case because you have a more powerful unraid machine than your other machine that runs virtualbox on? IMHO, the simplest way in your case is to duplicate the environment that your instructor/class/sources use (the distro, etc), so that you don't have to deal with discrepancies in OS/distro level, unless that is of your interest as well. >Or are you looking to run VM in this case because you have a more powerful unraid machine than your other machine that runs virtualbox on? Bingo! This is the bulk of it, but there are some other benefits. 1) I can have identical development environments whether I'm on my laptop or desktop, as I am interacting with the same kernels 2) I can work remotely (this will involve setting up some network configuration but I'll worry about that when I get there) 3) As you said, the NAS has far more computational power/memory than my laptop (or desktop for that matter). I am unsure if I would be able to run the python/R/julia development environments on the bare metal unraid OS (if I was a linux guru I could probably figure it out, but I'm not), but the state of python development (and R) is pretty good for Linux, where-as compiling python packages using Cython in windows is like pulling teeth. Yeah, in that case I think the quickest way to replicate what you have on virtualbox is by installing the same distro on a KVM instance. I think there's a way to copy and convert your virtualbox disk images to KVM too, so you don't even need to reinstall things; and would make a good bnechmark against your notebook since both environment are practically identical. Man, I'm awful at this explanation stuff. I mentioned virtualbox to indicate what level of experience of virtualization I have. I don't have a virtualbox image of the development environment I would like to setup. It's sounding like just running a KVM instance of ubuntu server (or something to that effect) might be what is easiest and gives me the most flexibility. Either way, benchmarks will be the order of the day when I get unraid 6 up and running.
May 12, 201511 yr AFAIK, Virtualization is primarily usually used to partition a machine so it can be used by multiple smaller instances, or to simulate multiple machines (mostly in lab scenario), and/or to run different OSes on a single machine. So running a single vm on a single machine is kind of pointless, unless you have plans to clone said vm to be run on multiple machines or you want it to be portable to be moved to other servers in the future. I couldn't disagree more with this. I run my primary workstation OS as a VM on unRAID and it's pretty much the only VM I run. Everything else is in Docker. Sure I have other VMs on there for testing purposes, but in a production scenario, this is by far more ideal than running a physical desktop. Why? 1) If my OS ever became unbootable due to a Windows error, my data isn't trapped inside. I could still mount the storage the VM used without having to resort to booting off a boot disk or anything like that. 2) If I want to upgrade my hardware, I don't have to reinstall my guest VM's OS like I would with a physical machine. The VM always sees the same emulated chipset (i440fx or Q35) and so even if I switch from ASRock to ASUS, my VM and the drivers it uses stay intact. I just uplift my storage devices and USB stick, install them on new hardware, and I'm rocking and rolling! 3) Running VMs on the same system as a my NAS gives me SAN like performance for applications because I don't need to traverse the 1gbps network. All networking is internal to the host, which means I can use SSDs for my cache pool and write data to them over SMB at their full potential speed, not bottlenecked by copper wire and a 1gbps network. 4) Backing up my VMs is easy as they are in a single file (vdisk). I can also make copies to test new builds before I commit my changes. 5) I can run a VM as my desktop / workstation while also running the NAS underneath and Docker containers as well to add complimentary services to my system without bloating my desktop OS. These are just a FEW of the benefits of a single VM use-case scenario that I came up with off the top of my head. There are many more.
May 12, 201511 yr Man, I'm awful at this explanation stuff. I mentioned virtualbox to indicate what level of experience of virtualization I have. I don't have a virtualbox image of the development environment I would like to setup. It's sounding like just running a KVM instance of ubuntu server (or something to that effect) might be what is easiest and gives me the most flexibility. Either way, benchmarks will be the order of the day when I get unraid 6 up and running. Sounds good to me. Maybe others can chime in with their thoughts. I couldn't disagree more with this. I run my primary workstation OS as a VM on unRAID and it's pretty much the only VM I run. Everything else is in Docker. Sure I have other VMs on there for testing purposes, but in a production scenario, this is by far more ideal than running a physical desktop. Why? 1) If my OS ever became unbootable due to a Windows error, my data isn't trapped inside. I could still mount the storage the VM used without having to resort to booting off a boot disk or anything like that. 2) If I want to upgrade my hardware, I don't have to reinstall my guest VM's OS like I would with a physical machine. The VM always sees the same emulated chipset (i440fx or Q35) and so even if I switch from ASRock to ASUS, my VM and the drivers it uses stay intact. I just uplift my storage devices and USB stick, install them on new hardware, and I'm rocking and rolling! 3) Running VMs on the same system as a my NAS gives me SAN like performance for applications because I don't need to traverse the 1gbps network. All networking is internal to the host, which means I can use SSDs for my cache pool and write data to them over SMB at their full potential speed, not bottlenecked by copper wire and a 1gbps network. 4) Backing up my VMs is easy as they are in a single file (vdisk). I can also make copies to test new builds before I commit my changes. 5) I can run a VM as my desktop / workstation while also running the NAS underneath and Docker containers as well to add complimentary services to my system without bloating my desktop OS. These are just a FEW of the benefits of a single VM use-case scenario that I came up with off the top of my head. There are many more. I was talking about vm vs baremetal in general. Of course there are many use cases of VMs once you get things going. Btw I think #1, #2 and #4 are due to 'portability' which I've already mentioned. I guess I was also thinking from the conventional 'plain vmware platform beneath' point of view (vmware, or just base xen/kvm underneath). Unraid provides disk redundancy and preset dockers environment that enables #3. #5 is the benefit of partitioning a single system and running smaller instances of other crap.
May 12, 201511 yr I was talking about vm vs baremetal in general. Of course there are many use cases of VMs once you get things going. That's what I thought you were talking about. I was replying saying that I can find plenty of reasons to use just one VM with unRAID on ANY capable hardware platform. Btw I think #1, #2 and #4 are due to 'portability' which I've already mentioned. Not sure how you're connecting #1 to portability as I'm not talking about moving the vdisk to an alternate host, although that is a viable option. My point was simply stating that if a VM is used as opposed to a physical machine equivalent for any OS, you are not dependent on that OS to access the data inside it. You could mount the vdisk to get at the data as opposed to having to go through some rigmarole. I guess I was also thinking from the conventional 'plain vmware platform beneath' point of view (vmware, or just base xen/kvm underneath). Unraid provides disk redundancy and preset dockers environment that enables #3. unRAID 6's use case as a plain-jane virtualization platform is really not even in my top 5 of use-cases. I don't see folks building virtual servers with remote graphics connections for management as a primary way of supporting service-driven applications. I see Docker as that platform. But when you want a more tunable operating system environment for custom applications / use-cases, VMs can offer a simpler way to do that than containers, which aren't really designed to be tuned much post-build. And benefit #3 isn't about data redundancy, it's about performance. It's about being able to read and write data to a system at the full speed of the disks, without being bottlenecked by the network layer like with a traditional NAS. Imagine editing photos or videos of large size on a pool of SSDs. Over a traditional network, this performance would be abysmal unless the user purchases 10gbps hardware. With unRAID 6, you can get full performance without the expense of a SAN. #5 is the benefit of partitioning a single system and running smaller instances of other crap. I agree. The benefit of partitioning is huge, but it's amplified when you're able to partition resources towards three various purposes on a single host: storage and protection of data, serving of applications, and usable workspaces. Apologies if this is coming of argumentative. It's not intended. It's intended to open folks eyes to see beyond the traditional virtualization world and into the power of localized virtualization for consumer / professional uses. I could see unRAID being used on desktop PCs with no intention of utilizing it as a NAS. Turn an ordinary workstation into a multi-purpose computing powerhouse. Same goes for something in the living room as an all-in one media player/server/storage system.
May 12, 201511 yr @jonp: I agree, all good points. Once we open up the virtualization tap, the sky is the limit. As for unraid's case, it becomes really special once it incorporated kvm and docker natively. In the past (pre-v6) i've never thought I'd install a distro called UNRAID for anything other than NAS purpose. I think they may need to rebrand or spin-off a separate product to maximize their market potential.
Archived
This topic is now archived and is closed to further replies.