Google creates two billion Linux containers a week which astute readers will realize implies that they can be created much more quickly than a VM. That is indeed the case, these Linux containers are very similar to Solaris Zones, BSD Jails and other similar ways of sharing parts of an OS across multiple isolated applications as opposed to VMs in which each machine has it's own OS. Even with prebuilt images it is orders of magnitude slower to create a VM than to simply create a new container. With the involvement of a startup called Docker, Google has really changed how they handle their systems; read about the impacts at The Register.
"That tech is called Linux Containerization, and is the latest in a long line of innovations meant to make it easier to package up applications and sling them around data centers. It's not a new approach – see Solaris Zones, BSD Jails, Parallels, and so on – but Google has managed to popularize it enough that a small cottage industry is forming around it."
Here is some more Tech News from around the web:
- Servo Stock, The Future Of 3D Printers @ Hack a Day
- Gigabyte, Asustek step up gaming notebook competition against MSI @ DigiTimes
- How to Sort and Remove Duplicate Photos in Linux @ Linux.com
- Is Emulation the Best Feature of the Nvidia Shield @ eTeknix
- Netgear R6300 802.11ac Smart Wi-Fi Router @ Kitguru
This might not be a good time
This might not be a good time to start investing in VMware stock.
It might be a good time to
It might be a good time to sell VMware stock. The point of containers is to avoid the hypervisor all together. The Linux kernel can handle directly all the discreet namespaces, and control groups for each container.
It might be a good time to
It might be a good time to reread the 1st post
The next time it takes all
The next time it takes all stinking weekend to get Windows installed, activated, updated, and apps installed, think of Google installing Linux 2 billion times a week. :o)
Most likely the customer
Most likely the customer still has to configure an OS image and all the system settings, permissions, and configurations and the install the application software, but once that is done and containerized, it is similar to an image. So any OS will take some time to configure, and the Image/container can be made, and loaded quickly.
There was an old IBM competitor that had a system and hardware architecture to do this on the fly, and much of this, container management as it is called, was handled by the hardware as part of the CPU. Why Unisys has not dusted off and implemented the old Burroughs stack machines architecture and taped out a version using the current fab process node, and produced a server SKU, I do not know. But why implement containers in software when there is an CPU architecture for running these containers/Objects(object oriented code) that is still advanced to this day. If anyone has the money to have this done it is Google, and Unisys is the old Burroughs and Sperry companies which merged in 1986. Burroughs Extended ALGOL, was one hell of a powerful programming language, and the Burroughs MCP/OS was written in this high level language, and code on the system could be modified on the fly, but always in a stack under the ultimate control of the MCP.
“Burroughs Large System architecture which was fundamentally a hardware architecture for Object-oriented programming, something that still doesn’t exist in conventional architectures.”(See 1)
(1)
http://en.wikipedia.org/wiki/Burroughs_large_systems