3 ms·
> A major point of containerization is more efficient resource utilization It's actually "more efficient resource utilization per instance". Perhaps, with incr
by pritambaral 10y ago
> A major point of containerization is more efficient resource utilization
It's actually "more efficient resource utilization per instance". Perhaps, with increased capability comes increased utilization. While before one would run only one or two VMs and try to run all their apps within that, now one can run many containers.
----
That, and shareable memory across host and guests.
But I guess that works only if the host and guest run the same kernel. Docker and vagrant on OS X still have to dedicate some RAM to the Linux VM running as docker host, and I don't know how efficient other facets of docker on OS X are; but on Linux, I find containerization's memory usage almost at par with running the apps directly on the host.
- debinguy 10y agoFor my situation I work for a large CDN and we have over 15k servers deployed globally and software I write has to run on all of them. This means various OS and kernel versions. This means in a given day I can work on 20+ different containers. The more time I spend spinning up or down containers to save system resources the more angry I get. I also travel a lot for work, so connecting to a cloud based instance from a plane is not possible, plugging into a power outlet in a plane seat is. For me the issue comes down to lack of choice. Previously lack of choice from Apple was not a concern because the choices I would make were inline with what they are offering. This is no longer the case. For those that asked, I am replacing 10 years of using MBP's with a System76 laptop. I don't have it yet so I can't comment but I am excited for the change as have been growing more and more frustrated with Apple.
- corv 10y agoThe System76 Oryx Pro looks very interesting.
- pritambaral 10y agoAside, but you say "various OS and kernel versions". One'd have to use virtualization in order to run a kernel different than the host. I don't think containerization can help you there, at least as far as spinning up an image with a different kernel is concerned.