Never expected this from the reg: release the commentards.
lmcfy (dead?) or lxd would have been better examples for linux.
Containers are great. Without them, the cost of shipping materials and goods around the world would no doubt be considerably higher. It's economical because intermodal container dimensions are clearly defined by an international set of standards (ISO 668:2013 and ISO 1496-1:2003 if you must know) that allow containers to be …
I'm not so sure that uptake will increase drastically in 2016. Containers are a great buzzword at the moment because of people like Amazon but in reality most people don't need them. Also most don't have the type of code necessary to run in a containerised manner and neither do most people want or need to rewrite that code. There is a weird belief driven my the media that containerisation is somehow necessary for CI/CD and DevOps which is simply not true. With the right configuration there are numerous options for these both on internal infrastructure and in the public cloud. Containerisation is designed around micro services architecture, and if you do some reading in Docker forums it's very clear that most developers don't understand this and are reluctant to do the necessary coding. A large percentage of posts are asking how to run stacks of software or traditional software in a container. Add to that the many and various security concerns of the Docker ecosystem which the security staff of most organisations would have a fit over and I don't see mass adoption any time soon, frenzied media or no.
I see so many DevOps (urrg, horrible word) articles talking about containers like it's the best thing and everything should be containerised because it's, like, so hot right now, yaa.
If it's not got your application code in, then in my view, it shouldn't be a container. For me the benefit of containers is being able to spin up identical worker instances of an application, rapidly, that can sit behind a load balancer and buzz away and die as quickly as they were born. The idea of enforcing a maximum container age of X hours/days is quite appealing too, frequently recycling. Granted many applications would not fit this architecture, then in which case don't Contain it.
There's such a lot of "legacy" software out there that doesn't really run in a "container" - or doesn't run well.
I always ask my co-workers what they want to achieve by "moving everything to containers" - or if it's one of these self-serving things you do to be able to say "Yeah, we did it (fist bump)".
Our customer's stuff sometimes runs months or even years in the same configuration.
No need to "spin" up more or less of it (they wouldn't pay more - nor do we want to charge them less).
There's also (to my knowledge) the mostly unsolved issues around storage (needs object storage or cluster-filesystem), overall persistence, user-management etc.
Very few companies can break-down and refactor any problem they face in such a way that it fits into their current "virtualization gospel". Most of our customers don't know why they should, don't know how the could and don't want to pay for it anyway.
We have one customer who is apparently trying to refactor their main application from a asp/dot net thingy into a linux container thingy. A project that is supposed to take two years and god knows how many man years. That's a bold move and I have nothing but respect for them (if they achieve it...).
But few people would and could do such a thing.
Though, arguably, it would be good if people had software written well enough as to allow such a feat without superhuman strength...
Totally depends on how you use them - Containers are for more than just micro-services (and the like).
I'll get some flack for this from both sides I'm sure, but there is nothing wrong with running a container in the same PID and Network namespace as the host or even running systemd etc in a container. The idea that a container can only run a single process is religion, not science.
No it's more that you gain nothing by doing so while losing a whole bunch of great features that virtualisation offers. People doing this sort of thing are generally trying to be trendy rather than meeting a real requirement. If you need systemd you're almost always better served with a server than a container.
Biting the hand that feeds IT © 1998–2019