Strategic Thinking (was: Re: Speculative: Rust for base system components)

[Alexander Leidinger]

Quoting Poul-Henning Kamp <phk at phk.freebsd.dk> (from Fri, 04 Jan 2019  
22:42:31 +0000):

> --------
> In message <201901042219.x04MJf4w085379 at pdx.rh.CN85.dnsmgr.net>,  
> "Rodney W. Grimes" writes:
>
>>> ... and RPi's, access-points, NAS devices, routers, televisions,  
>>> photocopiers,
>>> sewage-treatment-plant-monitoring, high-voltage-switching,
>>> stock-trading, air-traffic-control, scientific super-computing,
>>> antiproliferation-monitoring, laptops, desktops and ...
>>
>> As far as I am concerned Linux can have the datacenter...
>> I find this list much more interesting :-)
>
> Me too.
>
> Data-centers are booooring!

Which means that x developers with commit bits in FreeBSD are free to  
develop whatever they want.

This does not mean that all users of FreeBSD agree.
This does not mean that all developers with commit bits in FreeBSD agree.

Do you want to limit what y developers with commit bits in FreeBSD are  
working on?

 From what I hear here I get the impression that there are people  
which want to limit that y developers want to explore the benefits of  
feature A. Nobody told so war we have to import anything into base  
yet. The initial request was to get an idea about opinions. Nobody  
told we have to rewrite the kernel in rust, there were infos that  
there may be benefit in having parts of it in rust, which can be  
explored e.g. in ports. Nobody asked to replace a critical boot time  
component. As we are not a company were the people are paid to work on  
specific items (yes, there are people paid to work in parts, please  
forgive me that I don't count them here... we don't talk about them  
doing this work), we can not really tell that this takes away  
development resources away from other work (those developers may not  
work on something else, or they may work on something which is not  
"strategic").

And if you really think that containers (in whatever color...  
kubernetes, docker, "jails" or whatever) are only datacenter tools...  
well... have again a look at NAS devices, laptops, and desktop systems  
(and whatever).

I don't have a datacenter at home, but I use a lot of containers at  
home. I use them in the "jail"-color (every service his own jail, I  
even have a desktop-setup-in-a-jail...). I don't use them as is, I use  
tools. ezjail, iocage, whatever color you want. Would openstack be  
overkill here? Maybe. Maybe not. Would I give it a try if we would  
have openstack in ports in my basement? Yes I would -- why should I  
limit myself to linux to have a look at openstack/kubernetes/docker...  
we have the infrastructure to make it possible (I let it up to you to  
decide if we have a better infrastructure/base for this or not).

I expect in the long run virtualisation and containers arrive in a lot  
of places, even in those you have listed above as not boring. There  
are benefits in the upgrade path, there are benefits in handling  
dependencies (compared to an one box does everything), there are  
benefits in the security area (yes, we have capsicum which addresses  
some aspects, but not all as if each part runs in it's own jail).

FreeBSD comes from the "power to serve" area. You can off course tell  
that access-points, NAS devices (which also exist in datacenters...)  
and routers are "serving", but datacenters are the traditional area of  
"the power to serve". Basically if you tell that datacenters are  
boring, you tell that we shall turn around and that e.g. the CDN of  
Netflix is not the area we want to target (I would not agree that this  
CDN is some sort of NAS, for me this is more like a  
web-/ftp-/<protocol_of_the_day>-server, so something which resides  
traditionally in a datacenter).

Bye,
Alexander.

-- 
http://www.Leidinger.net Alexander at Leidinger.net: PGP 0x8F31830F9F2772BF
http://www.FreeBSD.org    netchild at FreeBSD.org  : PGP 0x8F31830F9F2772BF
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 819 bytes
Desc: Digitale PGP-Signatur
URL: <http://lists.freebsd.org/pipermail/freebsd-hackers/attachments/20190105/318b3cae/attachment.sig>