[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [cobalt-developers] 2.2.14 kernel
- Subject: Re: [cobalt-developers] 2.2.14 kernel
- From: Tim Hockin <thockin@xxxxxxxxxx>
- Date: Fri May 26 14:09:21 2000
- Organization: Cobalt Networks
Dennis wrote:
> Keeping the raq3 OS up to date (you are regularly several months behind..ie
> now releasing 2.2.14 when the world is on 2.2.15 ).....
As I have said in other posts - we are not bleeding edge wrt kernel for
several reasons. The least of which is time. (this all relates to
RaQ3). We aren't shipping 2.2.15 because 2.2.15 has several
instabilities that we need to work out. Add to that our thorough QA
cycle (which has turned up more than a few bugs in Linux). I don't
think you have to worry about keeping Raq3 current - it is far less
difficult than maintaining a whole seperate architecture port. When
Cobalt made 2.0.x run on Mips - there was no /usr/src/linux/arch/mips
tree. We helped make it.
Add to that this guiding principle, which has been very successful for
Cobalt: It is better to be stable and reliable than bleeding edge.
Also: solve 90% of the problems for 90% of the customers first, then
work on the rest.
> Regarding the RaQ3, its hard to recommend a product that 1) costs more than
> an equivalently equipped PC, 2) is underpowered and 3) is several releases
> behind and 4) may not support the next major release of linux based on the
> history of the raq2. Its a thought process thats building, in case you
> werent aware of it.
I appreciate your candidness, let me address these concerns.
1) We don't sell PC's. We sell server appliances. Show me a simlarly
equipped PC that comes with a full management software suite, a full
range of support options, pre-configured, fits in 1RU and draws less
than 40 watts of power. You may find a system that comes with some of
those, but not all of them. what Cobalt offers is a SYSTEM. Hardware,
software, support, and solutions. I honestly believe the TCO of a RaQ3
is less than that of a beige-box linux system. If I didn't, I wouldn't
be working at Cobalt.
2) We don't sell RaQ3s as compute servers. We never advertised them as
compute servers. We advertise them as email and web (etc.) servers. As
such, they are far from being overpowered. They can fill both 100
megabit networks without blinking. They can fill even more than that.
90% of web sites don't need dual 800 MHz processors. Those that do,
well, we can provide load-balanced solutions that not only out-perform
an SMP 800 MHz, but are more reliable and still easier to manage.
3) Again, stability over bleeding edge. 90%+ of customers never telnet
to their RaQ3, so they don't know (or care) that we don't have a Red Hat
X.x compatible upgrade for their system. Those who do care can get the
Red Hat packages they need from us (we do track the important bits) or
from Red Hat. being several releases behind is a side effect of what
our product is - it is not a PC - it is an appliance. You can look
under the hood, but you don't have to. You can tinker if you want, but
you don't have to.
4) RaQ3, being intel compatible, will always support the next-release of
linux. Just because we don't ship a new CD every 6 months to upgrade
all your software (needlessly) doesn't mean you can't do it. If you
have a legitimate need to upgrade certain components, please do. But
unless we (CObalt and whomever) have some partnership arrangement (and
therefore are aware of your needs) you can't honestly expect us to
support it, can you?
The people on this list are the small percentage of Cobalt customers who
do things with products that we didn't really expect. That is why
several developers monitor this list - to help you all out. It also
gives us a chance to learn how to make the products better.
I appreciate your comments, and I want to assure you that Cobalt is
listening, and we are doing what we can to make the products more
attractive for _everyone_. Remember that other 10% I mentioned...
Tim
--
Tim Hockin
Software Engineer / OS Engineer
Cobalt Networks
thockin@xxxxxxxxxx