[BlueOnyx:27341] Re: Celebrating 25 Years Since Cobalt Networks' IPO

[Rickard Osser]

Hear, Hear!

https://en.wikipedia.org/wiki/Cobalt_Qube

The Qube2700 was released in 1998. The Gen 1 and 2 (Qube2700/RaQ and
Qube2/RaQ2)
where both MIPS based and really fun to play with.

I bought my first Qube2700 in November 1998 after talking with Desa
Zraick (US) in September on the phone
from Sweden for about an hour. I paid for the first one using a CC.
Serialnumber (#77)...
Anyway, we made an agreement that I could distribute Cobalt in Sweden
(Nordic) and that 
the next orders would be on a credit-line.

My background at the time was:
Working in the IT-business since 1986, started with Apple and Mac (yes,
Apple II techie was my first job). Anyway, after using PDP/VAX and
various variants of Unix
I got into Linux in 1992. Around 1997/98 I was starting to program a
Web-gui to
manage Samba/NFS/Mail/DNS/DHCP and other necessary stuff you would need
in
an office appliance, I didn't get very far until I saw a small blurp on
the LWN i think:

Do you know what  David S. Miller is doing right now?

And then there was a short article on his porting of Linux to Mips for
Cobalt and
an explanation of the Qube2700... 

Well, I don't like to re-invent the wheel so I stopped my own
development and 
jumped on the Phone to talk to Desa.

Why did I follow Davem? He ported Linux to Sun which I was using at one
of my
consulting customers. What a difference in using the archaic Solaris
1.1 (SunOS 4.1.3) and a 
nice modern RH Linux port on the same hardware.

Anyway, thanks to George and Carley Korchinsky for a very nice and
fruit-full few short
years but fun. I won't forget to mention our own Taco (who worked in
Holland with George and Carley)
and all the other Cobalters like Gordon Garb (Fun Apple stories) and
Nicholas White (Many happy beers!)
and a lot of others.

For my own amusement I created some software packages for the Qube and
RaQ line, SSL for use with
both the GUI and virtual-sites, Cobalts own was only virtual-sites and
couldn't be sold outside of the US. The difference
to secure the GUI as well as the sites gave me quite a few customers in
the US aswell.

I also created a WIFI-Accesspoint integrated into the Qube, the
AirQube.
And then of course a host of applications running on top of Sausalito
using it for management.
I created a Door-access Appliance (which could be run in HA) which I
sold a few of. This version was based on Aventurine and
the payback to Michael was the first Qemu implementation and as faster
more secure way of uploading ISOs.

Michael, did you ever get the HA-code?

We still host LXCs using the same base with ARX from Assa Abloy.

I'm still creating appliances but today most of the stuff I do is MCU
(RP2040/Arduino) and Rpi.
I have made appliances based on RedHat (RHEL/CentOS/RockyLinux) as well
as Suse Linux.

As long as Michael will be around doing BQ, I will be lurking along and
put in a piece here or there...

Thanks Michael for all these years and to the others before, Hisao,
davem, Tim Hockin and many more...

Your humble servant,
Rickard


On Fri, 2024-11-15 at 14:59 -0500, Michael Stauber via Blueonyx wrote:
> Hi Taco,
> 
> > Let’s celebrate not just 25 years since the IPO, but also the
> > vibrant 
> > community and projects that continue to carry the Cobalt torch
> > forward. 
> > Here’s to the past, present, and future of Cobalt’s remarkable
> > legacy!
> 
> Hear hear! A toast to that! o7
> 
> Man, it's been 25 years already? Seems like yesterday.
> 
> I recall getting my first (rented) Cobalt RaQ3 in December 1996 and 
> kinda fell in love with it. I already had been dabbling with Linux
> for a 
> few years, so I felt right at home. And the web based GUI interface? 
> That was something else - at least for that time and age.
> 
>  From a purely technical point of view I'm still in awe with some of
> the 
> design aspects that went into the Sausalito architecture early on,
> which 
> got finalized with the late Qube models and the RaQ550's.
> 
> Sure, in some parts you could see that things had been rushed a
> little 
> and there were the beginnings of some more advanced concepts and 
> functionality that never made it into the final products.
> 
> But the security concept and the general architecture was rock solid,
> contained a lot of foresight and a modularity that made it easy to 
> expand on it and to keep it current.
> 
> Some parts of the code are THAT solid that they still remain in usage
> 25 
> years later with next to no or miniscule changes. Some of that is
> still 
> in the backend of the GUI and some of that original code is still in
> the 
> build environment with which we build RPMs and PKGs. You just go to
> the 
> respective level of the code tree and run "make rpm" or "make pkg"
> and 
> it spits out RPMs or PKGs that are ready to be distributed.
> 
> That this still works decades later is an incredibly testimony to the
> skills of the original designers. I raise my glass to them for a job 
> well done.
> 
>  > However, as history shows, the motivations behind Sun's
> acquisition
>  > went beyond mere competition.
> 
> Indeed. As chance would have it I worked at the Langen office of Sun 
> Microsystems in Germany at the time leading up to and during the buy-
> out 
> of Cobalt. I was moonshining Solarspeed.net during the nights and by
> day 
> I was a support engineer for the traditional "big iron" servers of
> Sun.
> 
> The corporate culture of Sun Microsystems was somewhat special and
> the 
> weekly company wide Scott McNealy emails to all employees were a
> thing. 
> I do recall him frequently dissing Linux, even though one would not
> need 
> a perfect hindsight to predict that it eventually might become a
> serious 
> competition to what Sun had on offer. And it was also clear that the 
> company was not well positioned to reap any benefits in that area.
> 
> Eventually that realization must have sunk in with Scott McNealy as 
> well. Probably because of a first (abortive) move that Microsoft had 
> made in that direction, too. So out of the blue a weekly Scott
> McNealy 
> email announced not only the acquisition of Cobalt Networks, but also
> that the company was now fully embracing Linux. The intent certainly 
> being to capitalize on the Linux experience and market penetration
> that 
> Cobalt Networks already had established.
> 
> I do recall another funny email a few days later, straight from 
> corporate HQ in the US. It went along the lines of: "Please stop 
> pestering the Cobalt guys with inquiries! Sun has 35k employees and 
> Cobalt has around 200. They get nothing done if everyone continues to
> bother them with requests for information!"
> 
> But in the end even the acquisition of Cobalt Networks couldn't save 
> Sun. The burst of the dotcom-bubble contributed to it, but also that 
> their marketing was ill suited to reap any meaningful benefits from
> the 
> Cobalt Networks merger.
> 
> A marketing department that was used to deal with banks and
> government 
> departments was accustomed to sales moving at glacial speeds at best.
> Finalizing the sale of a E16800 flagship "big iron" server back then 
> might easily have taken two to three years from start to finish.
> Getting 
> calls from ISPs that each wanted a truckload of RaQ550s by tomorrow? 
> That just didn't work so well as nothing at Sun (but the rumor mill) 
> could move *that* fast.
> 
> Without the burst of the dotcom-bubble it *might* eventually have
> worked 
> out (I have my doubts), but as it was: It didn't.
> 
> Burning through cash like mad Sun had to swing the axe and
> restructure. 
> The first victims were most if not all the new acquisitions. Even
> that 
> couldn't save the company from having become not only complacent, but
> also arrogant towards peers and clients. A few years later Oracle
> bought 
> the scraps - lock, stock and barrel. One of the last vestiges of the 
> company is the corporate sign in front of the Facebook US
> headquarters, 
> whose backside still carries the original Sun Microsystems writing
> that 
> once was facing outwards instead.
> 
> 
> Closing thoughts:
> ==================
> 
> I never found out who was behind the move that lead to Sun
> Microsystems 
> open sourcing the Sausalito code. Whoever that benefactor was:
> Without 
> that move we would not be where we are today and neither BlueOnyx nor
> BlueQuartz or TLAS would have existed in first place. And tens of 
> thousands of former Cobalt users would have been left stranded in 
> "vendor lock-in" without a clear migration path onto other platforms.
> 
> So in all this I want to take a special moment to thank your unknown 
> benefactor with sincere gratitude and admiration. Thank you!
> 
> In the early years of the release of the Cobalt sources there were
> quite 
> a few competing projects which tried to "make it work" and only our 
> friend Hisao Shibuya really had the expertise and the tremendous 
> tenacity to make it work and to be able to build not one, but two 
> success stories out of that: TLAS for his then employer and the open 
> source BlueQuartz as his own personal project.
> 
> This work was made extraordinarily difficult by there not being a 
> documentation for the inner workings about the heart and soul of 
> Sausalito: CCEd, CODB, CCEWrap and how these interfaced with PHP,
> Perl 
> and even Python (although Python wasn't used in any meaningful way).
> 
> There were also two competing standards for writing GUI pages: The
> more 
> modern UIFC and a legacy method that had been used prior and was (for
> quite some time) still used in some older pages like the DNS
> management.
> 
> As it was: There was only sparse original documentation that had been
> scrounged together over the years. But that wasn't always helpful, as
> some of that clearly had not been written by engineers involved in
> the 
> development, but perhaps been done by interns or marketing guys.
> 
> Any early adopter who sought to benefit from the open sourced code
> had 
> to start with a profound knowledge of the original Cobalt interface
> and 
> had to reverse engineer and analyze the code to piece together the 
> understanding of the "big picture". Hisao did that and managed to
> gain a 
> profound comprehension of the inner workings of the architecture that
> allowed him to succeed where others didn't.
> 
> Therefore: My heartfelt thanks to Hisao Shibuya, who carried the
> torch 
> for many years and allowed us to kickstart BlueOnyx from the legacy
> that 
> he had preserved and expanded.
> 
> Lastly: We wouldn't be still around if it were not for each and any
> of 
> you: May that be the old timers from the Cobalt days that stuck
> around, 
> or the many others who discovered us since then and stayed with us 
> through all those years. Many thanks for your support and your 
> confidence in BlueOnyx!
> 
> 
> Where do we go from here?
> ==========================
> 
> As you all know, BlueOnyx is constantly evolving and new features are
> getting added. Not too long ago we integrated Radicale CalDAV/CardDAV
> and web based FTP access is now (again) built straight into the
> BlueOnyx 
> GUI. We ditched the clunky Cobalt Migration Utility and released both
> Easy-Migrate and Easy-Backup to make migrations and backups easier
> and 
> more reliable.
> 
> Over the years a long list of features has been added to BlueOnyx -
> way 
> beyond what was included into the original Cobalt RaQs. And that list
> will grow even further as time moves on:
> 
> See: https://www.blueonyx.it/features
> 
> This year also saw the release of Aventurin{e} 6110R, which is the 
> newest release of our Container and VM virtualization platform. It is
> also based on BlueOnyx and can be used to virtualize not only
> BlueOnyx 
> instances, but also a large selection of other Linux distributions.
> 
> See: https://www.aventurin.net/
> 
> Finally: RedHat yesterday announced the release of RHEL 10 Beta, so
> we 
> can expect to see a RHEL10 and (of course) also an AlmaLinux 10 and 
> RockyLinux 10 sometime in the 2nd Quarter of 2025.
> 
> The release of the EL 10 based BlueOnyx 5212R is targeted for the
> summer 
> of 2025, but this depends on the general availability of AlmaLinux 10
> / 
> RockyLinux 10 and how fast we can port our code to those. Some 
> preliminary work on that has already started and will continue on a 
> moderate pace in the coming months.
> 
> With that: Many thanks to you all and have a great weekend!
> 

-- 
Bluapp AB
Rickard Osser
CTO
Solberga Ängsväg 3
125 44 Älvsjö
Sweden

Web: http://www.bluapp.com
Mail: rickard.osser at bluapp.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.blueonyx.it/pipermail/blueonyx/attachments/20241115/abf609c5/attachment-0001.html>