Saw this day coming years ago...

http://www.zdnet.com/article/sun-set-or ... uct-lines/ says:

Oracle's 2009 acquisition of Sun, which gave the company Solaris and SPARC, was a terrible move from day one. The rise of commodity Linux x86-based servers insured that Oracle buying Sun would be an all-time awful technology merger and acquisition.

Indeed, you'd be hard pressed to find anything that went right with Oracle's $7.2 billion purchase of Sun. Simon Phipps, former Sun open-source officer and managing director of Meshed Insights, gave a long, painful list of all the many once popular Sun programs that Oracle wasted. Among them are:
• Java was described as the "crown jewels," but the real reason for buying Java SE -- trying to sue $8 billion from Google -- has failed twice.
• Ellison said Java's role in middleware was the key to success, but Java Enterprise Edition (EE) is now headed to a Foundation.
• Bureaucracy over MySQL security fixes led to a decent portion of the user community going over to Monty's MariaDB [MySQL] fork, enough to start a company around.
• Ellison said he would rebuild Sun's hardware business, but its boss quit a month ago and the team behind it was part of the lay-off.
• Despite Scott McNealy's (former Sun CEO) understanding that Solaris had to be open to win in the market, Oracle hyped it up and closed it down. The result was this week's layoffs, foreshadowed extensively in January.
• Oracle renamed StarOffice (OpenOffice) and announced a cloud version, but it couldn't make it fly. Sensing the impending EOL of the project and alienated by heavy-handed treatment, the community jumped ship to LibreOffice.

When all is said and done -- and now all has been said and done -- Oracle buying Sun was a waste of money for Oracle and a waste of once valuable Sun technologies. Great moves, Ellison. Let's see if you can continue your good work with taking Oracle to the cloud.

RIP, Sun. Their hardware and software was the cat's meow from the mid 80's till the mid 2000's. We all saw how Intel was going to undermine the hardware. The mid 90's Pentium Pro was a shot over the bow, the Core and then especially the Nehalem of 2008 was the coup de gras. We all hoped the Sun products could find a niche role running on Intel CPUs, much like Apple made the transition from Power to Intel, but then Larry, Prince Of Darkness ( http://www.urbandictionary.com/tags.php ... %20ellison ) came along. IBM made a solid pitch, one that seemed so much better for the future of Java in particular, but all that did was engage Larry's testosterone-addled noggin that is wired to win at all costs. Well, now the costs are plainly visible.

Software wise, perhaps Solaris could survive with a solid set of customers and apps, but it never was going to thrive with Linux around.

Ahh, the sweet 80s. My grad school had a cluster of Sun 3 workstations using this thing called NFS that let you share files over the network as if they were local. A sweet BSD Unix so much faster than the time-sharing VAX 11-780s that we used to use. And we had a row of SGI workstations too, with our jaws dropping because it could rotate a light-shaded rendering of a Rubik's Cube in real time. And a nice trip to get trained up on those new-fangled Cisco routers and those funny Internet addresses. So many dotted decimals, and why do we need those netmasks again?!?

Seems like the so-called four horsemen of the internet boom ( Cisco Systems, EMC, Sun Microsystems, and Oracle ) aren't handling middle age all that well, sigh..

Queue the sequence at the end of "Patton": All glory is fleeting...

My brother worked for Sun after college, at that giant new campus they built in Burlington, MA, that they never managed to fill.

I still drive by it when I'm back in Boston, and it's still kinda sad.

vikkyvik wrote:

My brother worked for Sun after college, at that giant new campus they built in Burlington, MA, that they never managed to fill.

I still drive by it when I'm back in Boston, and it's still kinda sad.

Back when that building was going up I was contracting with DEC / |d|i|g|i|t|a|l| / whatever. They had a bad case of Sun envy. It didn't work out well for them.

A lot of both of those groups ended up working for Red Hat Linux in Westford. It didn't seem like the smart move at the time, but most of those people are doing quite well there.

Kinda surprised RH didn't make a play for the Sun campus. It'd be a lot more appealing to those within the 128 belt than their current locations.

I've never used Sun hardware (well I kind of remember having used a workstation once but I don't know where and why), however a few years ago I discovered ZFS. I'm running an OpenIndiana (OpenSolaris derivative) server with this gem of a filesystem and all my data on it (which is a lot of data, the thing has 57 hard drives for storage). I hope Solaris can be opened again so that ZFS advances can be freed. I'm obsessed with data integrity, running ECC memory even on desktop computers, so no other filesystem will do it for me.

Say NICE things about VAX 11/780's.

My wife worked for Sun as an executive admin for years, and the company was just a basketcase; every quarter there were rumors of more reductions in force, people she'd worked with who seemed safe would suddenly see their jobs disappear, etc. All while she was in an admin role making north of $60K/yr with a CORNER OFFICE ALL HER OWN in a high rise in the Biltmore area of Phoenix.

She finally got the axe in 2009 and moved on to bigger and better, throwing lifelines to former Sun people wherever she could, because Oracle wasn't much better about their random re-orgs and RIFs.

Airstud wrote:

Say NICE things about VAX 11/780's.

Great for their time and their genre, mini-computer. Home system for the great BSD Unix releases.

Preferred having my own PDP-11/34 before it and my own Sun-3, MIPS and then Alpha workstations after it.

Aesma wrote:

I'm obsessed with data integrity, running ECC memory even on desktop computers, so no other filesystem will do it for me.

And yet Intel decided with Desktop or Mobile platforms, your don´t get ECC. For a couple of years they didn´t even have an Embedded Plattform with ECC Support. Thank god they came to their senses with the E3xx series...
IF they now added Master/Checker support and TED/DEC ECC.... that be cool.

Back on Topic, i can remember the times when, in 1999 40% of all German Websites ran on a single Sun E6500 and while reliability wasn´t that good... performance was ok.

best regards
Thomas

tommy1808 wrote:

Back on Topic, i can remember the times when, in 1999 40% of all German Websites ran on a single Sun E6500 and while reliability wasn´t that good... performance was ok.

best regards
Thomas

Provides me with a flash-back to the days were all the RISC vendors wanted to have mainframe class products. Multi-board chassis, very expensive to develop and build, not much sales volume to amortize over. Not sure about the profit levels. However it usually did allow you to sell the storage arrays too, which were massively profitable. DEC used to take a hard drive it bought for $200, put it in a plastic sleeve, and sell zillions of them for $2000 each. 10x profit is great when you can get it! It worked pretty well, till EMC came along and ate everyone's lunch in the storage business.

Meanwhile, Intel just kept iterating on the single board computer till we got to the blade servers that rule the waves these days. And now, EMC is owned by Dell, an Intel vendor. Dell also owns VMWare, the premium virtualization product. If you look at things through a geezer's eyes, it almost looks like Dell is today's mainframe vendor, and clearly all the RISC players are dead or near dead. No more SPARC, no more PA-RISC, no more Itanium, no more Alpha, POWER is almost out of power, etc.

Oracle is SO hated by SO many, and with good reason.

I tend to use fairly cutting-edge companies when I need hosting or VPS services, and you should have seen how fast they moved to dump MySQL in favor of MariaDB once the latter was proven stable and assured to last - which in that case happened very quickly due to the particulars of the fork.

Free software types haaaaaaaate Oracle. Any time an Oracle property is forked, users coalesce around the new fork with amazing speed and unity. The only product I haven't really seen that happen with is Virtualbox.

EA CO AS wrote:

My wife worked for Sun as an executive admin for years, and the company was just a basketcase; every quarter there were rumors of more reductions in force, people she'd worked with who seemed safe would suddenly see their jobs disappear, etc. All while she was in an admin role making north of $60K/yr with a CORNER OFFICE ALL HER OWN in a high rise in the Biltmore area of Phoenix.

I didn't hear much about Sun in terms of hiring/firing. I only knew a few Sun employees and they were techie types back in the Internet boom era and were happy there. I have no idea how things went down after that time frame. If they're like a lot of companies, they probably off-shored a lot of things in a short period of time and then it all went to sh*t.

EA CO AS wrote:

Oracle wasn't much better about their random re-orgs and RIFs.

treetreeseven wrote:

Oracle is SO hated by SO many, and with good reason.

Oracle sucks, from what I heard. I know of at least two things they acquired in this area and the result was turmoil. Random re-orgs, all kind of management fads trying to "fix" things that weren't broken just so one VP could brag to another, sh*t being off-shored then falling apart yet management trying to make it look like it was going just fine, etc.

What a waste of resources

IIRC, Sun bought Silicon Graphics (SGI) which was a cool computer company too. Or Cray Research was also bought by somebody. Anyhow, they are all gone now.

I think the basic thing killing them was that consumer OS (including Mac OS or Linux) was sophisticated enough to multitask and do "big" workstation jobs. Mass-produced chips made workstations powerful enough, at MUCH lower prices than these niche players. The need for a $20,000 "graphics workstation" went away. As did the need for small servers. Now a $1,000 computer has more power than the hardware all these guys used to make. 16GB ram for one thing. You used to need to spend 1 million dollars to get a computer like that. Not true anymore. Plus, cloud services.

Revelation wrote:

and clearly all the RISC players are dead or near dead. No more SPARC, no more PA-RISC, no more Itanium, no more Alpha, POWER is almost out of power, etc.

There's one pretty massive exception, though: ARM.

Not a big player in the server space (yet), but it completely dominates the mobile and (upper) embedded segments.

By sheer numbers ARM-platform CPUs pretty comfortably outsell Intel and AMD combined.

Klaus wrote:

Revelation wrote:

and clearly all the RISC players are dead or near dead. No more SPARC, no more PA-RISC, no more Itanium, no more Alpha, POWER is almost out of power, etc.

There's one pretty massive exception, though: ARM.

Not a big player in the server space (yet), but it completely dominates the mobile and (upper) embedded segments.

By sheer numbers ARM-platform CPUs pretty comfortably outsell Intel and AMD combined.

Yes, ARM is thriving, mostly due to a totally different business model: readily license-able designs and intellectual property rights.

SUN stands for the Stanford University Network. The company was actually founded to establish the Stanford University computing network, but obviously the technology was broadly applicable in other settings and so SUN became quite successful and also became one of the founding pillars of modern computing and network design. By the time I arrived at Stanford in 1996, the entire on-campus network was run on an array of SUNSPARC servers named after the children of Leland Stanford (Marguerite, Leland, etc.).

But as all things must come to an end, SUN seems to be seeing its final days. I don't know what servers run the Stanford University Network now, but the university's Computer Science department will continue to be a major driving force in the computer and IT industry for the forseeable future.

DocLightning wrote:

SUN stands for the Stanford University Network. The company was actually founded to establish the Stanford University computing network, but obviously the technology was broadly applicable in other settings and so SUN became quite successful and also became one of the founding pillars of modern computing and network design. By the time I arrived at Stanford in 1996, the entire on-campus network was run on an array of SUNSPARC servers named after the children of Leland Stanford (Marguerite, Leland, etc.).

But as all things must come to an end, SUN seems to be seeing its final days. I don't know what servers run the Stanford University Network now, but the university's Computer Science department will continue to be a major driving force in the computer and IT industry for the forseeable future.

I agree.

Also, perhaps not as well known, but Cisco got its start at Stanford. The founders largely lifted the hardware design from the guys who went on to fund SUN, and the software from some of Stanford's in-house IT work. They were caught red handed but back then universities didn't have the kinds of commercial concerns they now do, so they got off with a slap of the wrist and some small commercial concessions. And the rest is history.

Revelation wrote:

Also, perhaps not as well known, but Cisco got its start at Stanford. The founders largely lifted the hardware design from the guys who went on to fund SUN, and the software from some of Stanford's in-house IT work. They were caught red handed but back then universities didn't have the kinds of commercial concerns they now do, so they got off with a slap of the wrist and some small commercial concessions. And the rest is history.

Of course I knew that. Google, too. I remember hearing how these two grad students over in Escondido Village (where the grad students live) had started this new search engine.

Stanford has never been too strict about students using university resources to start commercial concerns. The university has plenty of sources of money, including enormous donations from those same students once they are wealthy alumni.

DocLightning wrote:

Of course I knew that. Google, too. I remember hearing how these two grad students over in Escondido Village (where the grad students live) had started this new search engine.

The early search engine leader was DEC's AltaVista which was a lot of Stanford grads at DECWRL (Digital's Western Research Labs) in Palo Alto. The big difference was DEC thought it might sell some computers, but the Google guys (and their VC backers) were thinking on a whole different scale. And (again) the rest is history.

Revelation wrote:

POWER is almost out of power, etc.

Power is making a sort of comeback though, as they have become pretty competitive in terms of bang/buck, even if you buy IBM metal. Not for all applications, but when your application can make use of that enormous backend, applications tend to run quite fast.

best regrads
Thomas

I guess nobody has heard of ZFS then ah ah. A single zpool can store a maximum of 256 quadrillion zebibytes (2^128 bytes). As Bill Gates would say, we will never need more than that ! What is really great about it, and that a couple of filesystems are trying to copy, is that it has full checksums of the data. Every time you access a file it will check if there is corruption, and correct it live if found. OpenZFS is available on Linux and OSX on top of Solaris and OpenSolaris and derivatives.

Oracle bought Sun for Java and to kill other hardware manufacturers. By cutting Itanium support Oracle literally killed HP's high-end server market.

POWER was powered by IBM sales force and mostly by confusion at HP (Alpha, Itanium, X86-64, X86, Atom). At one point HP sales guys were confusing customers with too many products.

Aesma wrote:

I guess nobody has heard of ZFS then ah ah. A single zpool can store a maximum of 256 quadrillion zebibytes (2^128 bytes). As Bill Gates would say, we will never need more than that ! What is really great about it, and that a couple of filesystems are trying to copy, is that it has full checksums of the data. Every time you access a file it will check if there is corruption, and correct it live if found. OpenZFS is available on Linux and OSX on top of Solaris and OpenSolaris and derivatives.

I did a quick read of Wiki ( https://en.wikipedia.org/wiki/ZFS ) and it is impressive. I'm a big fan of snapshot cloning and replication, going all the way back to the Netapp products in the 90s. However the write up does imply a fairly deep learning curve and more than a few gotchas that might catch out people not willing to understand how it works and give it the proper provisions.

Thing is there are a lot of same but different technologies in the storage space. I cut my teeth on IBM LVM and JFS back in the old days, then Legato Networker, and more recently the mainstream Linux EXT4 and volume management tools. In an ideal world, it'd all shake out to a smaller number of best of breed solutions. In the mean time the approach the mainstream Linux distros take is to hide as much of the details of this stuff as possible away from the end user whilst leaving some room for future growth, which seems to be a good approach, IMHO.

dtw2hyd wrote:

Oracle bought Sun for Java and to kill other hardware manufacturers. By cutting Itanium support Oracle literally killed HP's high-end server market.

POWER was powered by IBM sales force and mostly by confusion at HP (Alpha, Itanium, X86-64, X86, Atom). At one point HP sales guys were confusing customers with too many products.

That's a fair rendition.

I still wish IBM had bought SUN. IBM has a tradition of supporting multiple different OSes and HW architectures. Java was an excellent fit for its enterprise business. But LPOD had a raging hard on for SUN and was not going to be denied, whilst IBM really had no choice but take a more sane approach to valuing SUN.

tommy1808 wrote:

Power is making a sort of comeback though, as they have become pretty competitive in terms of bang/buck, even if you buy IBM metal. Not for all applications, but when your application can make use of that enormous backend, applications tend to run quite fast.

I have a sentimental bias towards POWER.

Not only did I work on its first POWER implementation (the RS/6000 aka RIOS workstations of the early 90s) as an IBMer, I also worked on compiler development for its ancient predecessor, the IBM RT/PC. I also worked on various pre-cursors to AIX i.e. version 2 as briefly seen on the IBM PS/2, RT and mainframe.

I also worked on some firmware for the IBM 405GP implementation of POWER in the early 2000s after I left IBM.

I miss IBM's (and DEC's) clean and well documented products. The current world's products come no where close to matching the support we used to get back then. It's very frustrating to try to get things to work these days. So many nebulous firmware blobs you're supposed to download and then deal with the bugs that come along with them. In the old days vendors really did have to get things right the first time. Now it's so much about wading through support matrixes for firmware updates, hoping to hit on a combination of things that actually works, sigh.

I remember Sun Microsystems from their very huge presence in Palo Alto, my hometown. Their corporate campus, former headquarters on Charleston Road, which they vacated in the early 2000s, was a monolith, situated across the street from the place the first integrated circuit was developed at Fairchild Semiconductor. The building was finally torn down around 2005ish to make way for a community center. In my early tech career, I had the good fortune (or misfortune?) of supporting a number of V880s and SPARC systems - ancient history now. I'll always look back at them fondly... an old SPARC 5 was the first piece of Sun hardware I ever touched, back in 1996 or so, when I was 12 - which was running my school's mail server and everything else over a dual ISDN line (anyone remember those?)... gods, I feel like a relic. And I'm not even that old and decrepit yet...

Revelation wrote:

dtw2hyd wrote:

Oracle bought Sun for Java and to kill other hardware manufacturers. By cutting Itanium support Oracle literally killed HP's high-end server market.

POWER was powered by IBM sales force and mostly by confusion at HP (Alpha, Itanium, X86-64, X86, Atom). At one point HP sales guys were confusing customers with too many products.

That's a fair rendition.

I still wish IBM had bought SUN. IBM has a tradition of supporting multiple different OSes and HW architectures. Java was an excellent fit for its enterprise business. But LPOD had a raging hard on for SUN and was not going to be denied, whilst IBM really had no choice but take a more sane approach to valuing SUN.

LPOD is Larry, Prince of Darkness?

Flighty wrote:

LPOD is Larry, Prince of Darkness?

Yes. In the thread starter I mentioned "Larry, Prince Of Darkness ( http://www.urbandictionary.com/tags.php ... %20ellison )"...

I know POWER has made some inroads in space exploration with radiation-hardened processors.

It's interesting to see where certain products find a niche and continue to chug along. A lot of consumer routers and wifi access points used MIPS for a long time and may still to this day, I haven't been keeping up with the new stuff. Power of inertia I suppose.

treetreeseven wrote:

Oracle is SO hated by SO many, and with good reason.

I tend to use fairly cutting-edge companies when I need hosting or VPS services, and you should have seen how fast they moved to dump MySQL in favor of MariaDB once the latter was proven stable and assured to last - which in that case happened very quickly due to the particulars of the fork.

Free software types haaaaaaaate Oracle. Any time an Oracle property is forked, users coalesce around the new fork with amazing speed and unity. The only product I haven't really seen that happen with is Virtualbox.

Part of the well-earned Oracle hate is that many of their products serve totally miserable functions. I'm referring chiefly to ERP. Nobody makes a good ERP system - not SAP, not IBM - and everybody implements them worse. Nobody can make a good ERP suite because their purpose is to enforce the opaque web of compliance rules that defy human comprehension. The result is a mind-numbing and soul-crushing application that ensnares all business operations. It's a vivid example of how regulatory codes spawn into a shadow tax code on our collective productivity. Anyway, rant over.

It has been interesting to see Java market share slowly eroding. I'm sure Java will still be around 50 years from now just like FORTRAN, but it's noticeably fallen in favor as .NET and Python grow in popularity. At least, in my circles.

treetreeseven wrote:

A lot of consumer routers and wifi access points used MIPS for a long time and may still to this day, I haven't been keeping up with the new stuff. Power of inertia I suppose.

In the routers we are making, pure OEM, MIPS is already history and I hear the same from partner companies. But since we do more of it in house we have been using ARM, others may just have a good codebase.. or at least one easy to use. This is actually in a large part due to Raspberry pie, much easier to find people that know their way around an ARM assembler than MIPS.

DfwRevolution wrote:

I'm referring chiefly to ERP. Nobody makes a good ERP system - not SAP, not IBM - and everybody implements them worse. .

I actually quite like Navision. Everything else I had to usend in my life, SAP, Oracle and MyFactory, are pure evil anti-productivity chains. The last one especially, I am pretty sure it was developed by the KGB to bring down western economies. .....

Best regards
Thomas

tommy1808 wrote:

treetreeseven wrote:

A lot of consumer routers and wifi access points used MIPS for a long time and may still to this day, I haven't been keeping up with the new stuff. Power of inertia I suppose.

In the routers we are making, pure OEM, MIPS is already history and I hear the same from partner companies. But since we do more of it in house we have been using ARM, others may just have a good codebase.. or at least one easy to use. This is actually in a large part due to Raspberry pie, much easier to find people that know their way around an ARM assembler than MIPS.

Yes, it seems the critical mass in the non-Intel space has moved on to ARM. The shop I'm working for now is transitioning from MIPS to ARM. They are more of a trend follower rather than a leader, so it shows we're late in this transition.

PI is one good reason. Cell phone entities that purchase millions of chips a year is another. A third is that the vendors get to all claim they use this industry standard architecture, but then all seem to have found ways to do things differently enough to lock their customers into their solutions. ARM's open licensing policy enables this.

It's pretty clear the focus of the non-intel Linux development community has been ARM and it has been that way for a while. Some of this is paying fruit, but as above, the vendors are always going to want to have ways to lock their customers into their particular solutions, so it's far from utopia.

Revelation wrote:

ARM's open licensing policy enables this.

Exactly. Have a fairly light work load overall, but one algorithm eats all the computing power you can through at it, no problem, get an FPGA with an ARM core, or one with the right hard coded unit.... from stock, over night delivery if you need it.
Try getting that from Intel. Try getting another run of an EOL CPU....MoQ: 100k Units.

Best regards
Thomas

tommy1808 wrote:

DfwRevolution wrote:

I'm referring chiefly to ERP. Nobody makes a good ERP system - not SAP, not IBM - and everybody implements them worse. .

I actually quite like Navision. Everything else I had to usend in my life, SAP, Oracle and MyFactory, are pure evil anti-productivity chains. The last one especially, I am pretty sure it was developed by the KGB to bring down western economies. .....

It's the only rational explanation.

RIP Sun. I used to do 90% of my work on SunOS / Solaris SPARC. That was in the 90's; contrast to last year, when I think I turned on the basement Ultra 60 maybe twice to chase some SPARC specific issue. I haven't even felt the need to snag a $100 t2 off of ebay to replace it.

I guess Fujitsu will keep the SPARC architecture alive for a while; one wonders what will happen to Solaris, though.

DfwRevolution wrote:

Part of the well-earned Oracle hate is that many of their products serve totally miserable functions. I'm referring chiefly to ERP. Nobody makes a good ERP system - not SAP, not IBM - and everybody implements them worse. Nobody can make a good ERP suite because their purpose is to enforce the opaque web of compliance rules that defy human comprehension. The result is a mind-numbing and soul-crushing application that ensnares all business operations. It's a vivid example of how regulatory codes spawn into a shadow tax code on our collective productivity. Anyway, rant over.

Having worked on software expressly designed for collaboration within industry consortia to produce documents, which is similar but maybe 1/20th as complex, I can assure you that writing the code for it is just as bad.

I had a friend who worked at the company in question after I left, who said she was continually amused when my comments would come up in a code review since they were always so clearly me. Also they knew on sight the youtube video ID of the clip from Aliens where the colonist is begging "kill me" because I dropped the url in so often in comments in the calendar and balloting code.