Oracle Staff Report Big Layoffs Across Solaris, SPARC Teams (theregister.co.uk)

Simon Sharwood, reporting for the Register: Soon-to-be-former Oracle staff report that the company made hundreds of layoffs last Friday, as predicted by El Reg, with workers on teams covering the Solaris operating system, SPARC silicon, tape libraries and storage products shown the door. Oracle's media relations agency told The Register: "We decline comment." However, Big Red's staffers are having their say online, in tweets such as the one below. "For real. Oracle RIF'd most of Solaris (and others) today," an employee said. A "RIF" is a "reduction in force", Oracle-speak for making people redundant (IBM's equivalent is an "RA", or "resource action"). Tech industry observer Simon Phipps claims "~all" Solaris staff were laid off. "For those unaware, Oracle laid off ~ all Solaris tech staff yesterday in a classic silent EOL of the product."

Comments

Rule #1

[UndyingShadow]

"Don't make the mistake of anthropomorphizing Larry Ellison." -Bryan Cantrill

Re:Rule #1

[ShanghaiBill]

"Don't make the mistake of anthropomorphizing Larry Ellison." -Bryan Cantrill

Larry can be very cold hearted, but I can't fault him for this. Sparc/Solaris was a great choice in the 1980s, a reasonable choice in the 1990s, and a bad choice ever since. There can't be much more than a handful of legacy customers left. I am surprised that it took this long to finally die.

Re:Rule #1

[Anonymous Coward]

After Solaris was open sourced, there seemed to be a time when the use of Solaris could potentially expand greatly. Once Ellison closed the code again, there was no hope at all. Ellison should be blamed completely for this.

Re:Rule #1

[ShanghaiBill]

Then why the fuck did he buy it?

To kill MySql, which was a long term threat to Oracle's core business.

Maybe his company could have done less of a dick move

Without dick moves, Oracle wouldn't be Oracle, and Larry certainly wouldn't be Larry. Being a dick is his core competency.

Re:Rule #1

[thesupraman]

MySQL? A Threat a real database? Umm. No.

PostgreSQL perhaps, but not MySQL.

Just because its popular with 'toy database' applications (where the db is really just slightly glorified read only storage) doesnt mean it is a threat or oracle - different horses for very different courses.

Re:Rule #1

[Anonymous Coward]

MySQL? A Threat a real database? Umm. No.

A threat to real database needs, no. But a threat to real database sales, yes, at least in the eyes of people who see any alternative as a lost sale. Sure, MySQL is mostly a replacement for a structured file, but that's still one replacement for a structured file that Oracle wasn't paid for.

It's not much different than the kind of mind set that would claim that a 10 year old a yearly allowance of $1200 copying a bunch of MP3s is a loss to the music industry of $12 million.

Re:Rule #1

[TheRaven64]

PostgreSQL is less of a threat to Oracle than MySQL. If you start building a system with PostgreSQL, you have a reliable database with a decent query optimiser and you'll end up putting a lot of logic into the database. Once you grow to a big enough size that you might be able to afford Oracle, they send along salesmen and convince your least competent C?O that your data would be a lot safer if you migrate to a proper database and they provide migration tools to do this. They can probably demonstrate some speedups from their custom filesystem and query optimiser. On the other hand, if you start with MySQL, you end up using the database as little more than a key-value store and put most of the logic outside of the database. It's much harder to demo a place where Oracle is an improvement because your bottleneck is the application logic that's doing a load of redundant database queries, not the database itself.

Re:Rule #1

[Anonymous Coward]

Not sure if you've noticed, but Oracle have spent the last 2 years pulling volume discounts from under some pretty big companies. My entire multinational has set a deadline to remove all use of Oracle, and I know we're not the only ones migrating from Oracle to Postgres

Re:Rule #1

[TheRaven64]

I didn't, but I'm not surprised. Last time I spoke to someone at Oracle, they said that the revenue from the DB was up, but they had very few new customers. Their real problem is Moore's law and friends. In the early '90s, payroll for a moderately large company took a serious database on high-end hardware to manage, with several hours of compute time to complete the payroll calculations each week. Now, that same company is maybe 10 times the size, but you can buy a computer more than 100 times better in terms of RAM, storage capacity, storage speed, and processor speed, for a few hundred dollars. On a modern machine, you don't need carefully optimised queries and carefully designed on-disk data structures - you can probably fit the whole thing in RAM and run the computations in Lua and still get the whole thing done in a second or two. At one end of the market, most customers can now run their systems with cheap commodity hardware and software. At the other end of the market, companies like Facebook and Google have more data than Oracle can easily handle and couldn't afford Oracle license fees even if there was a viable Oracle product for them to buy. The middle is gradually shrinking.

Re:Rule #1

[MouseR]

He bought it mainly because of Java. Oracle has a huge investment in Java and there was no way in hell he could afford to let it die or, worse yet, let it go to IBM.

Solaris was a nice bonus because Oracle also had lots of investment in that hardware/software platform for it's own servers. From where I sit, I saw a gradual move to Oracle Linux over the years but I can not say wether that is responsible for dropping Solaris/Spark.

There was a time I had a Solaris virtual box. Mostly unused but was useful when I needed to install a particular server instance for the applications I was doing at the time. Nowadays, I have an Oracle Linux virtual machine on which runs one of our development tools, for our App development. Other members in the team have their own Oracle Linux virtual machine to host dev or qa instances of the server for our App development use.

Haven't seen an OpenSolaris box in years, but I know one of our training center in MDC has (had?) a few boxes on there. Not sure they are still there. Been ages since I walked in that office.

Re:Rule #1

[Anonymous Coward]

WHY would any SANE programmer put business logic into the model layer?

To uniformly enforce rules when accommodating multiple applications accessing the same data.

Another reason is that the database will usually be the longest living part of the software architecture.

In terms of things being tightly coupled, defining a set of stored procedures for data access hardly makes the application tightly coupled. Using raw SQL tightly couples your application to the schema. If the schema needs to change for database reasons the stored procedure layer could prevent you from having to modify the applications.

Yes, there are other ways to accomplish the same thing but it's not an insane decision.

Re:All Solaris Staff?

[theweatherelectric]

Can the codebase be recovered?

Doesn't need to be. Illumos is the open fork of (not so) OpenSolaris.

VirtualBox

[that this is not und]

Has there been any word about VirtualBox? That is pretty much the only former-Sun softwarevI use on a regular basis. Since the Oracle purchase of Sun I have wondered why Oracle was keeping it alive.

Re:VirtualBox

[Tailhook]

VirtualBox is difficult to explain. Oracle has never seriously tried to monetize it. They've never inflicted a Java installer on VirtualBox users. Most of it is still Open Source and they haven't driven off the entire user base to some fork. They haven't done any of the damage this sociopath or a corporation does to everything else they acquire.

I pointed this out to a moderately clever person once. He suggested that perhaps Oracle forgot about it. Maybe there is a small team of dedicated developers quietly enhancing their work, filling out their TPS reports and successfully avoiding notice.

Re:All Solaris Staff?

[theweatherelectric]

ZFS enhancements come to mind.

After the split with OpenSolaris, ZFS development moved to the Illumos project and OpenZFS grew out of it.

More information the Layoff

[plopez]

In case anyone wants more insider info. https://www.thelayoff.com/orac...

A problem of Sun's making.

[upuv]

Lets face it Sun made mistakes. These mistakes made it ripe for take over and plundering by Oracle.

The biggest mistake was Linux. When Red Hat launched no one took it seriously. Red Hat legitimised Linux in the eyes of industry. Companies faced with massive expansions of internet equipment simply could not afford the iron from SUN / HP / IBM. These small start ups went to Linux. One such startup was Google. All of a sudden massive new companies emerged on a platform that was not enterprise iron. Overnight Dell become a major player in the server room.

What did Sun do about this? Nothing. Even when faced with new unit sales that were almost zero compared to just a few years before sun still did nothing. Sun released Solaris for x86/64. But completely forgot to get 3rdparty shops and it's on internal application development teams to port to it. They only thing that ran on the x86/64 Solaris was open source software. Stuff that was already running on much cheaper x86/64 gear. Sun limped on for a few years making the occasional uplift sale for existing gear in the field.

Then Oracle pounced. Suns mistakes led to this point. They sold for far less than they were worth. Why? The market lost all faith in Sun's ability to generate a cash flow. Thus the negative impact on asset value.

Oracle saw something it liked very much. Java. Oracle instantly went on a predatory path of trying to extract money out of Java. We all know how well that went. Oracle just recently announced that they are looking to open source the Java EE specification. This predatory cash grab caused some other interesting market changes. The explosion of new languages resulted and they got market share. Ruby on rails in my opinion would have never had as much success as it once did with out Oracles legal threats over java usage.

Oracle also needed to save it's DB division. At that point Oracle DB pretty much was only ever deployed on pricey Sun gear. Also Sun owned the control of MySQL. Which was growing at an astonishing speed. This threat needed to be squashed at all costs. In my mind these were the real reasons for Oracles purchase of Sun.

There were some other assets that Oracle was looking to sell off. But in the end Oracles taint reduced their value to zero. OpenOffice comes to mind.

I still remember the day when Oracle purchased Sun. There was an audible groan in the office. Execs around the world were scrambling to find alternatives to many products. Sun certified engineers instantly saw there pricey certificates devalue in an instant.

The only reasons Oracle has kept Solaris and SPARC alive for so long are:
1. Uplift purchases still come in. But not many. These can be counted in 1000's world wide. Basically nothing.
2. The platform became part of the Exadata/Exalogic platform. ( An an holy creation in my mind. )

Re:A problem of Sun's making.

[whoever57]

The biggest mistake was Linux. When Red Hat launched no one took it seriously. Red Hat legitimised Linux in the eyes of industry.

I disagree. The root problem was that single-thread performance of Sparc lagged Intel x86 around the time the PIII came out. Linux made adoption of x86 possible for many applications, but few of them would have moved to x86 without the big performance and cost benefits of adopting it.

Sun tried to compensate with more cores, but that only made Sparc more expensive and many applications were not written to take advantage of multi-core hardware.

Re:A problem of Sun's making.

[bungo]

This is very true.

This wasn't just Sun's problem, IBM had similar issues with the AIX Power servers. The price/performance of Java on x86 was far better than anything that Sun or IBM could offer. They were still stuck with the mindset of selling expensive mid-range type systems. They could offer better scaling, with more cores in a single box, but when you could have a distributed application, with many small servers, the old vendors couldn't compete.

Once Redhat went public and management were able to convinced that Linux was a real alternative, instead of buying 4 AIX servers of $75k each, I was able to get 12 Linux x86 servers for $4k each (rack mounted, so a little bit expensive). The price for Sun SPARC servers were cheaper than the AIX servers, but still way more expensive than x86 servers.

The Linux servers were multiple times faster for the Java applications that they had to run, for just over 1/2 the price of a single AIX box.

Re:A problem of Sun's making.

[_merlin]

I see it differently. MS killed desktop UNIX by being good enough. Desktop UNIX was better, but Windows NT 4 on commodity hardware was (perhaps just barely) good enough, and far cheaper. This impacted all the UNIX vendors, as it killed off a significant revenue stream.

Sun was losing the performance battle badly with SPARC. This was painfully obvious with the Sun Blade 2500 tower. The CPUs were no faster than the Blade 2000, and it had a quarter of the RAM bandwidth as a cost-saving measure (they reduced memory from 256 bits wide to 64 bits wide so you could install RAM DIMMs individually rather than in groups of four). They were still trying to charge premium prices for SPARC, but x86 and POWER were walking all over it in performance. They managed to score some wins on highly-parallel workloads with the UltraSPARC T series (Niagara), but it was always specialised and Suns AMD64 boxes performed better on most workloads.

Sun was spread far too thin on the ground and faced with falling revenues. They tried open source as a strategy, but open sourcing something provides no value on its own. They couldn't maintain/enhance their OS (Solaris), compilers (SunPro), Java, and everything else. They were floundering, occasionally coming out with an interesting product, but no consistency.

Ruby on Rails was framework-of-the-month already by 2006, years before the Oracle purchase. The explosion of programming languages would've happened anyway, it's a result of all the people who studied computer science after the first dotcom bubble. J2EE had already been co-opted by Apache and Red Hat, and let's face it, hardly anyone even uses J2EE. Most of the time Spring is a lighter-weight substitute, you only really want J2EE if you need distributed transactional consistency.

Sun were lucky to sell out when they did. They really didn't have a viable plan, and taking the money and running probably was the best thing for the shareholders. They'd already open-sourced most of the stuff that had any value, so it wasn't going to matter too much how Oracle mismanaged it. Oracle were stupid to buy Sun, as it was already pretty obvious there wasn't really much worth buying.

Yeah, I miss Sun, but the Sun I miss had already faded long before the Oracle purchase. I just didn't want to believe it and willingly went into denial.

Only real question is

[oldgraybeard]

Why did they wait so long. We all knew this was going to happen. Oracle wanted the intellectual property not the actual projects or the people.

Re:Only real question is

[upuv]

The big reason they waited so long was they wanted/needed to migrate existing customers onto exa or cloud platforms. If they just killed it quickly they would effectively be loosing those customers, probably forever.

Also OracleDB was one of the only reasons to buy SPARC anymore. Database shops hate platform change in general. No one ever got fired for buying SPARC for the DB environment.

Re:sorry, not sorry

[upuv]

In the early days SPARC was one of the only platforms that you could run x64 Java. Linux x64 was considered unstable and untested at the time.

So we bought massive SPARC boxes to host java apps. A lot of java apps at once. And yes the performance was garbage. A dedicated engineer at least just constantly on the lookout for performance issues.

I still remember one team going OK we built those new linux boxes for you. Only to find out they built them with 32bit linux. They simply would not build them x64. I had to remind them that the whole point of the project was to evaluate java on x64 intel. When they finally did rebuild them properly the initial performance tests were amazing. That company never looked back after that point.