A few thoughts through the SUN product line: (Anyone else got ideas that I missed?)
SPARC - likely gone.. IBM will probably stick with Power.
JAVA - might be good for it.. IBM are big supporters of java, with Eclipse, IBM JRE's, etc
MySQL - interesting dilemma - IBM have DB2 CE or whatever it's called as a free DB engine - mysql is not at DB2 level (yet) by any stretch - but if it's good enough - then is it a revenue threat
Solaris - IBM have an awesome NIX variant in AIX. Of course - that's why they now push Linux so much.. :P Not sure what would happen to solaris here...
Disk Storage - The thumper, etc seem like pretty good products.. IBM have some storage kit developed internally (DS8000 etc) - and some I believe is OEM.. I think that some gear like the Thumper might be able to replace the OEM gear... (Would also be quite nice behind an SVC)
StorageTek (Tape Drives) - These will need some consolidation (although again - some of IBM's kit is OEM) - but at the top end - I would have said that the SL8500 (Sun's) trumps the TS3500 (IBM's) at the high end - so that could be dramatic..
Intel/AMD based servers - I'd guess they'll keep them seperate for as long as they can - eventually standardising down to one of the two lines? (Or perhaps spinning some/all out to Lenovo or someone like that?)
OpenOffice - since IBM are pushing it (or a badged variant of it) as Lotus Symphony - you'd assume it would be sticking around...
HPC - I can't see IBM dropped BlueGene - but I've got no clue what Sun are really doing in the space... (from wiki: In 2006, Sun built the TSUBAME supercomputer, which was until June 2008 the fastest supercomputer in Asia. .... Ranger has a peak performance of over 500 TFLOPS, and is currently the 6th most powerful supercomputer on the TOP500 list (November 2008).)
There is a whole vendor ecosystem around high-end SPARC/Solaris systems, especially in financial services and telcos. It could take a number of years for these corporations to move onto IBM gear if that was the key.
Dropping SPARC completely would be foolish, better just rebrand everything to IBM and take it from there.
Agreed. Remember, IBM already has quite a few different hardware platforms: x86 based servers (running Linux or Windows), System p (formerly RS/6000) servers running AIX, System i (formerly AS/400) servers, System x (formerly system/390) mainframes. Adding Solaris Sparc to that would be just be adding one more to a long list. I don't think IBM would have trouble supporting Solaris for a very long time so long as customers still want it.
For me, IBM has looked likely to buy Sun for many years, because doing so would give IBM two things that would be very valuable to it
There are other SPARC vendors (e.g. Fujitsu) that they could sell off to and focus on their own lines (probably with a provision to use the IP they see useful).
Edit: Btw, the ecosystem around SPARC is slowly eroding. Fin Svcs are going x86 more and more. The problem with SPARC is they aren't growing in new markets - it's becoming marginalised.
Enterprise Linux variants are the strategic Unix platform in the City of London (world's largest financial services market), and have been at most places for a few years now.
Just finished up working for IBM at a major ex-state-monopoly Telco, working at the worlds largest hedge fund right now.
I'm a huge fan of the SPARC architecture, and I would LOVE to be able to run 32 threads simultaneously on one machine but unless Sun gets that stuff in the cloud it is dead to me.
That's been true of every other architecture that was inexorably marginalized by x86. You're right that IBM isn't going to unceremoniously drop it. But it's dead, and good riddance.
I remember McNealy admitting they underestimated 32 bit architecture and placed too much on 64 bit.
Another error might have been overvaluing its vertically integrated SW/HW solution technically and financially, and not commoditizing it quickly enough through open-source and hardware cloning.
Spreading one's wealth goes against one's better judgement often, but the rewards come back in unseen ways.
In Sun's case I think it's simpler than that: SPARC hardware carries a steep premium over commodity x86, a good chunk of that premium is actually intended to pay for Sun's software stack, and Sun simply hasn't executed well enough on the software side to merit the premium.
Sun did a lot of engineering stuff "right" in the late '90s, and again over the past couple years: they were first to market with a mainstream-grade MP kernel and first to market with a mainstream 64 bit OS, and recently they have ZFS and DTrace to talk about.
On the other hand, I think you could have predicted probably starting in 2001 that nobody was going to pay that much for large-scale MP, 64 bits, and advanced filesystems; the entire server market swung to web stacks, which scale horizontally and move most of the systems management burden into a DBMS and out of the Adrian Cockroft and Chris Drake systems engineering realm where Sun has excelled.
I really think that if you compare the systems-level innovation at Apple to the systems-level innovation at Sun, Apple comes out ahead. And systems-level innovation isn't even their core competancy.
I think there was another big sea-change caused by AMD.
Before this, architecture and ISA were absolutely key - RISC was the way of the future. RISC architectures were theoretically running loops around CISC ones.
AMD bolted a CISC onto a RISC core and effectively made this architecture point moot. This approach let AMD and Intel use their momentum and market to catch up and overtake their workstation friends.
The big sea change that AMD caused is that 64-bit processors are now a commodity. Intel COULD have done that without AMD, but I doubt that they WOULD have if AMD hadn't essentially forced their hand.
And it's so ironic that people who were not married to x86 architectures enjoyed 64-bit computing a full decade before it became mainstream (as in "you could order a box from Dell")...
And that Sun had one of the first full implementations (I think MIPS got there first).
Yeah - although microcode had been around for a very long time.
The difference between microcode and RISC was the amount of synchronisation - RISC generally implied lots of uniform, high performance operations. Microcode could, but didn't necessarily. e.g. Microcode could still be doing funky co-ordination with lots of specialised execution units (and did afaik). This has some similarities to RISC, but RISC is a different philosophy really - namely fewer types, pipelined accesses, uniform instructions, more registers, etc.
(In my view) the big watershed moment was AMD's K5 architecture. This was literally based on one of AMD's pure RISC designs. It wasn't that popular in itself, but it set up a golden age for AMD... Which really only came to an end with Intel's Core.
Microcode is an implementation technique. Micro-ops (or whatever Intel formally calls them) is an instruction dispatch architecture --- an actual microarchitectural feature.
Solaris is way better than AIX, and a Power port is half done (historically there were a Power version).
I don't know really, but IBM could choose to drop linux or, my hope, merge solaris and linux... this would be terrific, if backed by a behemoth like big blue.
Because Linux is (and will be for the foreseeable future) licensed under GPLv2, which is incompatible with software under v3. There are also some issues on driver architecture that make ZFS and the Linux kernel a tough match.
btrfs is supposed to one-up ZFS anyway, and that's going into mainline as experimental with the imminent release of 2.6.29. I'm not sure how btrfs and ZFS they compare, just know that btrfs is being developed since Linux has been unable to get ZFS.
This article quotes the acquisition at around $6.5 billion(http://www.reuters.com/article/newsOne/idUSTRE52H1GS20090318). StorageTek acquisition was $4.1 billion(http://www.infoworld.com/article/05/09/01/HNsunstortekcomple...) and MySql AB acquisition was around $1 billion. That makes $5 billion. So is rest of the Sun worth $1.5 billion? Clearly, it seems that these acquisitions have decreased Sun's valuation, rather than increasing it. Also, I didn't see any point in buying MySql AB for $1 billion.
It would be a sad day if Sun stopped being independent, but IBM does seem to be the best match that will preserve the most of Sun. Maybe they can even fix some of the JCP and OOo policy problems. On the other hand, NetBeans will probably go the way of the dodo..
A friend of mine pointed out that commoditizing someone's product was a great offensive move. He thought that was why IBM funded so many open source projects, many of which were in the Java space. He pointed out that Eclipse is one of the slickest Java IDEs, and what is an Eclipse, but an occlusion of Sun?
IBM and Microsoft have taken two different tracks. IBM is trying to everything except custom software development and support contracts because those are the most profitable short-term. They give away low-quality software that is difficult to use and missing important features; then they charge every user the full cost of fixing these problems. Basically, IBM wants you to rent expensive people to run your free software.
Microsoft is trying to make custom software development and support contracts unnecessary by selling quality, feature-full software. I'm sure lots of people want to say something snarky about Microsoft's actual software quality, but at least recognize that it is the plan. Basically, Microsoft wants you to buy software that ends up being much cheaper than renting people. Ultimately, they want you to computing appliances that are powered by their software and which don't require any maintenance or custom coding.
By the way, there is nothing slick about Eclipse. I am using it every day to write S60 applications and I hate it. I am using IntelliJ IDEA 5.1 (three or four years old, three major versions behind) and IntelliJ is much, much better.
You've no argument from me about IntelliJ being a better IDE. But remember when Eclipse first came out. In my recollection, it set a new bar for Java UI slickness.
This is key. Open Source often fail to reach a certain level of polish unless they find some way to receive corporate funding.
No, I'm not. Microsoft does have consultants and certification and probably always will. However, every release of every one of their products is designed to reduce the need for those people. For example, the ultimate goal of SharePoint is to remove and/or substantially reduce the need for custom website development.
That's along the same lines, but my friend thought that IBM's strategy was to eliminate much of the profit potential of what Sun was developing in the Java (software) space by commoditizing it. (Joel points out that with Java, Sun is commoditizing its own hardware.)
This brings to mind Apple. Software, Design, and Integration are what distinguish Apple Computer, allowing them to charge a markup above the commodity level. The hardware is largely commodity hardware, except for the fact that it's slickly designed in a way that Apple's competitors have some difficulty matching, and that it's (supposedly) the only hardware that can run OS X. Various Hackintoshes testify to the true commodity nature of Apple hardware.
I've been predicting that someone is going to enter the low cost space and offer a truly integrated software/hardware experience based on Linux. I'm wondering if Apple is planning to do just that.
Owning VirtualBox virtualization software is a great reason for IBM to buy Sun. IBM is trying to be a major player in the cloud computing space, but it relies on Xen and VMWare for x86 virtualization technology. IBM doesn't have much influence on Xen and its open source development community. Obviously IBM has even less influence over VMWare. So buying a solid x86 virtualization stack like VirtualBox would make IBM a serious contender for cloud computing deployments built from commodity x86 boxes.
IMO VirtualBox is the worst virtualization tech available today (that's not to say that it's bad, just that the others are better); they just added 64-bit support recently and I don't know if they even have SMP guests yet. You probably shouldn't even ask about SR-IOV. IBM already has KVM.
I suspect you could have made Xen (on the high end) or Jails (on the low end) equally as powerful. I'm not a fan of Zones, and of all the Solaris innovations of the past couple years, I think Zones have the least staying power.
Yes, you have made your bias very clear on Zones here on HN every time the subject has come up. Clearly they don't work well for you. However, people do use Zones and they clearly do work well.
I suppose you have figured out how to use a jail to limit memory and CPU resources of the programs running in a jail? Ooops, that is a zone-only feature; like the time I set up Oracle in a zone and limited it to using no more than 4GB RAM .
On the high end, I'd use Xen, like all the major large-scale hosting providers do. On the low end, I wouldn't bother trying to do fine grained resource control.
It's not exactly insightful to point out my bias against Zones in a response to a comment where I basically say "I'm biased against Zones". I am indeed biased. And I can defend my argument that, as a technological approach, trying to compartmentalize all the kernel namespaces and building ad hoc group-level rlimits is a dead end.
For what it's worth, just as your bias in favor of Zones (which I don't fault you for) is based on professional experience, so too is my bias against them.
Sun were a good 5 years too late open-sourcing their software stack including Java and Solaris, and should have made their hardware easily licensed for cloning.
I love open source, and if I had the money, would work with nothing else. That said... for a big business like Sun, would that really have helped? Maybe their stuff would have been more popular, but it would have required a wrenching transition to services that might have hurt them very deeply.
I think they should have taken this alternate Open path and coupled it with business-as-usual in the rest of the organization.
So they lose revenues in the normal part of Sun because those customers now pickup their SW for nothing essentially, and instead of Sun then laying off employees, they are shifted into services. One area goes down, another goes up.
Hardware at Sun then becomes less focused on manufacturing, and more about building a community of external implementors by providing good HW designs and support.
> instead of Sun then laying off employees, they are shifted into services. One area goes down, another goes up.
The same people that might make great software producers might be lousy at services, so it's entirely possible that large layoffs would still happen, which would hurt morale. Indeed, 'services' is a different industry, and it's not like you can just pick up from one day to the next and become a services company. IBM managed to do that, so maybe they would be a good fit, but it's not an easy thing to accomplish in any case.
They are a hardware company. Why would they want people to clone their hardware is beyond me. The IBM PC was cloned and the result is that IBM is a relatively minor player in the PC arena and one that has no meaningful power (much unlike Intel and Microsoft).
It was Intel that gained from cloning, and IBM then suffered, you're right. But MS had the APIs into it for Windows so it gained.
The idea was that Sun becomes a designer of and licensor of state-of-the-art HW designs (which it may or may not implement). The designs come with software APIs too that different companies may want to implement.
The difficulty is that SPARC International is already a different organization to Sun. So, I don't know if they could've done this legally.
There is one nice way to boost Sun's survivability: to get the server room rid of Windows boxes - a segment where Sun can't compete. To deploy a solution that currently runs on Linux/x86 on a SPARC/Solaris or even SPARC/Linux is really easy and Sun makes some wicked fast machines for when you need throughput and not single-thread performance. To do it with Windows is a whole lot harder. The good thing: we would get performance and reliability, and in exchange, we would get rid of Windows (and, of course, Exchange and Outlook).
A second, interesting way, would be if Sun decided to make inexpensive SPARC desktops based on their T1 or T2 designs that encourage people to develop massively multi-threaded applications that run best in spaces where x86 boxes can't go just yet (and won't go until at least Larrabee). But they need to run and build something sexy like a Mac mini and they need it by last year or so. I doubt they could pull this one off.
As for me, I am going to stockpile Sun type 6 USB keyboards and matching mice, just to ensure my ability to have a keyboard with keys labeled Help, Stop, Again...
Sun was undervalued and even at the price IBM appears to be offering, Sun is a good value.
Why? Sun still has elaborate R&D, strong balance sheet, multiple revenue streams. Granted, it had overpaid for many acquisitions(I am looking at you MySQL) and hasn't figured out exactly how to leverage all that OSS into income.
IBM should turn Sun into a cloud division. With Java/J2EE, MySQL and their storage/data center expertise it seems like a good match?
A full featured J2EE cloud solution could get a lot of traction with all the many corporates heavily invested in Java and looking to reduce costs (IMHO).
Smart move. One of IBM's core competencies is hardware (assembly); one of Sun's core competencies is software (esp. virtual space software) and innovation.
The danger, of course, would be in losing that innovation when/if the acquisition occurs. If IBM does end up acquiring Sun, it would probably be a good idea to let it maintain as much independence as possible. There seems to be an inverse relationship between size of an organization and its ability to adapt and innovate.
Except that IBM has been selling their hardware units off for years... hard drives to Hitachi, laptops to Lenovo, etc. IBM's core competency now is consulting services supporting software.
I think Sun makes sense for them to acquire (their business models are becoming quite similar already)... but IBM's days of having a hardware focus are over.
Do you think their HPC and server hardware will go, too? I'm guessing they'll stay in whatever hardware business is not commoditized, which means a lot of Sun's hardware fits into IBM's portfolio.
Moreover, there is an important lesson to learn. Open source is not the solution for poor revenues. Sun has OpenSPARC, OpenSolaris, Java, OpenOffice, MySql and what not...But in terms of revenue, it hardly made a difference.
All I can say is Timer Warner / AOL. Not a popular opinion, but I really hope Sun dies. I'm probably not up to snuff on their product lines, but I know they've developed and promoted the atrocity that is Java and I think they're quickly killing Mysql. IBM, on the other hand, has proven itself as a company that can reinvent itself over and over and have been doing some very cool things. They're one of the few classic companies that still do actual long term R&D without the requirements that they crank new products out in the next 2 years. It would be really bad for them to buy Sun.
Sun has some amazing engineering staff. They produced such wonders as Dtrace, Sparc, and ZFS. If IBM were to buy them and take the good portions it would be a huge boon to IBM, Open Source and the software/computing community.
These people will find a good home elsewhere. Sun, as a company, has been floundering since the dot com crash. Their most popular stuff is free and many people choose cheap linux boxes over heavy duty sun boxes. The investment bank I used to work for is a good example of the heavy computing world dropping Sun.
A merge with IBM would solidify their irrelevance if you ask me...
Java is strong (while not most 'hacker's cup of tea) large amounts of buisnesses like it. I could see IBM cleaning up some of the less pleasant sides of it and continuing to support it slash push it under the Sun brand.
If they were to rebuild the Sparc line this could also be done under the Sun brand. I could see rebuilding sparc to be useful particuarlly in response to the latest Intel AMD debacle. If they made something like KVM (kernel virtual machines) more strongly built into the hardware this would likewise help them push the line.
IBM is a smart company, they could definitely create a decent buisness plan to use Sun to their advantage.
IBM has the POWER lineage of RISC CPUs for servers, which has consistently beaten Sparc in benchmarks for at least the past eight years or so. POWER also has the advantage of having a shared ISA (and thus shared R&D) with IBM's PowerPC-based mainstream products, e.g. the CPUs used in Xbox 360 and Playstation 3.
Thus I think that Sparc will be among the first casualties if this merger goes through.
I think you're right about this one. SPARC hasn't been competitive since the late 90's, so it's hard to see why IBM would care about keeping it alive in the face of POWER -- not to mention, IBM is also one of Intel's biggest server OEM's.
Say what you will about Java-the-language, especially after the world has moved on to higher-level languages, but it's hard to deny the extent to which Java has been a force for good in enterprise software development. The de facto standard for companies that didn't adopt Java? C++.
Yeah, I can deny it. I've worked in java shops for 2 years. The "good for enterprise" argument is bunk. All its lead to is inefficient and disastrous technology decisions, an inability to be agile and flexible in software development, and a general dumbing down of any smart engineers that happen to be stuck working there.
As a language, the best metaphor I can imagine is that Java is like the language of some Amazonian tribes. They don't have the concept of numbers bigger than 2. Its not about their brain power, just the way their culture and language has shaped their minds. That's basically what Java does to software engineers.
I would argue that this is what the Java culture does to software engineers, not the language. There seems to be this culture of the 'Enterprise Astronaut', that says everything needs to be 'Enterprise Ready'. I'm not sure what this means, but it seems to indicate incredible complexity.
But the language itself doesn't enforce this; it is very possible to write concise software in Java. What's better still, is you can use a Groovy (or JRuby, or Jython), and still use the HUGE class library provided by Java.
If you haven't played with it, give one of the JVM dynamic languages a go.
You've worked in enterprise Java shops for 2 years. I've spent ~4 years now parachuting into enterprises and doing security evaluations on Java apps. You may not like the "agile" "flexible" "dumbing down" "bunk" (and, hey, I'm not going to write something Java either), but you can safely take my word on it that the bunk is better than the empirical results of those same developers writing C++ code.
SPARC - likely gone.. IBM will probably stick with Power.
JAVA - might be good for it.. IBM are big supporters of java, with Eclipse, IBM JRE's, etc
MySQL - interesting dilemma - IBM have DB2 CE or whatever it's called as a free DB engine - mysql is not at DB2 level (yet) by any stretch - but if it's good enough - then is it a revenue threat
Solaris - IBM have an awesome NIX variant in AIX. Of course - that's why they now push Linux so much.. :P Not sure what would happen to solaris here...
DTrace - obviously kept - maybe linux/aix porting? (we can hope)
ZFS - again - you'd assume they would keep it.
Disk Storage - The thumper, etc seem like pretty good products.. IBM have some storage kit developed internally (DS8000 etc) - and some I believe is OEM.. I think that some gear like the Thumper might be able to replace the OEM gear... (Would also be quite nice behind an SVC)
StorageTek (Tape Drives) - These will need some consolidation (although again - some of IBM's kit is OEM) - but at the top end - I would have said that the SL8500 (Sun's) trumps the TS3500 (IBM's) at the high end - so that could be dramatic..
Intel/AMD based servers - I'd guess they'll keep them seperate for as long as they can - eventually standardising down to one of the two lines? (Or perhaps spinning some/all out to Lenovo or someone like that?)
OpenOffice - since IBM are pushing it (or a badged variant of it) as Lotus Symphony - you'd assume it would be sticking around...
HPC - I can't see IBM dropped BlueGene - but I've got no clue what Sun are really doing in the space... (from wiki: In 2006, Sun built the TSUBAME supercomputer, which was until June 2008 the fastest supercomputer in Asia. .... Ranger has a peak performance of over 500 TFLOPS, and is currently the 6th most powerful supercomputer on the TOP500 list (November 2008).)
And Wikipedia has a nice listing of all the goodness of sun hardware: http://en.wikipedia.org/wiki/Category:Sun_Microsystems_hardw...