Thoughts on Oracle buying Sun...

So Sun finally found a buyer. The IBM and Sun deal fell apart for various reasons (depending on who you ask) and now Oracle has swooped in and temporarily entered the hardware market. I've been writing this post in bits and pieces since I heard the news. The piece kept getting longer and longer and more unwieldy every day. So I've decided to break it up into three separate posts.

• What I like about the purchase



• Oracle as a hardware vendor



• The dead pool (what software products are going to die).

 

Over all I think it's a great deal for Sun. It remains to be seen how this works for Oracle. One interesting aspect of the acquisition to me, on a superficial level, is the logos.

 

Oracle Red

vs

 

Sun Blue

For my money, Sun has the superior logo and color since switching from that fisher price purple. A Red Sun logo looks pretty sad, IMHO. Huh, Red Sun, has an interesting ring to it.

 

The Dead Pool, who dies in the Oracle purchase of Sun?

Since Sun's primarily a hardware company and Oracle's primarily software, you'd think there wouldn't be a lot of overlap between them, but you'd be wrong. There's going to be a few hard choices to make for Oracle. Some of them are easy. Sun's Java Systems Web and App (aka SunOne aka iPlanet, aka Netscape) products will be killed. They weren't that heavily used anyway so no one will really miss them. Oracle can offer them an easy transition. The area of Identity management is going to be a sticky one. The fusion stack has an integrated LDAP already, but my limited exposure to it has been unfavorable. Sun's LDAP and IDM products only slightly better in my view. Both have good and bad points. But in the end, you only need one. So it'll be interesting to see what happens there. I suspect that most of the Sun products will loose most of these fights. They just don't have the momentum or market presence to stay.

Then you get into the sore spot for a lot of people. The open source products and projects that Sun supports or owns. Oracle already has Oracle Developer Suite. Sun has Sun Studio and the fairly popular NetBeans. And developers are the most cool-aid drinking crowd you'll ever find. You think Linux and WIndows folks don't like each other, just put a NetBeans user in the same room as an Eclipse user and ask them which is the better IDE. There's no good answer here. You can't keep them all, although because of the open source nature of NetBeans there isn't a lot of expense to keeping it around. Sun Studio is probably already dead, we just don't know it yet. Then there's MySQL... Seems like chicken little just did this dance about a year ago when Sun bought them. Some folks think it's not as bad as it sounds and I think they're right. Oracle could actually bring some real value to MySQL. The open source issue would still be sticky, and they don't want to erode their sales, but there could be a fairly sizable chunk of middle ground for MySQL and Oracle. Other open source projects probably won't be so lucky. Glassfish is probably going to get dumped. Not a lot of upside for Oracle with Glassfish. Then there's Oracle Linux and OpenSolaris. Although there's probably room for both, strategically they should pick one. My wishful-thinking bet is on OpenSolaris. The tie-in's with the hardware line are too strong to ignore. Oracle Linux hasn't exactly lit up the server market anyway. But any way you slice it I'm expecting forks of a lot of Sun sponsored open source projects at the first sign of Oracle playing rough with the Open Source community.

Other interesting areas are going to be in the grid space. Sun and Oracle have competing grid technologies. Not sure who wins out in the merging of those products. I suspect Sun Cluster will be put out to pasture in favor of ClusterWare (which also portends bad things for Symantec's VCS product). Sun's N1 management suite will probably be left with out a chair when the music stops as well. It will be interesting to see where Oracle simply drops a product vs where they merge a product. In some cases the products have different strengths, and merging them makes a lot of sense. But by the same token you don't want to create market confusion in your product portfolio, so they have to walk a fine line.

So my software deal-pool list (dead as in not a Sun product/project anymore., forks/spin-offs not counted):

• Sun Java Systems Web and Portal products



• Suns IDM/SSO suite



• Glassfish



• Sun Cluster



• Sun Studio



• NetBeans (*sniff*)

I'm sure I'll amend this list with a few that I've overlooked later.

Oracle as a hardware vendor

It will be interesting to see how long the 'Sun' badge stays on the equipment. Right now I think the Sun name will stay on for 2 or 3 years or until Oracle exits the hardware game. I'm betting that they will start cutting hardware sooner, rather than later. Oracle loves fat maintenance contracts, so I could be wrong about them leaving the hardware game at all. There's a lot of recurring revenue to be had on the support contracts. But the Red Sun won't be able to compete in as many hardware arenas as it did in the past. I suspect underperforming lines will be culled quickly. I don't know what lines those may be. If, for example the blade servers, aren't selling like hot cakes (relatively speaking of course), then I think it'll be taken out behind the shed, old yeller style. Eventually I suspect all the server hardware will go this way, but that's just a speculative guess on my part. I know Oracle would love to have an App to Disk solution on hardware that no one else sells. Unfortunately it's no guaranty of success. If owning the whole solution where all it took to be successful, we'd all be running IBM and Apple.

Product areas I see getting killed off first are the blade systems. I could be mistaken but I don't think the blade idea has much steam. The barrier to entry for most companies is too high. Plus there's the 'eggs in one basket' problem that make a lot of systems architects and engineers shy away from blades. I wouldn't be surprised if they kill off the intel and AMD based systems as well. Let HP and Dell have that. Although they could try and keep it going to have that bundled solution. Systems preloaded with unbreakable linux or Solaris and preloaded with Oracle goodies as a value add. The first spin off will likely be StorageTek, which will likely kill the brand. IT history is full of brands that diminished under a new owner and then died after subsequent resale (Who uses wordperfect anymore?). The reason STK makes a a good spin-off candidate is there are other tape/storage vendors who would love to get their ends on their install base. The same can not be said for the x86 business or SPARC.

My current feeling is we'll see product streamlining first and then reductions/spin-offs. Mostly scalpel type cuts, but few major changes in the first 18 months. A lot will depend on the state of the IT industry. If IT purchasing is ramping up, Oracle will likely keep things going for 3 years or so. But if IT spending is in the tank or they botch the merging of the two support/sales organizations they'll go at the Sun hardware lines with a battle ax.

What I like about the Oracle purchase of Sun

I'm not quite sure what the 'vibe' on the street is (and by street, I don't mean wall street, they're like headless chickens riding a herd of sheep). So far I think reaction is mostly neutral or 'wait and see'. The open source community is up in arms fearing a mySQL death blow. I think that fear is unwarranted at this time, and I'm not alone in that thought. I think there could be some real upside to Sun customers, especially Sun/Oracle customers. Oracle brings financial stability and customer base to the table. Sun brings a recently invigorated streak of innovation and solid hardware creation & manufacturing skills.

It would be interesting to see what kind of oracle appliance the combined team could come up with. I have visions of an appliance based on the T5240 that auto-configs via the enterprise manager grid control. It woud ship pre-installed with RAC and the Fusion stack. You plug it into the appropriate VLANs, it attaches to the grid and then you tell it what kind of work you want it to do. It'd have 128 threads, a good mix of disk and memory. If you're starting to run out of capacity, just add another pre-packaged building block to the grid. The devils always in the details, but I think it could work. It'd be a bear to patch/upgrade unless it was built into the process to automatically segment the grid into upgraded vs non-upgraded and once a threshold of nodes are in the upgraded status, switch traffic to those nodes en mass. Because no one else builds a T2 based server, you'd have to get all your kit from Oracle which could drive future revenue. If they make the entry point easy enough and the OS/Hardware simple enough, smaller businesses could run oracle products than before.

Oracle (which owns most of the products I encounter in my professional life these days) could optimize and integrate the Solaris kernel to provide added speed and flexibility on Sun hardware. Especially in regards to performance and scalability. If Sun can optimize it's kernel for the Oracle RDBMS or Java for the Fusion middleware stack or Weblogic with out customizing the respective products, it could be a big win for Oracle. I think the trick is going to be to not customize the products. If they do, they open themselves up to the same anti-trust talk that Microsoft faced. We're a long way away from that, but they wouldn't want to give their competition any ammunition. There was a time when the VOS (Veritas, Oracle Sun) stack was the way to go if you needed to scale big. They even created a joint center for a while to work on issues up and down the stack (not sure where that ever ended up). But there's only so much a joint operation can do, because patents and IP rights get in the way. The barrier between two of those three has just been removed (and the 3rd isn't as vital as it once was).

The other area I like is the cutting edge areas. For example Sun's doing a lot with Flash storage and mixed flash/disk storage. What if you could optimize the database to take advantage of that mixed storage pool? Oracle and a native ZFS pool? And perhaps most interesting (and the one that gets me giddy) would be native DTrace providers for Oracle products. Imaging the diagnosis option if you can natively probe an underperforming SQL query? Probe a 'lost' tuxedo session? And then the gui's that could be written to take advantage of the providers? You'd never buy Spotlight again.

Perhaps the most important reason I like the purchase is Sun survives. I selfishly want Sun to live on because I consider myself to be quite good at Sun. I'm fair at AIX and Linux, but Sun and Solaris is in my wheelhouse. I'm not adverse to AIX or Linux, it's just not what I've had the most exposure to. If Solaris goes away, I guess I'd have to go with Linux because I can run that on just about any hardware. AIX requires me to buy something from IBM.

At any rate, I'm glad Solaris lives another day.

I miss the rejection letter

A while back Louis CK was on Conan O'Brien. He had a great bit on "Everything's Amazing, Nobody's Happy." Basically how things are amazing right now and people still have room to complain. My cell phone has more abilities than the first five computers I ever owned combined. And yet I still find time to complain about what it can't do.


Similar thing with the job market. Searching for jobs has never been easier. In good job markets you can put a profile up on job boards and sit back and wait for people to find you. In tougher markets, you can search and apply from the comfort of your computer. No resumes to print, cover letters to write with awkward salutations, no envelopes or stamps. You apply, and 5 minutes later it can be in the hands of the HR department. It's amazing. You can sit in any internet connected location in the world and look for jobs in any major metropolitan city in the western world and then some. The 'web 1.0' way of doing it was a bit cold and non-personal, so sites like linkedin.com have stepped in to meet that need. (My profile's here) There are other avenues as well. EMC has a twitter account for jobs. It's never been easier for job seekers and employers to find each other.
As the Joe Walsh song says: "I can't complain, but some times I still do". I miss the rejection letter. I still get them every once and a while, but haven't gotten any on this latest round of job searching. And I think it's an artifact of living life at internet speeds. So I've sent resume's and inquiries to a handful of opportunities but haven't received any kind of response. Others I have recieved a canned response from the HR application they used to allow you to apply, or an auto-reply email from the HR@company.com email box. I guess I can forgive the lack of response. The flood of potential applicants and the number of dead ends therein would make it a fools errand to respond to them all personally. With the ease of applying for jobs, a person can apply to several dozen jobs in an afternoon. So the likely hood of someone in HR responding to someone who's already taken another job is pretty high. So I guess I shouldn't be surprised that HR doesn't get back to me personally. But it doesn't mean I have to like it either. Some times it would be nice to know that a real person at least received my application. It would be nicer still to know that I didn't get it so I wouldn't be left to wonder.
Oh well, even though everything is great, I'm not completely happy...

A rambling post about twitter, Steve Case and Revolution Healthcare (mostly twitter though)

I've been on twitter for a little while now. And I've had a few thoughts running through my head but never enough to warrant a post, until now. I want to cast these thoughts into bytes for later mockery. I expect when I retire I'll come back and read these posts (in what ever format they evolve into) and make a lot of fun of myself. Kind of like when I look at pictures of myself as a teenager now. The "what were you thinking!" kind of stuff. This could be a long and winding post, apologies in advance. On with the dreck!

So I'm on twitter, mostly as an easy method to update my facebook status, because Ecto (my blog editor of choice) has an option to set your twitter status to notify of new blog posts and lastly (and perhaps secretly the real reason) the pure geeky-ness of it. For the most part, I follow people I know, but I do follow some non-people, like The American Red Cross and a few celebrity types like Dr Tiki or The Big O and Dukes radio show. Recently I got back on the Diggnation bandwagon and decided to follow Alex, Kevin and a few others. Kevin Rose tweeted a note about a new website he put up: wefollow.com which is a twitter directory that ranks by number of followers. Good idea that fills a certain need for folks. I used it to find other interesting people to follow like Tim O'Reilly, Leo LaPorte, Snoop Dogg and Steve Case. Initially I started following people like crazy. Before I new it I had added a few dozen people. I did this late at night so it seems like a great idea at the time. Then the next day started. And I started to drown in an avalanche of tweets. Some very interesting ones, like Tim O'Reilly's or funny like Christopher Walken. The irony of cwalken to me is I distinctly remember not laughing very much when he was on SNL a few years back. Go figure. Anyway, the rest of them were just pure clutter for me. When I finally got back to check my timeline on my phone, I had 375 tweets waiting for me. In less than 12 hours. DOH! So I started dropping people like wet socks off a cloths line.

Brief interlude. Back in 2007 after leaving the Red Cross I was contacted by someone from HR at Revolution Healthcare. They made it clear that it was a Steve Case venture. Basically using his name and reputation as a recruiting tool. At first I wasn't very keen on the idea. Didn't seem like a great fit. But the more I thought about it the more I liked the idea. Healthcare, and in particular health insurance needs a revolution or at the very least a 'market correction' in this country. Steve Case's name could open the doors to companies that would otherwise ignore a startup. The ratio of insurance company people to health care providers in amazing to me (even worse when you look at just Doctors). So I thought working for Revolution Healthcare might be worth a second look. If Steve Case could use his name recognition and access to other execs to open doors, the venture could be very successful. Turns out it a good thing the HR person never got back in touch with me (although I was peeved at the time) since they've laid off a lot of people. It was nice idea. Hopefully they can weather the storm and keep trying. I would have liked to have been a part of it.

Back to twitter. Steve Case. Started following him because he came up in the top-10 for #tech on wefollow.com (doesn't seem to be anywhere on wefollow.com now) and because revolution health care. Wanted to see what kind of exec he might have been. Well turns out he likes the finer things it seems. He's had some expensive tweets (well at least for a lower pay scale guy like me). Pics from Sunset in Captiva FL, Advice on where to stay in Maui. Starting a resort in Costa Rica. Then there's the @user replies to people I don't know. Kind of weird only seeing half of a conversation. But I kept with it for about a week. Today, can't take it anymore, have to 'unfollow' Steve Case. Not that he knows or cares mind you. I wouldn't in his place. But he wasn't saying anything that was particularly interesting to me, especially in the context of what I use twitter for.

Which leads me to another Twitter thought. Twitter needs a feed filter ability. Not sure I've thought this idea out completely but I need two twitter feeds. One of my 'a-list' people I follow. Close friends, businesses that announce stuff via twitter (like MacHeist, w00t or Red Cross) and the like. Then the B-list twitter where all these other folks can go. I could then follow up with their stuff when I have spare time or an interest. Maybe another way to look at it is 'push' the tweets from group A but I'll 'pull' the tweets from group B when I want to see what they had to say. I'm sure there's a better way to say what I mean here. And who knows, maybe you can already passively follow people and not have them clutter up your timeline and I just haven't figured it out yet.

Anyway, time to get back to the business at hand. Time to push 'publish' on this ramble and have it announce via twitter, go figure.

IBM to buy Sun?

According to the WSJ IBM is in talks to purchase Sun. The MSNBC version of the same story provides a lot more detail. There's a bloomberg article that seems to be down as I write this that indicates that Sun asked HP they turned it down.

One thing that strikes me is that all of this is coming from one source, so there may not be much to this. This could be just a bunch of speculative talk. This same speculative talk also indicates that HP was asked and HP turned Sun down (which is a bit surprising because HP and SUN would be a better fit IMHO).

But if we pretend for a moment that it's all reasonably true some interesting things come to mind. The cross-country nature of the two companies will be a big issue. It's east coast vs west coast. It's suits and ties vs pony tails and sandals. IBM's cathedral approach to Suns bazaar approach. On top of that you can't have HQ's on both coasts. Which one goes? In the end, IBM will 'win out' eventually just out of shear entrenchment, in my opinion. There just isn't enough Sun left to change things. There is the possibility of it being a separate entity, similar to EMC/VMWare. That's likely how they'd start, but eventually they'd restructure under the IBM way of doing things.

Another interesting choice is what becomes of the hardware lines long term? Do they push SPARC out in favor of POWER? Do they do both? Short term they would try both I'm sure, but eventually they would have to drop one or the other. I suspect they would drop SPARC. The high-thread low-Mhz model of the T1 and T2 chips doesn't have as much traction as I think Sun or IBM would like. The POWER line on the other hand takes the same tact as the SPARC64's but with much higher clock rates. Perhaps a T1/2 and POWER lineup would be workable. But it's a tough choice. Neither CPU has the broad adoption of the X86 instruction set CPUs of AMD and Intel. I think in a merged company the sun x86 lines would be gradually dropped within a year in favor of the IBM setup. Hopefully a 'best of both' approach would be taken but I doubt it. As long as the OS's could be ported to both, any of the possible permutations and combinations would likely be workable, with the possible exception of keep everything as is.

The AIX vs Solaris issue would also be interesting. Sun's current sunset dates (if maintained) would have IBM support Solaris for up to 5 years after the decision to stop Solaris. If they decide to drop AIX they're faced with a similar issue (although IBM's EOL policy is a bit more cryptic). Either way you slice this one there are going to be a lot of hurt feelings. The AIX vs Solaris camps would be bitter enemies if it weren't for the common Microsoft enemy. My hunch is Solaris would be taken out back behind the woodshed and not come back. It would likely happen after all of the corporate structures are merged and aligned. Then Solaris would be stopped and open solaris would be the road forward for Solaris hold outs.

The storage area is the only area that seems straight forward to me. And is the only technology area that IBM and Sun can merge easily. Since neither company is storage leader (although IBM has a much larger market share) merging the product lines would be relatively simple (compared to OS or hardware). Neither company has a good story to tell in the mid-range market (although the 7000 line from sun is interesting, if not proven). The high end market is dwindling and I think both companies will not invest a lot in that space and leave it to Hitachi, EMC and NetApp (I exclude HP because the XP is pretty much a HDS...). Combined they could also aggressively go after Dell's market share (and nibble at HP's). Dell is growing by virtue of selling tons of little storage, basically doing it on volume. And since the high-end Sun arrays are re-branded HDS gear, the just have to maintain the support infrastructure. Plus Sun has the better play in the emerging flash market and IP knowledge that would come with that. Add onto that the StorageTek side of the house and it's an idea with some promise. So far, this would be the easiest (but by no means simple) merger.

The other software assets could also bring some value to IBM. Mysql and Java would join IBM's already impressive web offerings. Add Sun's middle-tier and middleware offerings to that mix and you have a very expansive software portfolio. And then there's the Sun ports of Xen and purchase of virtual box. Both of these virtulization offerings would give IBM a small counter to the VMWare juggernaut. And then you can add Open/Star Office to the Lotus brand (when was the last time you saw a SmartSuite purchase?) for an enterprise desktop solution. After a bit of blue-washing the code base and essentially free access to the IBM patent portfolio these products could really take off. There would be a few orphans here and there, like NetBeans. Not sure where Netbeans fit's in a Rational world. But the software portfolios could be a great combined force.

So far analyst reviews are mixed but trending towards 'thumbs down' and I can't say I disagree. The short run would be dreadful and IBM would have to walk a delicate line to avoid loosing Sun's existing customer base and its massive and active development communities. If end users and developers are going to be forced to change, there's no assurance that they will change to IBM. If you have to go through the trouble, you might as well look at all the options. Well, so far it's just idle speculation. Time will tell where this ends up.