tallsails

Why MSFT wants to buy ! Yahoo!

TallSails — Thu, 06 Mar 2008 12:41:38 GMT

A lot of folks figure MSFT wants the eyeballs, ability to generate a profit on the web. Well thats the obvious short story. What was obvious to me the moment i heard it, was MSFT was going to be able to fulfill a goal of thiers, which they have stated, in the only way I think they can....

MSFT has entered the era of realizing cloud-computing (thier term) or grid computing. There are a few companies out there that have massive scale data centers - Google, Yahoo, AOL, Ebay, Amazon. A few others, not many. And when I was at a MSFT developers conference this past fall, they announced they were going to build many of these massive data centers to join this business - and become a modern day telco.

Only, they are waay late to this game. Nothing Scales. (tm) How you heat, cool, layout cables - this is all very specialized at this scale - how you setup install, de-rack, maintain a massive data center (lets not forget all the cool things about redundant cluster computing with fail in place components, etc. Just try to build one of these suckers - )

So how can they catch up? Buy Yahoo. The street will make them pay for the online ad revenue, a profitable business model, a web portal and the number of yahooligan eyeballs, etc.

But the street won't charge the premium its worth the MSFT - how to build, run, massive scale computing, which there will be 3-5 (rule of 3) players in the coming years, and MSFT needs to be one. Or they will be bought by one.

So they want to buy Yahoo. So far its not been working, but I now realize how they plan to get into massive scale computing - which I didn't believe they could....

Multi Core CPUs and micro-scopic biology.

TallSails — Thu, 15 Feb 2007 10:00:48 GMT

By now even my ma has heard about the multi core, multi CPU revolution. 2 Core CPUs are the norm on shipping laptops and towers, and 4 core CPUs are just around the corner. Dual and four socket mother boards, which allow for 2 -4 CPUs (each with multi cores) are no longer exotic, and the memory and OS, and database are engineered with this computer-factor in mind. Soooo.... now what?....

First, many more cores. There are 8 and 16 core chips already in the pipeline. And before you start yakking about dimishing returns - let me point out a few things. First - a "context switch" where the CPU is running one thread and switches to another, is the slowest thing on the planet. So just having a core "sitting" for that thread, with all the registers set, and switching among cores for each "context switch" is an easy way to ramp processing power.

Second, we are not done improving the interconnect, cross bar, technologies between all those CPUs and sockets. Research labs are playing with chip sub sets that don't even have traditional circuits between them, the circuits are so close together that so much noise is introduced, the sub chips effectively train and communicate similar to error correcting modem techniques, and since the circuits don't always align, the systems have a sort of "roving" cellular tolerance to the communications.

Wireless communication from one part of the motherboard to another is also another "in the lab" interesting idea. You are not constrained to the two -dimensional plane of circuits...lots of interesting things can happen.

So lots more CPUs? You bet.

What else tho?

Well, look at what the chip companies are doing, and look at mainframes in the 80s and lets play 2+2=?

AMD bought one of the largest and most advanced graphics cards companies. Now a graphics card is a really hardware pipelined vector array math card. AMD has already announced they will package that on-or-near the CPU. Also recall that there are already math packages (and screen savers, folding@home, SETI, etc) that will "do the math and processing" on the graphics card (not the graphics math or processing, general purpose stuff).

And I have read hints from AMD and INTEL that point to a non-homegeneous (ahem, mixed irregular) cores.

So your 80 core CPU in 2010 will have some cores that are optimized for floating point, some for communication, some for integer math, some for large memory applications, some for I/O bound subsystems and subroutines (mostly waiting on the bus, but can spend those cycles re-optimizing thier assembly code) and other cores that are dedicated to "helping actual cores" - these background cores never run a user or OS thread, they do things like re-order instructions and re-optimize (hot spot) instruction queues for cores that are dedicated to running your user threads.

IBM is close to this with thier "cell" processor, from cellular automita, but from what I understand, all the cell processors are the same, they just specialize thier workload and assignment. Think about the "array processing boards" some mainframes would install to run SAS...

So you have a bunch of vector processors, and groups of risc cores, each risk (reduced instruction set) group optmized for a kind of processing. And your programs, sub routines are allocated to certain cores that are suited for that kind of processing.

Sounds a lot like the human brain / nervous system eh?

Sounds like technology evolution is mimicing biological evolution. And right now, we are still in the simple C-H-O-N (carbon, hydrogen, oxygen, nitrogen) chain state, perhaps up to the single cell microbe state.

The "technology" life forms coming will be much more complicated, much more powerful, and each phase will need the processing power of the previous to design it. All but the simplest microscopic life forms have different cells dedicated to different functions. Thats how early we are in the technology evolution. Single cells Amoebas and Euglenas.

Thats where its headed, and everything from aquisitions, to lab work, to announced pipelines foretell it.

(And you thought I was going to rant about how multi core CPUs are helping medical and biology research.)

The good news is, this will greatly expand the kinds of problems we solve with software, the "circle" will get larger.
And, it will give the operating system vendors something worthwhile to focus on for a good 10 years, instead of adding and removing dancing paper clips from our screen.

Some companies won't take your money

TallSails — Thu, 15 Feb 2007 15:02:39 GMT

We got new laptops at work. Dells, corporate thing. So I decided to look up a few things about my dell laptop, and check out thier online store, see what the everything-old-is-new again CEO Micheal has done.
Micheal, visit your own on-line store. Yourself, with no one in the room, you will be shocked to find out....

First, I have to say whether I am small-business, medium business, large business, or government entity. Those are the only choices. I had no idea that large business buy laptops that are wildly different than small businesses, but what the heck. And what of the moms, Dads, college bound senoirs, grand mas, grand dads, video gamers, web surfers, unemployed but looking for work? What about people that might want to buy a laptop? Umm, no door for you to go thru.

I want a laptop. You only sell like 8 different builds. Many of your customers are not any of these, they are dads, moms, college bound kids, and plain folks buying thier own stuff - or do you not sell to the common folks anymore?

Second, supposing you guess that right, you then have to pick from coined names like "inspiron" and "lattitude" to further "guess" where to head. I figured i wanted lattitude, but I don't know the difference - which is high end, cheaper, student model etc. Thanks for putting me further off, making me feel more lost.

But the kicker is this page I finally get to, after my third choice, where I pick a model number of a lattitude, still not knowing what the heck a 620 varies from a 440 lattitde....but nonetheless I finally! get to see a laptop and some (some) specs.

Here is the page

Study it carefully, cause at this point, moms, dads, college bound kids, you know how much money you would like the spend. 3 guesses of navigation in, and still - NO PRICE. thats right.

Now, even worse - you can get the price - just click on the BUY IT button. Every web catalog has a buy-it button don't they.

Not this one. Not even close.

If you are a purchasing manager at a large government entity, maybe the
"Quote to order"
button makes sense to you.   Is that even english? I don't want to quote anything. The verb is what I do (print, copy , paste) - I don't want to quote. I want to "read a quote" "read a price" , "View a quote". I may or may not want to order, and perhaps that next screen makes me order it. Or worse. (actually, worse).

You finally figure out thats where you want to go.

only you can't get a price here either!!!   You have to "register". And if you try to "guest checkout", which works without registration at every other place I have bought stuff on the web - No price, no dice, no sale, it asks you for your previous quote number!!!! (Which means, you are already registered? - did the designers really test this?)

You are in an endless series of mazes which all look alike.
Go North.
You are in an endless series of mazes which all look alike.

I was pretty familiar with the Dell web site of old, and you could get products, many, side by side, with prices, very easily. I know the "Web 2.0" consultants tell you to put "role based navigation" on your web site - but you forgot a serious role. I want pricing info. I can't get it. I have X dollars to spend. I don't want to "quote to order".

Price. Buy. Features. These are navigation buttons I want.

Turns out they exist.

They are hidden under the green "customize it" button in the flash movie. Only I typically don't navigate or pay attention to flash movies, when I have them enabled at all.

So, If I have "been trained" on the dell navigation, work at a large company, and know what "quote to order" means, don't want to price shop, don't want to comparison feature shop, and have flash enabled and like watching flash movies, and know that "customize it" also means "prices and features hidden under here" (even without customizing....)

You have one very nice web site.

Only I am not that above person. I am used to your old site, and about 50 other web sites, and your competitors web sites - They have buttons like "Price", "Buy", "features" , "compare similar products".

Different is not always good. Sometimes we like the gauges just where they were the last time we went flying.

As it turns out - the old, line-up-the six laptops we sell, with features and prices screen exists!!!

You just can't get there from the main dell screen.

You have to navigate "guessing" as described above, to some product, then once on that product, the tabs at the top (which make much more sense BTW) have a tap called PRODUCTS -> LAPTOPS. and bam there you are.

Right where you wanted to be, all along. And I would argue, this is where every Dell customer wants to be, no matter if they are a small business, medium business, large business, or government entity. And even if they are a student, mom, dad, grandma etc.

Nothing makes a person feel less wanted than having 4 categories as the only way of navigation that they don't fit into.    When you started your company Micheal, people bought your computers. Start selling to people again.

Many of us also work at businesses.

update 2/15/07
Turns out, if you say you are "home and home business" you only see the two low end laptops from Dell. Too bad anyone that is, or thinks they might be, this.

And even worse, if you say small,medium,or large business, to actually see the higher end laptops (Who would want that? PRogrammers? Your future tech purchasers?) - - YOU DON"T SEE PRICES.

THats right, large business people, or anyone who says they are (for purposes of browsing your store, to see those computers) don't get to see any prices. Not list, not retail, no comparison possible at all.

So on that "See all 8 laptops on 1 screen" screen, you may or may not see prices. Depends on if "you should".

Wow. Cause new people to dell, be they large or small businesses, never want to check prices.

This web site might work for people used to ordering with Dell (well, not the ones I just talked to) - but it sure is a bad way to aqquire new customers. In fact, its a multiple-slaps-in-the-face way.

Steve Jobs lets out a big secret - How to innovate

TallSails — Wed, 31 Jan 2007 11:14:23 GMT

The Iphone is already a hit - its an innovation, rule-changing, category killer , and its not even shipping. But what I find most interesting about the keynote and Jobs interviews is when he said (rough quote) "We were going to partner with a different phone company, but wanted to retain total control of the hardware and the software". Thats a gem we can all consider when trying to foster innovation.

Why do "startups" seem to be easier places to innovate than large corporations? Why does "incubation" or a small off to the side team sometimes lead to innovation inside large companies, but sometimes not? Why are the Xbox, the Ipod, the Iphone, the Tivo all successful at being rules changing innovators? (And yes, I include the Xbox, while I don't have one nor a Tivo.)

The Xbox did convince consumers, as well as Sony and Nintendo, there is a market for expensive game consoles. The Tivo broke ground on all the problems with VCRs we have been "putting up with" for years. Same for the Ipod. And it dawns on me - to innovate, you want total control of the entire "stack" - hardware, software, OS, etc. You can't innovate with too many other parties, other companies saying "no , that can't be done" or spending tons of time convincing them to add a feature "no one else wants" (by definition, you are going to try to do that when you innovate...)

Which also explains why its hard to "outsource innovation" or rather - innovate when parts of your stack or process are outsourced. Its a total contradiction. No phone company would agree to all the hardware demands Jobs made to make the iphone what it is. So he kept total control.

Innovation comes from talent, and isolation from executives, third party suppliers, anyone who can say "no". And you have to control your entire stack to truely innovate - if your OS, hardware, drivers, anything, comes from outside your company or work group - its hard to innovate - you are slowed down, beaten into marginal tweaks and improvements, and also your ability to keep things trade secret are greatly reduced.

There are hardware demands that the xbox software teams needed to make the xbox what it is, same for the tivo, ipod, and iphone. And they would not have been able to get where they are without total control. This explains a lot of the new "devices" that we all love, and where they are, and are not, coming from. We are starting to see a lot of them (not a newton or palm every 5 years, we are seeing 5 new devices every quarter...) because the costs and intellectual experience curve to control the whole stack (hardware, connectivity, software) are getting easier, the development and design environments are getting better.

Look at the Zune - a compromise between what the customer wants , lots of (too many) folks at MSFT want, and what the recording industry wants. Loses out. Ipod - more of what the customer wants, few folks at apple, very few demands of the recording industry. Ipod wins.

Too many cooks spoil the soup.

It also makes you look at companies like SAP and DELL differently - as we move into the decade of innovation and intellectual value, not production value - our service based economy matures - companies that don't own thier entire production and engineering and stack have trouble innovating. Folks along thier production "supply chain" (that we all started in the 90s - the extended corporation) can borrow your new ideas into your competitors products as soon as you ask them to place them in yours...

Its a big failing of the "just in time", "outsourced", "distributed" corporation that has become popular. You are dependant and controlled and influenced by these partners. You can't innovate in that situation.

Iphones, Tivo, Xbox, Ipod, and Java - all great game changing innovations - partly because they had control of the whole stack.

There are a number of technologies that are getting so easy to use, a single small team of geniuses can handle it, from design, hardware, software, to application. And this contentration of talent, of innovation, is going to really change the product space in the next 20 years. Breakthrus in ease of use of the prior list, robot platforms (roomba ships a platform kit now), and nano technologies are going to cause an "explosive" innovation age.

The "guilds" are being broken up.

So how does this affect my trying to deliver software, and perhaps yours?
Own, understand, and control as much of your "dependency stack" as you can. Open source OS, languages, platforms, tools. Give your teams this "freedom" to innovate and not be curtailed by your "partners".
And keep the number of execs, and thier wishes, to a minimum, and to 0 once requirements are frozen.
I have seen many a project in my days what went run-away from the hidden cost, of the quiet requirements that never stopped, from executives that the team did not want to "say NO" to.

To many cooks, and the soup never ships....

Too much Magic spoils the project

TallSails — Tue, 16 Jan 2007 17:32:14 GMT

A lot of writing good software is, pure magic. A good design that can hold up to (anticipated?!) changes, innovative "ways of seeing" - or organizing a GUI, workflow, data elements - how to attack the problem. Its what makes the decision, optimization, and simuation solutions that my company writes, to me anyways, more interesting than simple database transactions. And there is true magic in how we develop parts of these appplications. Too much magic, and the project never ends.

See, thats the problem. Take one of our simulations, that operates the entire electric grid, generators, buses, powerlines, fuels (delivery, burns, accounting), pollution, weather, load, new generators being build, price forcasts - the whole enchalada - there is true magic in how the center of that engine works.

But just outside that, there is a large engine of simulation, that while advanced in its design, after our company has been writing this kind of application for over 34 years - - we have some pretty time tested patterns and methods to get it done. And just outside that is a nice big I/O engine, and GUI, and other - -impressive, but not mystical stuff.

And the non mystical stuff can be reliably estimated, clear requirements can be written, test cases can be documented prior to construction, and testing can occur. CMMI level3.x, nothing to heavywieght in process - but enough to provide progress transparency to management and our customers.

problem is, your best programmers focus on the critical center of the engine, the magic part. And they should, its high risk and mentally challenging. Only don't use the "process" (or lack of it? ) - iterative discovery and eureka! moments, for that part of the application, as your process for delivering the other 1-2million lines of code.

If you use too much "magic process" - your project may never end. You won't have transparency into where you are , what parts are running away in cost and time, what parts are not being vetted in requirements and design with others (we can find our design errors in the field! we are manly programmers!).

Magic is a nescessary part of our applications. But just a pinch will do.

Will Google exist in 10 years?

TallSails — Thu, 11 Jan 2007 14:35:06 GMT

A lot of folks, from investors to technology analysts, and CTOs like myself that have to gauge where things area headed - ask - will google exist in ten years? Its a short and simple analysis.

First off - what keeps Google alive? Stock hype - not anymore. They are making money quarter to quarter - mostly off adsense and other web advertising. They have made a market in monetizing web eyeballs. If you have a web site, and want to make money off it, the quickest easiest thing to do is place google ad-words on it, and they pay you revenue for views and click thrus.

They also make money off folks wanting to drive traffic to thier website, utilizing traffic driven from search and other google apps.

So they can persist, as long as folks continue to need those services and continue to go to Google for them. In fact, thier cost structure is going down, and demand for these services is going up (See my entry "the google curves" which details this).

So displacement by someone else is largely the risk here.

And you have have read of recent folks coming up with "better search" , powered by "social response networks as well as algorithms". Sounds great. It could actually be better search.

But google will still be here, big as ever, in ten years.

Even if 5 times in the next ten years someone actually does come up with better search, better email, better calendar.

For many of the same reasons that Ebay and Amazon are still here.

Cause having a capability is one thing. Being able to deploy it, re-deploy it, support it, and run it on a massive "millions of clicks" scale, is something else. Nothing scales.

That is to say, you can't just take 1,000 servers running the same thing to serve 1000x more customers. Their searches affect each other, etc. You can't generate hundreds of (usable) megawatts stacking portable honda generators. Massive generators are built differently.

So that little upstart with "a better search idea" has a lot to learn about massive web applications. And customer support. And first mover advantage. And compelling me to learn a new place to go and re-type my contacts for gmail, etc.

But its the massive web scaling that keeps amazon amazoning, and ebay ebaying. They know how to run massive applicaitons - and I have had dinner with technical folks from Ebay, and even over dinner its jaw dropping impressive how they do it. And if you want to know why google is so secretive - its a hell of a trade secret.

Google has the mindshare already, first mover advantage. Yahoo , Google, MSN, Ebay and amazon are here to stay.

I asked one of my top notch developers today which would be easier given $50million - improving Googles search to include social feedback, or deploying from scratch a new compute farm and search engine that can handle google sized traffic. His answer was quick - improving Googles search algorithm. Something google has done dozens of times already. Oh you didn't know that? Thats the beauty of web applications. They keep chaning under thier veneer API of the screens you and I see.

If you don't think google can match someone elses search algorithm improvements far before an upstart can manage to run a massive server farm, just ask jeeves......

Google is a confluence of cash, genius of simple web applications, massive web application deployment infrastructure, a large number of smart people, and a market that can monetize web eyeballs.

Having a decent web application means you only have 1 of the 5 things google has, and the easiest one for them to match you on.

Which is why Ballmer is not sleeping nights. He has only 2 of the 5, perhaps 3 if you count MSN and hotmail as a mega scale web infrastructure.

Its already happened and you missed it

TallSails — Fri, 27 Oct 2006 20:58:32 GMT

Psychologists tell us that our "self view" lags several years, how we physically and emotionally see our selves lags to our prior self. Management training is that it takes people several months to start performing a new job, they tend to perform thier old one for a while.
And so todays rant is about some things that we still see in the old light, but have clearly changed - what we spend our IT dollars on....

Reliability
Folks will pay for reliabilty. Ask Toyota how they won over the US car market. Ask yourself why people pay a premium for Apple MaCs and Ipods - they just work, and the customer experience is nicer. I gladly paid more for a camry in the 90s so I would not have to deal with car breakdowns. And so it goes with IT equipment - lowest price is not going to win. The internet, electricity, compute cycles critical to your business - part of thier compelling attractiveness is they are always there.

Try going for a few days without electricity or your cell phone.

The counter to this is, folks won't suffer un reliability, given other options. Ubuntu is an option, Mac OSx, Linux, Solaris, and people are using them in droves to gain reliability.

I am reminded of the joke the Navy flight deck crews tell the pilots as they close the hatch, just prior to catapulting the jet and pilot off the deck - "Remember, each component is built by the lowest bidder..." Not something you want to tell your CEO when they ask how their IT infrastructure decisions are made....

And on the people side, someone on my staff that I can count on 99% of the time, is more valauable than a 50% hit and miss genius. Well, in most cases.

Support
Look at how much we spend on monthly support, on IT. Initial purchase costs of hardware and software (often 0) are dropping all the time. The cell phone, the server, the database, the OS, are free - - its the support you are willing to pay for. And its really nice (Thanks Larry Ellison) when more than one company offers it - price and performance competition is a great thing for the end customer.

Energy - power
Lets plot some data center costs- 1970 - 2007
First plot the cost of a CPU cycle, 1 floating pt instruction.
Then the cost of RAM, 1 MB.
Then the cost of Disk, 100MB.
Then the cost of Tape, 10Tb.

Then the cost of Electricity, or power, say - the price of a gallon of gas or a barrel of oil.

Extend the curves.

It is a clear certainty, that IT spending will be dominated by energy and support costs. No question of if, only when.

And it may have already happened, and we are just getting around to noticing it.

What the DELL is going on here?

TallSails — Thu, 01 Feb 2007 09:15:24 GMT

Updated at end of article

0.6.2006
This weeks news, HP sells more PCs than Dell. What happened to Dell? Lets take a short historical look at thier value-add, thier reputation (from one persons, mine, perspective) and look forward - and it isn't going to be pretty going forward....

(Disclaimer of context - I was a Dell fanboy for much of the 80s and 90s, made a LOT of money from thier stock, from the PCs limited days thru the late 80s. Totally divested in the past 6 years of it. And feel pretty firm about it at this point.....)

In the Late 80s you bought a PC from PCs Limited (Dell) because

Thier catalog and phone ordering was the easiest to understand and customize
It was about $1000 cheaper than IBM, HP
It was more reliable than local-built shop ware
It was faster and about the same price as building it yourself
Support was great
It came with some useful software
It was a reliable PC

In the early 90s you still bought from Dell because

Thier web site was the easiest to use and to customize from
Thier support was still good
They were about $500 cheaper than IBM, HP
They were more reliable than local-built show wares
It was now more expensive than building it yourself
Support was OK, not great
It started to ship with tons of pre-load software you needed to remove / shutdown
Its reliability was average

In the late 90s and early naughts (2000+) you bought from Dell, but

Other websites were getting easier to use
Thier support reeked and was off shore
They were often the same price, or $50 to $100 cheaper than HP IBM
They were same reliability as local-build shop wares, perhaps under HP IBM
More expensive than building it yourself
The pre-loaded software was really a problem, especially for your mom who ended up with 3 AOL accounts
Reliability was average

And then, you bought from someone else for the late naughts because

You were familiar with other easy to order websites
They were the same price as HP, IBM, SUN
More expensive than building it yourself
The pre-load, including the OS, was of no use to you, you switched to open office
Dell support reputation still bad
Reliability reputation still bad
You wanted dual core, 64 bit, or Opterons before they were outdated

and thus, thier hold on you was broken. Several mis steps. Where is the Value-add, the offer-advantage for Dell?

Are they the cheapest on price - no
Are they more reliable - no
Support better - no
Easier to buy - no
Novel features or innovative technology, or intellectual property - no

So dell is in quite a pickle. Companies like Apple, HP, SUN are innovating and leveraging thier IP - - thier products have great end-user first experiences. Compare your first hours with a Mac over un-loading your payloaded drive on a Dell PC.....

Why were they so big ?
Dell was a great supply chain optimizer - and you recieved quite a large amount of savings and better components, more reliability than local shop builds as a result. But unlike Andrew Carnegie, Micheal Dell did not spend his billions buying up his vertical - those shops that assemble his PCs in the morning, are assembling Lenovo and Fujistsu laptops for the afternoon shift.....the rest of the world caught up.

But the rest of the world has engineering, and design capabilities, and Dell is late to that game, and its not a 2 year effort to enter that game. And that experise, that value-add, that intellectual property is core to Apple, SUN, HP, (IBM) Lenovo - they don't out source thier offer advantage. Dell did.

Its OK to out source your software (eek I am a software guy) efforts if you are a bank. Or your accounting or boxes if you are a choclatier...but if your central IP is software, or making chocolate - - keep that in house and protect and improve it (for your company) over the years.

Can Micheal Dell keep improving his supply chain - yes. But the economies of technology are reducing how much the hardware cost, and how much the services costs are. How much did you pay for your cell phone, or TV set top box, vs. how much you pay monthly to those companies?

Does anyone pay Dell anything monthly? Some service contracts on servers, yes. But they don't have a great reputation there......Its not even in thier corporate culture "we will win with better service"....

Solaris is free - your cell phone is free, many databases are free, Linux is free, great office suites (open office) are free - its the monthly support that you are willing to pay for. And the economies of writing software help - they only have to write it once and can distribute it to millions, so they can charge $0 for it. Everyone thought the initial Ipods were losing apple money....but the Itunes music store topped a million downloads far ahead of anyone's expectations....

And here is the big "why I don't own dell stock anymore" grenade....breathe deepy before you read on.

Microsoft competes with Dell.

Has Micheal Dell figured out that every Xbox that ships is one less PC he sells? (Its a thinly guised PC).
High end gamers used to order from Dell to a large extent....
Not to mention PS2s, PS3s, Nintendos. McDonalds realized in the 80s that thier biggest competitor was the microwave oven, not another burger chain.

And will MSFT stop with Xbox? Not hardly. Apple ships PCs. Windows is losing its lock-in abilities and will generate less and less revenue over the coming years. MSFT will start shipping more "appliances" (ZUNE?) and these are computing devices. And finally thier own branded servers and PCs. No question. Similar to the phone companies branded cell phones - a full MSFT preloaded laptop. With services revenue.

I use web based services (like right now blogging) more and more - and need less and less a PC to do it. My 5 year old 256k RAM PIII does quite nicely for gmail, playing web based games, running open office. The web has killed the high replacement cycles of many PC users. And this changes the MSFT-INTEL-DELL revenue picture considerably.

And Wall street is slowly figuring that out. Web services, web portals, web PCs. Web PCs almost never need a rolling upgrade (one of the really nice things about web services). (Got that one from Mcnealy.)

Throw in some spice of flaming batteries, Bad service reptuation, HP selling more PCs and even the street is wondering what the DELL is going on?

But thats the small tip of the ice berg. The real under current is that Dell has nothing special anymore - - other vendors meet them on supply chain, ease of ordering, and beat them on engineering and innovation.

And the web services trend has reduced the PC upgrade cycle considerably. Sorry Micheal - should have read about Carnegie and owning your supply chain, or found some other value add with those billions. Its gonna be a tough tough row from here on. Building your own engineering eh? HP, SUN, Apple, IBM have a big head start.

Its a tough world out there. I am not sure with HP, Apple, SUN, IBM and MSFT shipping servers and PCs you need a Dell for the eco-system to work well. MSFT is your competitor, and by the way a popular component in your product. Thats gonna make it doubly tough.

Its a shattering prediction. Dell will cease to exist or start selling other folks products, or be bought out by MSFT.
Thier current value-add is not enough to make it in this jungle.

Update - 2/1/2007 Mike back as CEO, and Dell prepping a hand held game device
the register reports dell is prepping a hand held game device to go up against the Nintendo DS, and PSP.
Whats my take? If the article is correct, its a winMobile PC with 1 screen. I already own that, its called my cell phone. (Granted my video, look and feel, graphics speed are likely less). But it has no new compelling "Gots to have it" feature (that I can tell). - Wii remote? Dual Screens? Large library of games I already own?

Now, if its a portable version of the XBox, then hats off to Dell.

As to Mike back as the new CEO, there is a saying about an old dog and new tricks....unless Mike has some ambitious new plans that have something in addition to pipeline supply management (which his company is already world class at...) - - I have my doubts

We are always hiring great people - and why the anonymous blog?

TallSails — Wed, 06 Sep 2006 09:27:13 GMT

Joel on software has it pegged - there are only a few ways to hire great people - and going thru resumes in response to a huge board posting is not one of them. And my company is always hiring great people. And since its a natural question many have sent - why the anonymous blog? Well, the viel lifts a little ....

First, I think about the really great developers I have hired - many "community referrals" - great nerds attract each other - that covers about half of them. One other was a referral from a great prof I had getting my Masters at Ga State (clue #1). Another was a resume from a post we did at Auburn (Clue #2) - - and many others were right out of college or co-ops and interns with my company - go to the mountain.

We had a great hire from a recruiter - he left in one year. Cause people that had been here 10 years were treated better. But I may be simplifiying things....

Joel was right.

Why the anonymous blog? First, it slows the solictications down (folks selling me copiers, phones, paper, desks, staff - none of which I need from these sources, much less the aggravation. Stock brokers and recruiters - thank you I am well taken care of already - I have family members on both.) Second, the mega-corporation that bought our little company a few years back has an unclear blog policy. (By unclear, I mean over 40 legal pages. really unclear). Call me a wuss. And a distant third, I am not using this blog to troll traffic to our company or its website, so it gains another hundredth of a percent of validity.

But - - I can always use more great programmers - so feel free to send a "comment" to this website which won't get posted (they are moderated, by me) and I can email you, if you are interested.

More details if you might be interested:
Software company that has been around more than 30 years.
Headquartered in Atlanta, software groups in Houston and Atlanta.
Benefits - you get to work with some of the brightest people on the planet.
Like Folks from the java posse

I don't want tired, "I know THIS syntax" folks. (Notice no syntax is listed? How many of you are using the same language and dev tools for more than 3 years?). I don't want "cherry picking - give me a raise" folks either.

If you really want to work on massive scale software systems, and J2EE development, then maybe this is a good place for you.

You must be smart, and have a good work ethic. We can prop you up to the firehose and teach you the rest.

Some things are clear

TallSails — Thu, 24 Aug 2006 10:09:38 GMT

Its a crazy time in IT right now - spending is picking up from the post Y2k draw back, Companies are finding out how to leverage the web, how to make money on and via the web, and the "services" Web2.0 bubble is upon us. I haven't blogged entries lately cause its hard to say where things are headed - but some things are very clear. One of them is, Dell-Intel-MSFT - while still huge players in the industry, no longer control it. There are signs of this everywhere.Its a different thing - being a dominant player in a massive industry that you don't control, from being the controller. And all 3 are having trouble making the adjustment.

How crazy is it?
Here are a few things you would call me bat-headed crazy for predicting even a year ago:
    MSFT helps Firefox work well under Vista pre-release
    Dell starts shipping AMD chips, even in laptops
    SUN sells more servers than DELL

Here are a few things, that everyone predicted, but still make it a crazy time in this industry:
    VISTA delays (again?)
    Apple Macs gaining traction outside the "creatives" world
    sub $1k laptops that perform well


Web2.0 Bubble - services!
There is a lot of Venture cap money that is headed into the web services startups right now. Those bi-zillionaires just can't hold cash for very long. There are clearly a lot of web services that are not going to make it - and the economies of transactions will kill them. Sounds great - lets start a web service that can be a small part of several larger web services. Only, if you are the web-calendar-reminder-queue-mgt service that netflix or audible uses, how much of my $10 a month are they going to pay you? (Answer - not much. They are going to write their own, due to the economies of scale of the massive customer base of the web/planet, and make more money that way.)

I am reminded of the "shopping-cart on-line catalog" infrastructure vendors of the late 90s - google "calico" for a great story. They made a killing, selling their "get you up and running" infrastructure (lisp based, loosely, could handle complicated on-line component-subComponent configs like the DELL store (early versions)).
Nowadays - adding a shopping cart to a web site is....free in most web IDEs (even on line ones.....that come free with your DNS registration....just to get more registrations....) Adding a catalog capability is also nearly free.

And far before that, even DELL realized they had a large enough customer base to start writing their own online cart and catalog systems, and cut the ties to Calico....(calico Stock shoots from above 100 to below $1 in the first web bubble). Now calico was also victim to the IXL like accounting of "keep signing lots of fixed price contracts for custom development" - this works as long as each month you sign more contracts than last month - but eventually all those projects run out of money long before they run out of work and commitments.

My company does custom software work in addition to license sales, and we watch those projects and commitments closely - and the sheer volume of them compared to our other revenue. Call it risk analysis.

So calico and IXP sales pace drops a little and their company crashes. Especially if you throw in "commoditized creeping price pressure" - everything that is custom and expensive, but lots of folks need, becomes cheaper and a commodity.....(see Linux, Solaris, IIS, for examples - are you listening Oracle, MSFT?)

Many services companies will make it (don't everyone flame this). Many will sign too many projects, get hit by price compression, be immature in managing projects for profitability, and fail. Many will find out thier prospective customers (ebay, amazon, salesforce) can write thier own services due to the economies of thier massive customer base and transaction loads.

Dell-intel-Msft troubles and adjusting to being "just dominant"
Intel held a recent briefing where, when questioned about their "evolving" roadmap, they said they didn't think anyone needed 4,8,16,32 core CPUS.

Sounds a lot like the familiar quotes of the big 3
    No one wants AMD chips - dell
    No one wants multi-core (dual) CPUS - dell, intel
    No one wants 64 bit - Dell, Intel, MSFT

Well, we do.
We really want 64 core CPU chips. We could use them today. And once we have them, we will start to exploit them. I could make a pedantic list of all the tasks (43 of them) running on my laptop right now - and this is on a single core CPU. Image what I would be running "oh but if I could". I am talking the normal PC user, the business user, and the scientific computing user. (I am all 3, in a single day...)

Home video, photo editing? Give me thousands of CPUS - this is an easily paralleled processing problem.
Home video watching? Need lots of CPU to compress, store, render several HD channels at once.
    (its nearly 2 or 3 channels per household member at some hours in our house)

"640k ought to be enough for nearly everybody" - Bill Gates, circa 1980s.

Don't make bounding predictions in this exponentially explosive industry.

I am sure there are hundreds of Nascar.com , Major league baseball.com, HBO.com, disney.com out there that would stream custom HD video and audio to everyone they could for $4 per customer per month, if they had 128 core CPUs to do the rendering in real time.

Itunes is just the begining folks, MP3s are small and you listen to them many times.

The real push-pull media will be massive steams for single or few uses - and we all need tons O CPU to pull that off. And 64 bit processing and addressing (Stop putting OS calls and video memory at the 640k, and 1.7gb line MSFT - other operating systems don't use my program address space and ensure thier memory model becomes obsolete in a mere few years....)

People are dropping thier expensive land lines - paying "by the minute " for cell phones is actually cheaper.
Watch cable TV go down the same way - in a few years, you will pay by the byte to have "streaming colors of video goodness wash over you" - - and save money doing it.

People love thier stuff "net based". Google mail, salesforce.com, audible.com, itunes, netflix. Save my stuff "out there" - where I can always reach it. Soon that stuff will include CPU intensive things like my home movies, my scientific computing.

The web economy serves ala carte with the efficiency of mass production. Because there are hundreds of us that want to watch that strange indie flick about a red head running thru munich. And its only bandwidth and CPU cycle costs holding us back.

There are plenty of pressures pushing bandwidth (telco, video cable, web access, business data transfer).

The pressures for massive CPUs are just mounting - and the companies that are architecting for them will win the day in the next wave.

What will massive core (64 and up) CPUs drive? Storage demand.   Maybe someone should buy StorageTek. Too late.

Don't fear the CMMI monster

TallSails — Wed, 06 Sep 2006 09:22:05 GMT

Lots of shops have "gone process" and "Gone CMMI" (Capability Maturity Measurement Model), and many folks have gone berserk when it happens. Well it happened to us, and it was not the worst thing that could happen. It actually helped in a number of areas. Maybe we did it right. Maybe we did it light.

Disclaimer - I am not a process wonk. I am the fastest coder you ever met. Cause I don't sleep at night, I lie there paraniod about my deadlines and outline my code over and over for hours - so when I hit the IDE in the morning, its 2 to 3 hours of straight 60++ wpm typing of code into the IDE. (Ok, none of this is true anymore, its how I came up thru IT, and my wife will attest I sleep at least 3 hours every night.) Now I type memos and PPT for hours straight when I come into the office, if not giving live presentations....

Our Parent Company, (One of the largest corporations in the world) "suggested" we switch our evolved software and version management to CMMI. And it worked. Here is a short discourse on how we did it and what worked well, and (I think) why.

First - training - as CTO I attended a good bit of CMMI training and did a lot of research ("Distilled" and other titles). And I learned what this thing is, and where it helps. I produced a lot of samples for folks, and held in person PPT walkthrus of different aspects and processes. I vetted them against myself and other tech ringers. Then we used Camtasia to produce in-house video reminders of those sessions, showing mock up updates to those tracking documents, etc. (Train thrice...)

Then, we moved our current methods into it, not replacing them. Also, we used excel spreadhsheets for tracking. I liked that approach since its not as scary as a closed on line system, and since we were still very dynamic during the learning phases, we could change them often. I wanted things "compellingly convienient" for everyone.

We have a Wiki to place and update our policies, and links to our project documents. We went to a strict V Model BUFD (Big up front design.) No coding until all requirements are in and approved by the business side. Well you can be granted an exception with the proper paperwork, but that low cost hassle is enough to keep the process on track. You are allowed on all our policies to violate them, with minor paperwork and we track the violations. That was our overall philosophy.

We have been "switched" to level 2 and level 3 processes for CMMI now for over 6 months. Productivity, as best we can tell is up - the statistics from before don't compare well. Business leads are not complaining they are not getting thier applicaitons constructed fast enough, if thats a measure.

We have much higher transparency into what we are working on, and when it will finish. We have fewer people working OT.

Some of the best anecdotes are when we work on large projects for clients - we have a formal methodology in place and nothing special changes - its not new and different, we just use our normal approach to those projects and the customers are happy we are planning our work and tracking it. People like to know what they are paying for.

The complaining of all the process overhead dropped off very quickly to the usual level of corporate crankiness, once people learned how-to in the new methodology. My memory is about the 3 month mark, when some versions were wrapping up thier first shipping modules under the new methodology.

And the business people have learned how to drive in this new system - push to baseline - that magic point that we have all the requirements detailed and signed off, and construction (and testing) can begin. So they push to baseline, they actually tell us what they want, they actually cut features if they can't tell us, and they don't yell "type faster" to the coders as an approach to getting higher quality software out the door faster.

Quality is up on every formal and informal measure. We are spending far less time on a dot-zero (8.0, 9.0) shipment fixing things. The traditional "quick patch version after a dot-zero version" is a thing of the past, which yields a 30% savings right there, if my old project plans are any measure.

And one quote from my CMMI training I will never forget is already true. "Once you work at a level 2 or level 3 shop, you will never work at another shop that is not that level". We have our business people educated on where they can lever software, (feature selection, feature cuts, select the ship date) and where they cannot (don't try to drive estimates lower, the tech side owns those - thank you xtreme planning...)

It was a lot of work to switch over. And we are not done. Our next CMMI rating will be about a 2.5, some of our level 3 processes are not perfect, some of my teams are not even doing 2.5 consistently.

But its already worth it. And I did not think that 6 months would be the point I would be saying that.

Reminder of my success points

train senior IT managment - this is not "a staff problem"
evangelize / train senior business manament

impossible. So have your trained IT senior management at all early project meetings, which turn into training sessions

make the new processes "compellingly convenient" - no new online closed system that people "work around"
explain what data you are gathering, what its used for, and how its anonomized
admit process errors early and often, fix the descriptions and process on the wiki in seconds and broadcast the fix
Keep helping your IT staff
Go light on process punishment for quite a while, give people time to learn.
Tracking - what you don't track, people eventually stop doing and go back to old habits.

Track everything, conveniently.
take attendance at training.

We use electronic sign off (your initials and date). No one has tried to cook the system. But its real convenient
We use wiki for fast process documentation and adaption, and easy surf and search for newbies
train everyone in 4 modes

wiki process reading
live walkthrus
movie files for offline walkthrus
lots of 1on1 support anytime they need it

Never shoot the messenger - you will never get valid data again
Never harp on dates - harp on what features should be in our out to make that date
update plans as soon as we think an estimate is too low or too high
Estimates are soft until baseline

The transparency to senior managment and the ability to hit a target date are amazingly valuable to any company.
Its also like a detailed home-quicken (or MSFT money) budget - be careful what you ask for, the details can be scary.

If your software shop is large, or getting large, this thing scales - its the way to go.
If your shop does outside (custom work for clients) then this is really really the way to go.
(Even if only some of your work, like our shop, is that way).

Someone once told me (in referring to sales pitches, but it applies to IT processes as well)
"You may as well write the script down, cause by the time you do it over and over, you are following a script anyway, you just never wrote it down and refined it".

Thats CMMI in a nutshell.
Well, at least our version of it.

SAP realizes they are out in the cold

TallSails — Fri, 19 May 2006 07:35:00 GMT

SAP has all but asked to be taken over, by, as they list, Google, MSFT or IBM. Sounds like they are limiting their thoughts to companies with billions in cash on hand. Well they need to be taken over, they know it, and here is why.

They don't have a stack, their own platform. Right now there are arguably 4 stacks out there, MSFT, SUNs, Oracle, and LAMP. Oracle's has kinks - parts are excellent, parts you prefer tomcat and mixing stuff. Or you can go "best of breed" and pay consultants to pick BEOS and other things, and deal with the high costs of all those elements versioning or revving at differnt and incompatible times.

MSFT, SUN, LAMP are complete stacks, MSFT and SUNS rev in concert - a big advantage over the open source field. Thier whole stacks rev together.

So why is this so important?

Enterprise software, global platforms. When you sell a massive enterprise application, what is your operating system, app server, database, web server cost? (Or when you buy?). MSFT has entered the enterprise software business with dynamics. and they will get this massive application correct, eventually, its thier forte (massive applications). So they compete with SAP, or soon will. And they can leverage thier office and windows brand to gain entry into every IT shop in the world to sell it.

So when the 3 bigs (that magic rule of 3 major companies usually shake out in a commodity business - see sodas, airlines, etc) all offer enterprise enabling suites of applications - who has a distinct price advantage? The ones that have thier own stack and can bid 0 on the operating system, database, app server, web server. And MSFT can. And SAP can't - they use SQL server and MSFT technologies. They even tout sharepoint portal. (WOW)

I was surprised they didn't see MSFT coming, but now they do - they cannot survive another ten years, even as massive as they are, with this price disadvantage in every single deal. And lets not even add in the innovation advantage as one side of MSFT knows whats coming and when in the stack ahead of SAP, and the possibility of tricks to make sure MSFT stuff runs better. But the former reason is the biggest.

MSFT has a 20% price advantage on every deal. SAP has figured that out.

Which is why, the industry gorilla, knows they need to partner up with someone with a stack.

Or migrate to an open stack without lock in technologies - LAMP or SUN. But the time and cost to do that might be larger than just being bought.

Enterprise software is a very interesting game, with its own rules. Its wierd to see an industry leader scrambling for a date to the prom, but in this case, SAP has done thier homework and they are smart to get moving now before all the handsome princes are taken.

Watch for Oracle, who also sells enterprise software (people soft, sielbel, Oracle financials) to start courting someone with a stack as well. Or maybe they already have.

SUN does a reshuffle, what does it mean?

TallSails — Tue, 16 May 2006 14:11:29 GMT

Everyone has heard Schwartz is in as CEO, and I think thats great. Read my entry "Who you gonna bet on" for why I think its great, even though I wrote it before the move was announced. But a new announcement, has Fowler now running the entire x64 and sparc chip shops, and Yen now heading the storage (tek!) shop. What does this mean?

I have had the great fortune to spend a couple of days with both Fowler and Yen, doing presentations at a SUN analyst conference this past February. They are both very capable leaders, and very customer driven, while understanding the details of technology - a rarity. SUN is fortunate to have a beevy of great leadership.

Over the past years, with the guidance of Andy Bechtolstien, the SPARC and X64 product lines have found a number of cost and risk savings common design and components - from chassis components, to circuit designs, to process and off shore (er, global) teams.

And Yen and Fowler each headed the Sparc and x64 groups (respectively) that merged a lot of these costs, risks, and designs. So now, I expect more of the same, the storage group will benefit from a leader that is plugged into the chip-server group leader, and they can combine components and design elements, wringing further cost structure benefits (big dollar words for "savings") and risk reductions from the storage line thru the server line at SUN.

After all, a large component of storage systems these days is the embedded servers (controllers, etc) that control them. The Storage tek folks now have Yen heading them up, and he knows how to lead a large group of engineers and hit dates, and how to navigate the SUN eco-system, he is an insider. They will be able to leverage that to stay fleet, and not get bogged down in merger corporation hi-jinx.

I think its great. You can't run the company as two camps.

Google hooks up w SalesForce and Oracle

TallSails — Wed, 19 Apr 2006 09:00:16 GMT

Google Inc. announced new partnerships with business software suppliers Oracle and Salesforce.com as it makes a big push into selling enterprise search features.

This goes square up against SAP, and Microsoft Dynamics. Its going to get interesting - enterprise application suites - on the web or otherwise, are an entirely different market than monetizing eyeballs on content (ad revenue). While I am impressed with Google on many fronts, this one is definitely going to wake the Redmond giant from its sleep. Massive applications is a space that MSFT eventually gets right, and I expect they will be a very formidable force in this area.

The question is, how do you purchase your enterprise infrastructure?
Traditional C level initiative - SAP wins
RollOver momentum from OS and office - MSFT
Organic (sometimes exponential) migration web service at a time - Google-Oracle-SalesForce?

And how will those decisions be made in the future?

Why did ACT! and Goldmine let SalesForce Happen?
Cause corporations are slow to change direction. It will take MSFT a long time to turn in this direction.

But Woe to SAP when they do. It appears thats the direction (Dynamics) they are headed.

Leaves space for the services approach from Google....

Who are you gonna bet on?

TallSails — Wed, 19 Apr 2006 08:23:04 GMT

As a CTO, and as every developer is thier own CTO, we make bets on what technologies are going to make it. Its terrible to toil for months and find out no one wants your great new app, built on technology "that is soooo last year". And in part, you are betting on the personalities of Ballmer, Ellison, and McNealy when you make some of those bets. I have had the pleasure of meeting some of the industry leaders.....and I bet on McNealy - and here is why

Full disclosure - I have met McNealy roughly 10 times in settings of 20 or less (a few time 4 or less) people, and a couple of times we have dialogued 1on1 - it was pretty nerve wracking the first time I realized he knew me by name. I have met Ballmer three times in person, groups of 50 ish. I have never met Ellison in person. I have viewed interviews of all of them, perhaps you have too.

So why do I prefer McNealy? Is he a nutcase? Isn't everyone at that level of earnings and success a bit different than the rest of us? I hear the critics - and he shares some of thier jokes about his living in his own "reality distortion vortex".    I think Scott M could be a LOT less acerbic when talking about the competition.

So why do I like him? First, he is transparent. He talks about where the industry is, how his company fits in that, where they are headed, and why. Ballmer talks about things that may never happen. (Mostly he talks about his company, not the industry forces and dynamics) Ballmer explodes - - and rants, but does not inspire me with any insights as to where the industry is headed.    Ellison is a modern day Sun Tzu, but he is in a land war with china when he goes up against MSFT.

Frankly, given how pervasive Java is - from cell phones to applications - we should all be glad SUN does not practice predatory tying and lock in approaches to thier technologies - sure they try to sugar coat you into thier other technologies - but those are not lock in either.

I don't know many developers that think Ballmer is dialed in on technology. Most of them think of the monkey-boy videos and send them to each other. McNealy makes off color remarks, but its part of his transparency, and frankly entertaining - when you sit thru an 8 hour analyst briefing, the occaisional frank remark helps keep you awake. But Scott does have a way of simply describing what is going on, and what will go on, in this industry. So do his staff, from Greg Papadopoulos <Greg Matter blog> and Schwartz <JSchwartz blog>on thru. At times I think they are orders of magnitude optimistic on how-fast all this may occur - and at times they prove me wrong.

In fact, props to SUN the company - to a person SUN has brilliant, upbeat people working for them. Yes, I am impressed.   Its a rare combination to have in a staff - admittedly I have it in mine as well - which could bias me into a self affirming judgement.

Ellison goes for the dark and mysterious approach, and while I love sailing, his pricing model is so complicated you can't get the same answer from any two Oracle reps as to what you should pay. I like his product, rather parts of it - - but lets get a pricing model that fits the current market and is comprehendable.

I also give props to Scott for making Schwartz (the anti-Scott) the president of SUN - in many ways his compliment, which helps. I as well am lucky to have complimentary skills near me in my org - and it really helps. Schwartz really knows that developers drive this industry - and not HIS developers. Oh and they have done the math on how many developers, servers, applications are coming from India, China, and Europe - - and those folks do not like spending hundreds on operating systems, office suites, or IDEs - so MSFT - figure out how to live without that revenue.

(Dynamics is a great idea - MSFT will kill SAP - you read it here first - its thier strong point - massive applications - it will take them 10 years to do it. But I would not want to be SAP or peoplesoft for the next ten years). Google wants the ad revenue and seems to have mastered it, that can drive those free applications (and thousands of others).

Sun doesn't care who strikes gold, they know that most of the millions made during the gold rush were made by folks selling shovels and blue jeans...

I don't really think Scott McNealy is a nutjob - but many of you do.
So as CTO when you pick your STACK of technologies, you are largely picking "which nutjob" you believe.
Given 3 nutjobs, pick the one with the industry insight.

Oh, and if you are a CTO or a developer, and not a pundit - you have to pick one, you can't sit back and say they are all whack. (well, you shouldn't. Some developers still build thier own stack, but customers are getting wise to homegrown--proprietary-- no economies of scale costs of maintaining those systems..)

Virtualization is the COOLEST thing

TallSails — Tue, 25 Jul 2006 15:02:28 GMT

What do you do with 18 servers, containing 22 power supplies, 26 processors, 31 gb of memory, and 51 disk drives, consuming over 14,000 watts of power?

Replace it all with one SUN 4200. Thats what my company did and we couldn't be happier. Using VMware and virtual machines running all the same applications, Operating systems, licensed third party stuff we did before.

Oh but the new 4200 has integrated lights out management, so remote support is much easier. And we can fall over the virtual machines to other 4100s if this one fails. Our data center is 17 degrees cooler and our UPS uptime is now extended over 30% - we avoided buying a new airconditiong system and new UPS - saving over $180,000.

The money, heat, and power savings are allowing us to add more servers, nice cool 4100 and 4200s, to add compute power. The end of the blow-dryers we installed from the Gigahertz-wars of the late 90s and early naughts (20-00s).

We used Platespin to automate creating the virtual machines - it went like clockwork. The virtual machine images themselves are stored on a Lefthand Networks iSCSI SAN cluster, which is also used for file storage by the virtual machines themselves. The funny thing is, much of the data didn't move, it was already on the SAN.

All our back office functions are now on virtual machines - not taking up power, creating heat when they are not in use. Performance is higher than before. Some of the servers replaced were dual core Xeons- some admittedly older PIIIs.

My developers, who chew up servers like skittles, love centralized workstations in VMs on the servers too - they can work from home, from a park, they can collaborate several folks for a few minutes on the same VM (they don't like working together for more than a few minutes....). Our multi-city development is doing great. Centralization is now hip, even with the "I want big iron on my desk" crowd. The large number of servers it takes to support older versions of our applications in the field - virtualized. Its eco-friendly and they spin up in seconds - much faster than a cold boot.

The hardest part of the whole project? Finding out what to do with the other 50% of the SUN 4100 - its running well below 50% capacity.

Congrats to SUN and AMD - you have really changed things. I can only imagine with the cost of gas and oil shooting up, and the cost of power in developing countries like india and china (and yes, thats where the majority of servers are going to be installed in the next 20 years, IMHO...) that the compute-power-per-watt of these new machines is going to really change things. Its amazing how much cooler these servers run.

Don't take my word for it - try one for free - free server from SUN (with free Oracle. Double WOW)
I was never a big fan of centralization, coming up as a developer.
But my budget forced me to try it. And there is no doubt its the way to go.

oh , and a quick calculation on that 14,000 watts of power. Thats 14kw, x24 hours yields 336 Kw per day, times 365 days a year, roughly 122k (kw) or 122Mw, and at a retail price of roughly $40 (updated) a Mwhour - $4,800 a year in power savings alone. Actual power savings are about 3x that since you need to power the UPS which is a lossy and heat creating device, and also the air conditioner (we just calculated the power used by the power supplies here - removing that heat takes more than 1x the energy to create it, typically 3-4x...).

So a measly SUN 4200 is saving us roughly $12k a year in power, half that if you have a cheap rate on your power next year - - but the blackouts in houston this week portel and expensive, hot summer, not cheap power....

For technical details, check my IT wizards blog entry

The Google Curves

TallSails — Mon, 10 Apr 2006 15:36:53 GMT

There are two curves that drive everything at Google.

For the first curve, plot two (added) factors - advertising rates (super bowl ads, TV prime time ads, even click thru ads have not dropped in 5 years - thats right, internet eyeballs are worth MORE per now than years past...) TIMES the number of users on the internet, which is rising by something on the order of 4million people a month. And has plenty of headroom to go, something like 3/4 of the planet has never (much less than daily) been online.
This first curve is the sum of two growing factors - both growing at above linear organic rates.

For the second curve, plot the two factors of - the cost of computing and the cost of bandwidth. Bandwidth is amazingly cheap, I am paying $30 a month for DSL, which is a lot more bandwidth that the 1200BAUD i used to pay the same amount for.

Now the fun part. Overlay the curves. Anyway you want. Even if you use different timescales, not sure when all those people get on the internet, etc, not sure the cost of writing internet software applications -

it becomes very clear, once you look at those diverging curves, that you can make a boatload of money eventually, and still make a lot of mistakes on the way.

Microsoft is trying a whole different, how many can we lock in, approach.

Get on the curves and you win. Volume wins. Check out Jonathan Schwartz BLOG on volume wins. (or the value in volume)

Thats what google is up to. Thats why they can provide lots of stuff free - computing and bandwidth dropping live crazy in price, revenue shooting up like crazy, and the cost to develop software scales across billions of users very easily with thier infrastructure. You might say thier amortized cost to develop software per user is cheaper the microsofts - becuase it is, by a large factor.

Now read the rumors about what google is doing, and it all makes sense....

Apple has another wave to go - Audio Books

TallSails — Wed, 19 Apr 2006 14:09:04 GMT

I was lucky enough to have a hip enough wife that she gave me a Nano(tm) for our anniversary this past year. I love many kinds of music and was glad to have it - morning radio has gone the way of jerry springer, and I travel with my job a lot - listening to music during these extended commutes is a great way to pass the time. Makes me hate the "wait for the plane now that I have cleared security" less.

And it changed my life. But shockingly - not in a listening to music way. I use my Nano to listen to audio books far more than I use it to listen to music. Drive time, extended commutes (travelling for work), last hour of the evening (I don't watch TV - about 2 hours a week tops), yard work. I find I can't listen to audio books when doing cognitive tasks (even blogging) but I am addicted.

I know, you are thinking, aren't those boring edited version made for tards and the blind? Nope - life has changed, as as we no longer go to the town square to listen to an orator, reading a dead-tree version of a book is passing quickly - and the audio books I download are un-abriged.

This is an industry changing phenom - I am averaging two books, at least 10 hours a week of "reading / listening" to them. I have spent more money on audio books than on electronics and music combined in the past 4 months. I have rediscovered my No-time-for-it love of history, and am listening to stuff I hated studying during prep school.

I also listen to the java posse podcast, www.javaposse.com (a great summary of java news and views)
and diggnation a summary / comedy-commentary on the social news website, digg , to keep up with technology as my passions and my job require it. I get my books from audible, since they have the best price points I know of, and I just can't bit torrent books - there are very few rich authors (rich in the "Does Harrison Ford really need more money" kind of rich) out there, especially the ones that write history books....

Oh, and my five year old daughter likes to "listen in" on one of the ear buds as I tuck her in, she is now full of questions about the civil war - which while tough to answer, are surely easier than other questions she will ask later in life. Its a funny thing trying to explain our society, past or present, to a five year old. Try it sometime.

Next week - the crazy list of books I have listened to.

Intel vs AMD chip designs

TallSails — Wed, 08 Mar 2006 15:10:36 GMT

http://news.com.com/2061-10791_3-6047412.html.

Very interesting, intel makes a higher profit by keeping the memory controller off its precision fab CPU, gets more CPUs per die than AMD. The opteron NUMA (Non uniform memory architecture) is tricky, but even its slowest 3 hop memory fetch is faster than the normal 1 hop memory fetch on the intel chipset.

The tradeoff is smaller L2 cache, some of the transistor budget for that is used for the integrated memory controller on the opterons. Oh, and intel makes more profit. Thats a tradeoff too, on another budget. Yours to thiers.

Who wins? Right now opteron. When memory sizes (RAM) shoot to 2-4 Terabytes for a server? The ratio of cache to RAM will continue to expand, and the tradeoff intel is making could very well not show well as RAM densisty and sizes take off again. Your hit ratio goes down as apps use more memory, unless you can keep expanding cache.

It also depends on the app - small apps - cache. Massive RAM simulations, like I have spent (is it twenty?) years of my life writing and architecting - integrated memory controller. And the benchmarks prove the theory on this one.

Whats funny to me, is this is clearly a piece of Intel PR, and yet, even with thier hand on the pen, makes me prefer the AMD CPUS. But my benchmarks had me biased already. So did my air conditioning bill, and the fact that intel rigged thier compilers to produce "not as optimal" code when producing code for AMD chips. (see slashdot for details).

Google buying SUN

TallSails — Wed, 08 Mar 2006 11:14:02 GMT

First, pundits like to post, or talk about (Fox news. We report. You Cringe) explosive things. Google buying SUN is one of them. Its based on Mcneely might resign (don't we all one day?), and he is selling shares.
www.digg.com has it as a highly dugg (voted for) article yesterday. Its exciting, but not realistic at all.

I am pretty plugged in, via personal contacts and work, to both companies. It won't happen. And the best reason has not been posted yet, Google uses a ton of java and needs it to stay alive. Lucky for google, or Sun realized it, java is open source, so google can in-house it if they need to. As can the rest of us. So they won't buy Sun for that.

Another good reason would be to keep secrecy high, and secure thier supply chain of opteron motherboards. But they can do that and not aqquire sun, and actually keep sun more viable as a customer.

Once they buy Sun, "sun the hardware vendor" is now trying to sell hardware to lots of google competitors. Not going to happen.

Google has realized that the cost of computing is going down (and bandwidth) while the revenue per advert (superbowl ad or mouse click thru) has not. They are exploiting those two curves.

Recall the old days when "Mutual of omaha" would air (as would lots of other shows) - NBC did not produce it, they ran it for the priviledge of running 3 of the adverts during each break. The producers (Mutual of omaha) ran the 1 advert and paid for the show.

Thats where google is headed. They will serve up the content for a portion of the ad revenue. Right now its a large partion, but that will change.

Content is free if they can monetize your eyeballs looking at it. Which is why the RIAA ruckus will die down as the music industry figures out how to show me ads as I download MP3s and organize them. Which is why cable and satellite TV service will be free, and broadband keeps getting cheaper with scale.

Here's to google, they get it. And since they use java, they don't need a specific OS or hardware vendor to stay alive to keep thier money printing machine running. Same for my company. We have applications that have been selling for over 30 years, and we are tired of re-writing them everytime MSFT uses rolling obsolesnce to crank revenue. So we are cutting the depencency cord. Have been for years.

I don't mind companies monetizing things. I do mind them monetizing, and leveraging, sheer dependency - thats not value add. MSFT had better move fast.