November 7th, 2009

The real pros and cons of server virtualization

Posted by Paul Murphy @ 12:15 am

Categories: Enterprise Policy, General, Linux, Productivity, Software Licensing, Strategy, Sun

Tags: Data-processing, Server, VMware Inc., Server Virtualization, IBM Corp., VM, System Utilization, Virtualization, Hardware, Paul Murphy

First, lets be clear: this comment is about server virtualization through ghosting - the business of using one OS to run one or more ghost OSes in lieu of applications each of which in turn is able to run one or more applications - it’s not about desktops, not about N1 type technologies, and not about containerization.

The pre-eminent examples of ghosting OSes are IBM’s zVM - an OS that originated in the late 1960s as one answer to the memory management and application isolation problems confronting the industry at the time- and VMware’s more recent rendition of the same ideas for x86.

Back then, IBM was caught between rocks and hard places: lots of people (including IBM’s own research leaders) were developing system resident interactive OSes aimed at using the computer largely as a central information switch, but its commercial customer base absolutely refused to countenance any advance on the batch tabulation and reporting model around which its management ideas had evolved in the 1920s and 30s.

Thus when the Multics design effort started at MIT in 1959/60, most of IBM’s people didn’t even know there were two sides to the argument but the research people lined up with science based computing while those who made the money for IBM almost unanimously choose the data processing side - and ten years later, after MIT’s people had first won their design battles and then lost the war (by letting data processing get control of the Multics development effort), IBM’s own fence sitting solution: VM, ended up roundly hated by nearly everyone.

Nearly everyone, that is, except people limited to IBM 360 class hardware who had no other means of achieving any kind of interactive use (this was before MTS and a dozen later solutions) - and they, essentially over the objections of IBM’s own management, made VM the success it still is.

All of which brings us to the 90s when available x86 hardware mostly wouldn’t run NT 3.51 and Microsoft’s emergency iVMS port, aka 4.0, contained a misconstrued uaf derivative known as registry that effectively limited it to loading one application at a time - thus forcing buyers to choose between a lot of downtime or rackmounts of dedicated little boxes.

The rackmounts won - at least for a few years; but then data processing got took control of the wintel world and VM, in the VMware incarnation of its ideas, soon became the preferred tool for reducing the rackmount count in the name of their professional holy grail: higher system utilization.

Unfortunately there are two big problems with this:

1. first, NT 4’s limitations went away with NT 4 - addressing them today with VMs achieves a level of absurdity no audience would accept in musical comedy -it’s right up there with using a licensed terminal emulation on a licensed PC to access a licensed server running a licensed PC emulation; and,

2. it is very nearly a universal truth that every gain data processing makes in improving system utilization produces a larger loss in IT productivity for the business paying them to do it.

The reductio ad absurdum example of the latter is Linux running under VM on a zSeries machine: data processing can get very close to 100% system utilization with this approach, but the cost per unit of application work done will be on the order of twenty times what it would be running the same application directly on Lintel; and every variable in the user value equation: from response time to the freedom to innovate, gains a negative exponent.

You can see the latter consequence in virtually every result on benchmarks featuring some kind of interaction processing. For example, the Sun/Oracle people behind their recent recent foray into TPC/C, both demonstrated their own utter incompetence as IT professionals by achieving less than 50% CPU utilization and the user value of this “failure” by turning in response times averaging roughly one seventeenth of IBM’s:

The counter argument I usually hear about all this is that virtual system images are more easily managed than real ones - and this is both perfectly true and utterly specious.

It’s perfectly true that VM style virtualization lets you bundle an application with everything it needs to run except hardware, and then move that bundle between machines at the click of an icon; but the simple fact that this applies just as well to Solaris containers as it does to VM ghosts shows that this is an argument for encapsulation and application isolation, not for ghosting.

Worse, the argument is completely specious because it bases its value claims on two demonstrably false beliefs: first that the only alternative is the traditional isolated machine structure, and second that virtualization lets the business achieve more for less. Both are utter nonsense: Unix process management has worked better than VM since the 1970s, and because virtualization adds both overheads and licensing it always costs more to do less than modern alternatives like containerization or simply letting the Unix process management technology do its job.

Again the quintessential example of this is from the heart of the data processing profession: when you take a $20 million dollar zSeries installation and achieve a 60 way split to produce 100% system utilization from 60 logical machines running applications or ghosts, what the business gets out of it is roughly equivalent to what it would get from four Lintel racks costing a cumulative $500,000.

A more down home illustration is provided by VMware itself - their competitive value calculator computes a cost advantage for their products over those from others on the basis of their belief that their VMs impose less overhead and allow you to get closer to 100% hardware utilization. Thus if you enter values saying you’ve got 200 applications running on NAS connected quad core servers, want to manage virtually, and have average infrastructure costs, they produce a table with this data:

All of which should raise a couple of questions in your mind:

1. first, if the consensus that ghosting doesn’t have significant overhead is right, where is VMware getting the third of the box it claims you can recover by getting its ghosting software instead of Microsoft’s?

2. and, second, wouldn’t the money VMware wants you to spend on ghosting ($241K in this example) be better spent on hiring people who can move these applications to free environments like Linux or OpenSolaris?

So what’s the bottom line? Simple: the real ghost in ghosting is that of 1920s data processing - and the right way to see this particular con job for the professional cost sink it is, is to focus on costs to the business, not ideological comfort in IT.

October 24th, 2009

What Windows7 could mean for Linux

Posted by Paul Murphy @ 12:15 am

Categories: Apple, Applications, Development, Linux

Tags: Marketing, OpenBSD, Microsoft Corp., Linux, Operating Systems, UNIX, Open Source, Software, Paul Murphy

I’ve had people using Windows 7 for about three months now, and everything about it so far seems to confirm my first impression that it’s a lot better than Vista: effectively reprising the consolidation and debugging Windows 98 offered over 95.

Once you get past the sheer shock of using a Microsoft OS that doesn’t fail daily, however, you start to fret about the things that aren’t there: as a Mac/Solaris user, for example, I find the absence of multi-screen capabilities and the relative inflexibility of working panes and icons extremely frustrating. Still it is usable; and that’s a long step forward - at least until you get to development work.

Then the frustrations set in: Visual Studio is very slick, but very limited. Specifically, it’s great if your application is going to use a super-computer desktop as a graphics terminal but pretty much counter-productive if you want to sidestep client-server and produce genuinely integrated multi-host applications.

So why? Well, mainly because Microsoft’s inability to transcend its own 90s focus on helping its sales force make money selling client-server into businesses has left the whole .net thing Microsoft promised to integrate into Longhorn and its successors implemented, along with the promised PICK-like file system and security conscious display frameworks, only in marketing documentation.

Organizational disfunction aside, I think the key technical reason for this has been that getting those things done within the underlying memory and process management paradigm Windows NT+ inherited from VMS has proven, if not actually impossible, at least too hard for Microsoft to make a commercial success of.

So now it wants to sell cloud computing and applications rentals but doesn’t have the OS foundation on which the development of these products has to rest - and that’s going to force Microsoft into a build or buy decision.

They’ve been trying to build a network based, vaguely Unix like, OS for PowerPC for about six years now -with no success to speak of, so my guess is that the build exponents will eventually lose the argument - leaving Microsoft with three mutually exclusive choices:

1. get there through a licensing deal with Apple;

2. do it by adopting and extending OpenBSD; or,
3. do it by adopting and extending Linux.

Each approach has pluses and minuses: the Apple approach would cost the most upfront, but drop a leading competitor out of Microsoft’s desktop markets; the OpenBSD approach combines low cost with a high quality code base and a well deserved reputation for security; and the Linux approach capitalizes on the breadth and capabilities of its community while threatening IBM.

You’d think Microsoft could do the Apple deal at the drop of a phone call to Mr. Jobs - who clearly wants to be out of the traditional PC business anyway - but my guess is that the emotional barriers to rational behavior on this will prevent that phone call.

If it comes to shootout between the OpenBSD and Linux options I suspect Microsoft’s techies will line up favoring OpenBSD as offering the stronger foundation for all the neat stuff they dream of doing, while all the marketing types will favor Linux - and in that company marketing trumps technology every time.

So the bottom line for Linux on Windows may be simple: Windows7 is probably Microsoft’s best OS yet and will therefore slow the move the Linux in the short term, but the limitations built into Microsoft’s development stack show it to be a dead end that will leave Microsoft marketing magnificent visions of its unfolding future while quietly figuring out how and when to abandon that code base for something else - and because that something could very logically be Linux it might be time for the Linux community to start paying a lot more attention to legacy interoperability with Windows.

September 26th, 2009

Linux as Wintel parasite

Posted by Paul Murphy @ 12:15 am

Categories: Apple, Development, Enterprise Policy, General, Linux, Linux Desktop OS, Sun

Tags: Wintel, Register, Linux, UNIX, Open Source, Operating Systems, Software, Paul Murphy

Last week Linus Torvalds told a seminar group at Portland’s LinuxCon that Linux is getting a little bloated - a consequence of the big blob kernel architecture required by his decision to prefer the efficiency of directly using x86 interrupts to the much more hardware independent architecture Tannenbaum developed Minix to teach.

Sun blogger Joerg Moellenkamp said something particularly interesting about this:

Of course it’s a nice sign of success, when people port more and more stuff to an operating environment and into the kernel. Perhaps this is the price of success. But at foremost it’s a problem. Bloat isn’t just about using more memory, it’s about speed as well.

The Register delivers another interesting piece of information:

Citing an internal Intel study that tracked kernel releases, Bottomley said Linux performance had dropped about two per centage points at every release, for a cumulative drop of about 12 per cent over the last ten releases.

…

Should they rearchitect Linux for the future (the SunOS/Solaris moment for the Linux community). And as refactoring, optimization and rearchitecting are tedious and boring tasks: Who will do it? I think, the next few years will be interesting ones for Linux.

Another speaker at the same event, IBM’s Bob Sutter, really needs to spend a few minutes looking at the history of his own company’s VM product line, but other than that came up with another absolute shocker: Linux won’t succeed on the desktop, he said, unless it creates a unique Linux desktop - or, in translation, that Linux can’t lead by following.

Personally I think that the SuSe business desktop does lead Microsoft in some areas, but, of course, Mr. Sutter wants to sell cloud computing - and so does Eric Mandel, CEO of a company called Blackmesh, providing managed Linux hosting services. He does a very sad and funny presentation on doing what they did: implementing a couple of open source deployment tools (Puppet and Cobbler) to make it fairly easy to configure and deploy Linux server/application combinations. This can be very important in their business, but I thought the retrograde nature of both the solution and its markets unhappily captured the essence of Linux today.

Basically he’s using an open source evolution of the old Jumpstart stuff to provision gear for customers who haven’t figured out yet that letting other people control both their data and their most critical business infrastructure is a recipe for coming to a quick and unhappy end. Cool stuff, for five years ago - but completely obsoleted for customers by today’s cost/risk trade-offs in doing it themselves and for techies by Solaris zones.

When you look at this kind of thing the contrast with BSD could hardly be greater. That group’s focus, despite their many divergences and disagreements, is always on better, faster, smaller - and Apple’s posture as the anti-IBM in personal computing carries over to its relationship with the BSD community: it’s the world’s biggest producer of Unix personal computers, but it doesn’t try to direct BSD research and it hasn’t tried to build services revenues on its own limitations.

Mr. Torvalds set out to build a “free Unix for the 386″ and succeeded brilliantly in doing so - but both its internal architecture and its market success depend on the peculiar dynamics of the wintel market in which x86 forms the common ground between the huge majority using Microsoft software and a rebel group looking for something to call its own.

Thus looking at it as an outsider, I’d say that much of what made headlines at Linuxcon 2009 was in one way or the other about the chickens associated with the reinvention of old technologies for commercial gain starting homeward -with all of it demonstrating that if the Linux community didn’t have Microsoft both to be against and to prop up their shared x86 foundations, it’d wouldn’t exist.

And that’s sad - but not irretrievable because at this point it’s fundamentally a leadership failure, not a community failure, and therefore something that could be changed.

August 22nd, 2009

Save money: buy from the enemy

Posted by Paul Murphy @ 12:15 am

Categories: Apple, Enterprise Policy, General, Hardware, Linux, Sun

Tags: Car, Dell Computer Corp., Sun Microsystems Inc., Apple Inc., Sun Sparc, Sun Solaris, Processors, Servers, Operating Systems, Software

When Spain opened its passenger vehicle markets to competition on entering the EU, the average new car price fell by about 20%, while the Obama cash for clunkers program, by reducing the availability of used cars, is expected to add about 3% to the value of the average used car still on the market - and significantly more for SUVs.

Those are obvious macro effects of competition. In IT there are similar effects: it’s competition that drives the innovation that drives the cost part of what we see as the operation of Moore’s law: Intel didn’t choose to produce the “nehalem” technologies because it wanted to sell faster processors for less, it got forced into doing this because AMD was eating away at its mindshare within the x86 community and both PPC and SPARC offered more for less beyond it.

As a result of competitive pressure a February 2009 TPC C result posted by Dell shows a per transaction (i.e. per “bang”) cost of just $0.60 compared to the $1.53 the same transaction cost according to a Dell report dated about five years (12/10/04) earlier - a gain of 2.55 times in bang for the buck.

It works the other way too. When Apple dropped the PPC for use in its traditional computer products processor competition declined: in 2002 Apple’s laptops were faster and cheaper than their (comparably equipped) PC counterparts, but today’s Apple laptop is physically indistinguishable from the PC equivalent - and costs more.

IBM has a closed mainframe market and its true believers are almost completely impervious to competitor appeal: thus a 3096 cost $4.5 million in 1982 while today’s version runs about $12.5 million -a larger jump than you’d expect from the consumer price index which grew by only 2.3 times over the period.

I get a fair amount of mail from nitwits rejoicing in what they see as they death of Sun - and particularly of the SPARC/Solaris combination. I don’t know what drives people to hate like that, but here’s a free bit of advice to everybody else: if SPARC/Solaris dies, you can expect to pay far more for your future x86 products than you will if Sun/Oracle gets it together and pushes these technologies as fast and as far as they will go.

Take away competitive pressure from the SPARC/Solaris combination and the rate of price/performance improvement for both wintel and lintel will decrease - imagine that it falls by only 20% (wildly optimistic, in my opinion) and your 2014 cost per “bang” will fall to only $0.29 instead of the $0.23 it’s headed for now. Six cents per “bang” - doesn’t sound like much, does it? The Dell that delivered the $0.60 I mentioned earlier produced 104,492 “bangs” - and that six cents works out to $6,270 for the system.

For you and me that difference won’t amount to much - perhaps a few thousand bucks we could otherwise spend on something else. For big players, however, this can add up to real money real fast - so much, and so fast that people making decisions for thousands of servers every year should be thinking that maybe the best way to get price breaks from the monopoly of their choice over the next few years might be to call their Sun/Oracle representative right now and talk about placing some serious orders.

June 13th, 2009

Murphy collezioni

Posted by Paul Murphy @ 12:15 am

Categories: Applications, Database Management, Enterprise Policy, General, Government, Linux

Tags: Airplane, Calgary, Corporate Governance, Linux, Networking, Business Operations, Corporate Law, Operating Systems, Software, Paul Murphy

Huh?

I spent last Saturday at the Wildrose Alliance (think Republican) annual meeting in Calgary. More than 300 people in one room almost all of whom agree with me on most issues. Wow! - I mean, seriously, how often do you think that happens?

Anyway, at one point (apparently apropos of an egg salad sandwich) a guy I’ve never seen before tells me he “digs Linux”. “Digs”? Say again? Anyway, I think there’s a bottom line lesson here: this world is getting way too small and too interconnected.

Interesting weather too: 30cm of snow in the Cypress hills, heavy snow and hail in Calgary -obviously global warming, as I think Ed Morrisey said somewhere, now really is something you can shovel.

The 21st Czar

This week was marked by a need to spend a little bit of time reviewing the mandates various national network control authorities have. In both Canada and most of Europe undisclosed police access to just about everything networkish is taken for granted, but the formal centralization of political power to hunt for and shutdown offending network operators or users is something few countries have undertaken - in fact, here’s a list of of the ones I found with internet security czars with that kind of mandate:

Communist China
Vietnam
Iran
North Korea
Cuba
Venezuela
The United States of America

Bringing down airplanes

Absent “black box” information - and one of the many things I don’t understand is why those are kept on board airplanes instead of having live telemetry recorded at an airline operated central site - the recent Air France crash now seems likely to go down to the unhappy coincidence of one too many natural factors.

On the other hand, I wonder whether airplanes will soon need much better internal electronic shielding to prevent the insane from bringing them down by frying the (non optical) pump controllers all fly by wire aircraft depend on.

Notice that I’m not suggesting this happened, only that the technology both in airplanes and at Radio Shack has evolved to make this possible.

Model based management

A couple of phrases from junk I was scanning earlier - things that had people “executing well against our model” in a context of “intensified mandates” and the desperate need to “implement the most agile BI strategies” - reminded me of a comment I’d seen, by “Jeremy”, on watts up with that:

The photo is priceless. This to me summarizes the complete lack of scientific approach to everything these days. Everyone sits round a PC and blindly believes whatever nonsense it spews out.

No first principles. No cause and effect. No understanding of physics. mathematics or even statistics. Just run any number of widely available computer modeling programs, fit the historical data and hey presto another science breakthrough.

It would be funny if it wasn’t so very pathetically sad.

(FYI: the article spoofed climate modeling by reproducing a press release from 2006 - proudly celebrating predictions that proved entirely wrong - but how many times have you seen otherwise smart people impressed with some “corporate cockpit” BI display where, in reality, the second letter should be an “S”?)

Fun with the drupal

I’m not sure I want to say that I, uh, “dig” Solaris and open source, but I did have an interesting demonstration of the power of the two come out of Saturday’s meetings. This particular agency has spent well into seven figures on a custom consumer information site that another agency now runs for it on a mainframe IFL. Response, of course, is pretty poor but the big issues she complained about are things that reflect the differences in thought processes and priorities between editors and data processing managers - and the thing that struck me during the conversation was that drupal will do most of what she wants out of the box - and the CCK (content construction kit) should make implementing the rest pretty easy.

So I did a very basic drupal demo (not, please understand, that I understand Drupal - in fact, it’s often so backwards to my thinking that it takes a very frustrating time to see the retroactively obvious) - downloaded and installed a clean 6.12, gave it a private database and apache virtual server address, added the modules she needed, added some time and date information others think she needs, copied and restructured some sample content from her site, and turned her loose - all in just under three hours and fully justified in terms of amusement value when she reported that her home access to the demo running on my ancient Sun 150 via Shaw cable was significantly faster than her access to the z9 IFL at the office.

Out of which an abberant thought: there are lots of editorial professionals out there facing daily frustration because the people who built their website tools simply don’t think like them (and, admittedly the notion that editors think is a bit overboard, but go with me on this, ok?) - and could therefore improve their own lives, along with system reliability and flexibility, simply by replacing all that eons evolved home grown stuff in favor of open source products like Drupal and Apache on Solaris or Linux.

And one more thing

From newsmax:

La. House Backs 15-Cent Charge on Internet Access

BATON ROUGE, La. — A 15-cent monthly surcharge should be levied on Internet access across Louisiana to fight online criminal activity, the House voted 81-9 Thursday, over the opposition of Gov. Bobby Jindal.

Rep. Mack “Bodi” White, R-Denham Springs, said he sponsored the bill for Attorney General Buddy Caldwell, to raise money to finance a division in Caldwell’s office that investigates Internet crimes, particularly online sex crimes against children.

The measure would raise $2.4 million a year for Caldwell’s department, according to a financial analysis.

“I don’t think that 15 cents per month is too much to ask for our children’s protection,” said Rep. Simone Champagne, D-Jeanerette.

So, quick question: if you just use your iphone -or cableco VoIP phone- for phone calls, do you pay the tax?

May 2nd, 2009

Oracle Office, MySQL, and other dreams

Posted by Paul Murphy @ 12:15 am

Categories: Applications, Database Management, Enterprise Policy, General, Linux, Software Licensing, Strategy, Sun

Tags: Oracle Corp., MySQL, Sun Microsystems Inc., OpenDocument Format, OpenDocument Format (ODF), StarOffice, Open Source, OpenOffice, Databases, Emerging Technologies

Most people don’t know this, but at one time Oracle Corporation ran an extensive internal beta adding a pretty good word processor and spreadsheet to the Oracle Office communications product most recently renamed Oracle Beehive and now positioned against both Exchange and Domino.

At the time I didn’t see the combination as competitive with Applixware or even Q-Office, but recommended it for sale because it did one thing better than any competitive product I knew of: it stored everything as rows in the standard database.

Sun does not own OpenOffice.org and Oracle, accordingly, isn’t buying it - but because Sun does own the StarOffice product and is the primary contributor to OpenOffice.org, Oracle is buying both influence with, and responsibility to, that community.

If Sun and Oracle now recreate the full Oracle Office by combining Beehive with StarOffice, ensuring standardized database storage for all data, and open sourcing some of Oracle’s forms based development tools into the bargain, almost everyone stands to gain.

The obvious effect, of course, is to give OpenOffice a tremendous technical advantage over Microsoft Office while, at the same time, adding communications technologies and a real Access competitor.

I think, however, that two rather more subtle effects would dominate the future impact such a product would have.

First there’s the impact on the whole ODF scene. Right now, and for the past ten years, I’d argue that the eminently logical separation of storage from format is what’s enabled Microsoft to stall on ODF by keeping the brightly colored ball bouncing in the public eye while quietly and effectively selling internal integration as the justification for knee capping ODF at every opportunity.

Force content and format information to be stored in a single, consistent, way however and Microsoft’s wiggle room get reduced while accurate format conversions become easier - meaning that a new Oracle Office would ultimately have an enormous impact directly in terms of getting truly open document format standards widely accepted and used - and indirectly in terms of empowering the Linux desktop.

Second, for this to really work in light of Sun’s existing licensing commitments, Oracle would have to open source the database and communications components for Oracle Office. Since MySQL is well suited to the job and already open source, my guess is that the pros and cons of using it would then tilt in favor - meaning that my hypothetical Oracle Office would boost the MySQL community first by creating long term support commitments and secondly by putting it at the core of a lot of Exchange replacements.

One note:

1. I’m assuming that Oracle will neither want, nor be able, to sell Sun’s MySQL asset as a going concern.

   The reason I’m making this assumption is that Sun’s billion dollars didn’t so much buy an asset as deny IBM the chance to increase Sun’s dependence on Oracle for hardware sales - meaning that there isn’t much there, and simply spinning the commitment off into a tax exempt foundation would return far less to Oracle shareholders than using MySQL to gain credibility and support in the open source community will.

April 25th, 2009

A question about Linux

Posted by Paul Murphy @ 12:15 am

Categories: Development, Enterprise Policy, General, Hardware, Infrastructure, Linux, Strategy, Sun

Tags: HP-UX, IBM AIX, Intel X86, Data-processing, Linux, UNIX, Operating Systems, Open Source, Sun Solaris, Software

Every once in a while I get email like this one:

I’m interested to get your opinion on Linux and how it compares to UNIX and other UNIX-like operating systems. The desktop distributions I have tried have all had their issues, but most of those problems are cosmetic in nature. However, I’m wondering how Linux compares with the other operating systems from a kernel and design perspective. To me it seems like Linux has a lot of backing and buzz, but I also hear rants from people such as Theo de Raadt who say things like “Linux has never been about quality. There are so many parts of the system that are just these cheap little hacks, and it happens to run”. However, the same could have been said about Unix 30 years ago. Personally I prefer FreeBSD due to its simplicity and cohesiveness. I also think that AIX/HP-UX/Solaris are much more capable but will eventually disappear because they cost money (except for Solaris) and Linux doesn’t.

If Linux really is crap and it’s going to kill “better” operating systems, is it a bad operating system in the long run? Thanks for your insight.

To which my typical response runs something like this:

1) kernel comparisons take a lot of time - time I don’t have, sorry.

2) however, Linux is NOT crap. It works -and it works well. Compare Linux to Windows in terms of its failures and susceptibility to attack and it’s insanely great.

What de Raadt and others (including me) mean when we say Linux is a second rate Unix is that Solaris is better for users and OpenBSD is more secure - but remember that the gap between Solaris and Linux is quite small relative to the gulf between Linux and Windows.

The biggest thing both for and against Linux is that it has always been heavily optimized for x86 - it started as a minix kernel hack aimed at making it run better on x86 by replacing object messaging with x86 interrupts. On the ++ side, it’s very efficient and reliable on x86, on the - side the internals can be seen as a collection of appeasements aimed at x86 design weaknesses.

3) and please keep a sense of perspective: Linux works - and if kernel designers like de Raadt get excited about a competitor’s bad choices - well, they’re right, but that should only really matter to people who share their interests. From a user perspective, the weaknesses built into Linux can be seen as operational strengths (provided your hardware is x86).

Notice, however, that this response doesn’t address his last paragraph:

I also think that AIX/HP-UX/Solaris are much more capable but will eventually disappear because they cost money (except for Solaris) and Linux doesn’t. If Linux really is crap and it’s going to kill “better” operating systems, is it a bad operating system in the long run? Thanks for your insight.

And, of course, it’s the claims he makes here, and the question he raises, that I want to address today.

First he postulates that free Linux will kill AIX and HP-UX while hurting Solaris.

HP-UX was always another Unix that sucked horribly but worked extremely well on PA-RISC machines with the memory and bus architecture the 10/11 kernel was designed for, but like Linux with respect to x86, isn’t efficient when ported to other architectures. As a result it scales well on Itanium, but everything runs well below processor potential - and even on the N-Class machines (which had PA RISC CPUs in boards designed for Itanium) performance never rose to two thirds what it should have been based on cycle for cycle comparisons to the predecessor K class.

Thus HP’s slow strangulation of the HP-UX golem on Itanium is simply collateral to their decision to kill PA RISC in pursuit of a strategic role as the service supplier in an HP, Intel, and Microsoft computing world triopoly.

Linux had nothing to do with this - it’s HP’s executive, not Linux, that’s killing off x86 competitors; and, correspondingly, it’s the opportunity to sell x86 against SPARC, and Power (along with Alpha and PA RISC) that powers the company’s support for Linux.

AIX, in contrast, probably has a future simply because Linux on Power hasn’t been a big winner for IBM.

It’s my belief, by the way, that the reason for that has little or nothing to do with technology or applications lock-in and almost everything to do with mindset. Look closely at AIX and what you see is two intermingled, but very different, kernels: the core NCR/AT&T Unix that makes it all work, and twenty years of encrusted competitive and authoritarian enablement making it saleable to data processing people.

Thus what makes it both possible and necessary for IBM to upsell its adherents to AIX from Linux on Power is largely that Linux lacks the control orientation inherent in the data processing mindset - think of it this way: the AIX default on just about everything is Deny, the Solaris default is generally Allow, and lots of data processing people get very uncomfortable around Linux because it falls closer to the Solaris standard than theirs on trusting users.

If, for example, you wanted to migrate a Sybase ASE 12.51 ERP instance running under HP-UX 11i to 15.1 on both Solaris and AIX, you’d find that the Solaris port takes about two hour’s work with a day’s worth of waiting around for things to happen, while the AIX one takes six to ten days - with most of that spent getting disk storage issues resolved (volume management and network storage are performance killers; the job boundaries imposed by the typical IBM data center mean that even trivial changes require meetings and memos; and, of course, the smart thing to do - working directly with AIX local devices - is a lot like hanging wallpaper in the dark) and making sure that both user and system processes have all the right authorizations - because they never do, and you don’t find out until things blow up.

So while I agree that HP-UX appears headed for the scrap heap, I don’t see the data processing mindset relinquishing all the controls and procedural detours they’ve managed to get into AIX any time soon - and in that context IBM’s efforts to get their cell based processors working properly without Linux have something of the slow motion train wreck about them that’s deeply reminiscent of their 1972 decision to can the “future systems” project.

So what’s the bottom line answer here? Linux is a pretty good Unix, and contrary to popular wisdom it’s not killing any other major Unix variant:

• It’s HP, not Linux and not the market, that’s killed PA-RISC and killing HP-UX;

• Linux isn’t killing AIX either because IBM’s decisions about AIX are likely to reflect data processing preferences - and their struggles with Power7 are really about keeping those customers buying.
• Contrary to Red Hat, Linux isn’t killing Solaris either.

  You hear about this all the time, but it’s mostly just FUD being spread by people who want to sell you a free Linux - when you look at the actual decisions people who replace older SPARC/Solaris installations with new Lintel stuff are making, it turns out they’re generally doing it because x86 is now performance competitive at the loads those machines carried, easier to get approved by senior managers who read drug store ads for PCs, and preferable to actually going Wintel.

The second part of his question was whether the role he sees Linux playing in the deaths of HP-UX, AIX, and Solaris makes choosing Linux counter-productive in the longer term. The answer to that, I think, is No - because Linux isn’t killing any other Unix; it’s giving people who want to move to x86 a better, and often cheaper, alternative to Windows.

April 11th, 2009

Using open source to reduce business risk

Posted by Paul Murphy @ 12:15 am

Categories: Applications, Database Management, Enterprise Policy, General, Linux, Strategy

Tags: Open Source, Paul Murphy

Everything considered, especially the fact that it’s Easter weekend, you’d expect this blog entry to be about Sun’s nearly miraculous escape from IBM and its future as an employee owned business - but it’s not, because I think it’s too early to tell just how much damage IBM managed to inflict and only the sheer viciousness of the personal attacks now being made on Jonathan Schwartz and other key Sun players supports the idea that Southeastern has played its last big card, and lost.

What I’d like to talk about instead is using open source to protect your IT operation from some of the third party risks attendant on the current recession/depression.

The proposition is that if you spend enough on licensing and support for proprietary software that could be replaced by open source products that not paying those fees would let you hire and retain at least two open source developers, then doing so makes sense because it reduces operational risk while increasing downstream flexibility. Read the rest of this entry »

April 4th, 2009

Betting on the future: T2 vs. Nehalem

Posted by Paul Murphy @ 12:15 am

Categories: Applications, Database Management, Enterprise Policy, General, Hardware, Linux, Storage, Strategy, Sun

Tags: T2, Sun Microsystems Inc., Sun Sparc, Advanced Micro Devices Inc., IBM Corp., Nehalem, Sun Solaris, Processors, Servers, Operating Systems

It’s April, quite a lot of senior IT people in larger business and government organizations have signed-off on the typical third to half of the fiscal 2009/2010 IT capital budget they pre-committed during fiscal 2008/9, and are now looking at what changed, and what didn’t change, since the assumptions underlying the other half to two thirds of that budget were made - particularly those with respect to purchases from Sun and the Wintel partners.

My advice? wait - wait until the change implications deriving from Intel’s Nehalem server releases are made clear by others, and wait until the fat lady really does sing on IBM”s attempt to kill Sun’s SPARC/Solaris combination.

One of the things to notice about the Sun mess, by the way, is that none of the assumptions and extrapolations being made by pundits trying to ingratiate themselves with IBM by describing what’s going on as initiated by Sun’s management and then interpreting that assumption as proof positive that Sun’s own senior people think the company should be put out of its misery, are correct.

What’s motivating IBM is precisely the opposite: they know they can’t compete with Sun in the marketplace and are using the opportunities created by the moral meltdown in Washington to shut Sun down via the legal and financial systems.

Sun’s salaried management believes fiercely in the strength of the company -but are employees ultimately responsible to the board. So when a major shareholder first forces that board to consider selling the assets and then “facilitates” that process, people like Jonathan Schwartz have no choice but to mutter “Aye Sir”, saddle up, and march off to do their best at the opposite of what they know is right for the company.

As I’ve said before, the right answer for both IBM and Sun would be for Sun to sell its x86 businesses to IBM along with guarantees on Solaris support - followed immediately by a move to employee ownership, and then putting all the wood behind Solaris and the CMT revolution. That would be good for Sun, good for IBM, good for the OpenSolaris community, good for AMD - and good for the country and the industry because it would increase both technology and price competition in the market while not just letting about 25,000 Sun people keep their jobs, but making those jobs both more demanding, and more rewarding.

Sadly, it’s not likely to happen - but the fat lady hasn’t sung yet and, meanwhile, there are lots of technology planners with money to spend and no real clarity about what to spend it on.

Intel’s Nehalem processors look pretty good in this context: an optimized Solaris (and matching compiler changes) for Nehalem is due out next week, the server processors are a manufacturing step ahead of AMD, and nearly all current x86-64 Linux and Windows software should work out of the box.

Overall, right now, the risks on Nehalem are lower than they are on SPARC - but that’s entirely an artifact of legal and financial manipulation, not the technologies. Thus people considering Nehalem as an alternative to SPARC need to think about three things:

1. Nehalem does not appear to address any of the x86 security issues: from ring 0 vulnerabilities to BIOS based boot processes and the reality that you cannot know what software is hidden in various hardware components, Nehalem offers significant performance improvement on throughput and power use, but does nothing new on security.

   In other words, if you care about security, your choices remain SPARC or Power- and installed Power systems cost close to ten times what comparable SPARC systems do.

2. Neither the Windows software environment nor that for Linux yet takes full advantage of Nehalem’s multi-core structure. As a result, Intel’s advance on its dynamic cache sharing facility in which one or two cores in a four or eight way chipset can speed up while others drop in frequency, can be very valuable in terms of getting some jobs done - but is more useful in workstations or intermittent duty servers than in heavy weight, continuous processing, applications.

   What you need to think about, therefore, is what your workload looks like - if you’re doing file or mail serving to a small number of users Nehalem will be a winner for you, if you’re serving Java pages to a large number of users both Sun’s T2 line and AMD’s latest products will give you more throughput for less money.

3. You also have to think about the consequences of virtualization. Can Nehalem help you consolidate a dozen or more older x86 servers to one new one running many virtual servers? Yes - provided that you don’t have performance killing network or storage bandwidth limits to contend with.

   If, like most people, you do, you’ll find that the T2/super thumper combinations offer hardware support in all the right places, cost less than Nehalem to begin with, and could let you put off fixing your network and/or storage technologies for at least another year or two.

   Still, there are lots of variables here so the right answer is extensive testing - think of it as visible due diligence on the decision and chuckle quietly because it also buys you what you need most: time to see which way the dust settles on both Nehalem and Sun.

In summary:

• Nehalem adds minor technical improvement to what’s really a combination of Intel’s manufacturing and sales capabilities with AMD’s technologies - and the results look pretty good, except that the wintel industry still isn’t ready for multi-core, and the general purpose data center change implications of going this route have yet to be worked out.

• Sun’s T2 and forthcoming ROCK cpus continue to lead the market but are under threat of extinction by IBM.

So what should you do? Wait.

For those of you convinced that Nehalem thumps the T2, I’d suggest reading Anandtech’s typically laudatory report on Nehalem performance. Written by Johan De Gelas, this is a nice piece of work (despite being on a dedicated Intel puff site :) ) but what I suggest you pay particular attention to is the benchmark results he cites - and not just the numbers he shows. Follow up on the public benchmarks cited, and you’ll find many include other, rather less flattering, results relative to SPARC and AMD.

The first one he cites, for example, is the SAP SD2 “ERP benchmark”. Follow that to the sap results site and you’ll find that the dual 2.93Ghz Xeon 5570 (8 cores) Nehalem’s score of 25,000 “saps” is about five times that obtained by a low end IBM Power 570 (2 Processors) but only about two thirds of the 37,650 reached by a Sun 5440 (4 T2+ processors) at 1.4Ghz.

March 28th, 2009

A recommendation on leasing new equipment

Posted by Paul Murphy @ 12:15 am

Categories: Enterprise Policy, General, Government, Linux, Strategy, Sun

Tags: Lease, Hedge, Leasing, Inflation, Currency & Foreign Exchange, Finance, Paul Murphy

You’d think that investing in new equipment just when we might very well be entering the second great depression would be suicidal - but I think getting other people to invest in new gear on our behalf may well be the right thing for many of us to do.

Specifically, it’s time to look at your capital spending plans for the next year or two and ask which bits of it you could sensibly acquire on a leased basis today or tomorrow.

The reasons for doing this are compounded from the following:

1. leasing provides a great hedge against inflation - and if you don’t think inflation is about to become a big issue, consider this graph from a PowerLine discussion of the federal deficit.

   All that cash has to come from somewhere, and simply printing a few trillion American dollars is going to appeal to the weathermen setting up this disaster.

   The inflation hedge here is that if you take a five year lease on new gear now and are still in business in 2012, you’ll be paying today’s rates in 2012 dollars - and every time inflation lets you raise your prices, your real infrastructure costs will go down.

2. Some of that cash will, however, come from greatly increased taxes - and, in most countries, payments against a carefully structured lease are deductible from income.

   The tax hedge here is simple: better to spend money on infrastructure you can use than on bureaucrats whose net effect on you is negative.

3. Most new equipment is more efficient, particularly with respect to power and space use, than old equipment and usually comes with significant warranty coverage. The leased gear should, in other words, directly reduce existing infrastructure maintenance and support costs.

   The “Moore’s hedge” here is that technology has been getting better - but Intel’s Nahelem plays catch-up with AMD and if IBM’s takeover of Sun turns out to be real that puts SPARC out of business, and so the marginal gains available from buying next year’s product will likely be considerably less than those available from buying this year’s products.

One of the less obvious ideas in this compound is that the new gear you lease should be grossly over specified for your current needs, be obtained on a long term lease, and use a technology likely to survive years of neglect - because, if the recession deepens and extends into depression, you will not be able to afford the staff needed to keep things current - and even if it doesn’t, you’ll still find yourself competing with the government for IT staff: pitting whatever you can offer against a secure, inflation proof, government benefits package.

Specifically what to lease is a different issue - as usual, I’d suggest taking a close look at Sun’s stuff because their gear, from modular data center down to the single T5200 box, is likely to continue working whether or not IBM gets to destroy the company.

The negative on this strategy is that you’re throwing away some residual value in what you have and trapping yourself into hardware/software configuration that works today and is configured for tomorrow, but will force you to live with hopelessly obsolete gear in the last year, or years, of the lease. The operative word here, however, is “live” because the real bottom line is that if things go further downhill, then early and effective implementation of this strategy could save your company significant downstream monies - and if a miracle happens? Well, you can always buy out your leases just as you would any other financial hedge.

Starting Monday I’ll be taking a break on blogging - and plan to file commentaries only on Saturdays until September. Why? Well, did you look at that graph above…?