It's Happening Again

This week is VMworld in Cannes -- a big deal in our world, and maybe yours.

I've been in this industry long enough that I can occasionally see very long cycles repeating, and I think I'm seeing another one.

The story is familiar: established ways of thinking are challenged by new ways of doing things.

But this time, the players are different.  And in some cases, there's the irony of role-reversal.

A Bit Of History?

I came into the workforce during the Golden Age of UNIX in the mid-1980s. 

Having cut my teeth on BSD 4.x, AT&T's System III and later System V, I found much demand for my skills amongst the various UNIX minicomputer vendors of yesteryear.

As part of my duties, I would occasionally find myself in front of people who had invested big in the last wave of minicomputer technology: VAX, S/36, DG, HP MPE et. al. 

I had to convince them that -- yes -- the game had changed, and they needed to look differently at their technology roadmap going forward.  The future was going to be UNIX.

Part of the pitch was functionality: UNIX could do most of what you were doing today, and had some unique features that you could use.

But most of the pitch was based in economics: use of commodity technology, standardization, portable applications and larger talent pool.  You could perhaps argue functionality, you couldn't argue economics.

Fast Forward To Today

As part of my duties, I occasionally find myself in front of people who have invested big in the last wave of UNIX technology: Solaris, HP-UX, AIX et. al. 

I find myself trying to convince them that -- yes -- the game has changed, and they need to look differently at their technology roadmap going forward.

The world is heading towards applications that run just-enough-OS (think Windows or Linux) running on a virtualization layer.  VMware uses the term "virtual data center operating system" to describe the emergent environment. 

Legacy OSes do what they're good at: supporting existing applications and information.  The virtual data center OS takes care of the plumbing -- whether internal, external or a dynamic combination.

And, once again, I find myself posing two arguments: a functional one and an economic one.

The functional one is pretty simple: here's how a virtual data center OS does all the things you are doing today, and a few interesting ones that you probably just can't live without. 

Although the debate goes back and forth about what's possible today, there's absolutely no arguing that boundary of "what's possible with VMware" moves up and to-the-right with each release. 

The trend line is inarguable.

The economic argument is strangely familiar as well.  The use of commodity technology - in this case, x64 processors from multiple server vendors.  Data center standardization -- one way of doing things.  Portable applications -- although this time around, we're talking *really* portable and it includes desktops! :-)

Regarding the talent pool, no surprise, VMware certification is a red-hot skill these days.  Even within EMC, it's the most popular certification going.  Again, look at the industry trend line -- it's up and to-the-right.

Are We In The Midst Of The Changing Of The Guard?

It's early days, but I'd offer -- yes, we most certainly are.  The dynamics between 1987 and 2009 are strangely familiar to me -- even to the recession we were experiencing at the time, which tended to -- ahem -- accelerate certain kinds of change.

The grinding noise from the legacy OS vendors with their legacy processor architectures is beginning.  We'll be hearing more from HP, IBM and Sun on this topic-- I'm quite sure.  Itanium, anyone?

And when change is in the air, it tends to affect people differently.  Some smell the opportunity, and dive in.  Others circle the wagons, and see if they can wait it out.

No one is right, no one is wrong.  Everyone makes their choices as best they can.

But, as we look back from five years in the future, will we see this particular period of technology history as the "Golden Age of VMware?"

Wish I could have made it out to VMworld in Cannes -- for those of you who are out there, have fun!

Comments

[Brainy]

"But most of the pitch was based in economics: use of commodity technology, standardization, portable applications and larger talent pool. You could perhaps argue functionality, you couldn't argue economics."

"And, once again, I find myself posing two arguments: a functional one and an economic one."

"The grinding noise from the legacy OS vendors with their legacy processor architectures is beginning. We'll be hearing more from HP, IBM and Sun on this topic-- I'm quite sure. Itanium, anyone?"

Well well, interesting to hear this from you, Chuck.

Isn't EMC Storage legacy by your definition, that will be replaced with commodity components? Proprietary Hardware running proprietary software, with only a handful of skilled people able to operate it.

I can already hear the OpenStorage guys moving into your direction...

[Chuck Hollis]

Hi Brainy

Umm, today most of EMC's arrays are built mostly out of what most people would call "commodity components", so I guess I'm missing your specific point here.

But there is a good observation here to be had: as data center architectures change, the storage technology -- and how it's managed -- will change as well.

And I think we're up for it!

-- Chuck

[Brainy]

Chuck,

I wouldn't call the CPUs, the Memory, the Operating System (Firmware?) etc. used in the Symmetrix and other arrays "commodity".

Otherwise we could call IBM Mainframes commodity as well. I hope you get my point.

-- Another point. Hypervisors are getting more and more powerful, providing support for lots of different devices. Until they are, (surprise! surprise!) an OS themselves. Hey, someone still has to do all those infrastructure services!

There has been an enormous uptake for ESX. But interestingly, 90% of the installed base is virtualizing MS-Windows (talked about it lately with folks at an ESX training. Hey, I'm certified too!).

I can tell you why. Windows was never good in running more than one application on one server. This was never a problem under e.g. Unix. This is why Unix deployments under ESX are rare.

And why would you pay for ESX, when virtualization under Unix is free, and has less overhead?

Which brings me back to our legacy discussion. With all those open and free Hypervisors coming out (Xen, KVM, xVM Server), wouldn't you consider ESX to be the legacy one?

Now, a lot of people just got rid of closed-source software (MS-Windows...) and replaced it with Linux, OpenSolaris, *BSD why should they lock themselves into closed-source and legacy software again (ESX) ?

[Chuck Hollis]

Hi Brainy

The CPUs, memory, drives, power supplies, etc. are all bought from the same suppliers that the rest of the industry uses.

The software is, of course, EMC developed.

Regarding the second part of your discussion, I think you're missing a key point -- it ain't just about hypervisor efficiency, there are larger issues that come into play when you think about the big picture.

Thanks for sharing your thoughts ..

-- Chuck

[Tommy]

And as with last year, no local EMC people was present.

[Chuck Hollis]

Tommy --

????

Best regards,

-- Chuck