Oracle's Virtual Play

Of course, you probably saw Oracle's announcement regarding their support of Xen, the hypervisor alternative now part of most Linux distros.

Monday morning, I hopped on a plane for Hong Kong.  When I got off, my blackberry immediately filled up with all manner of commentary about this announcement.  Of course, I couldn't sleep last night, so I had plenty of time to think about things.

And, I'll tell you -- I went from being curious, to being amused, to being downright disappointed.

The Basics

Most everyone realizes that virtualization of IT is one of those once-in-a-career, gee-this-changes-everything types of technologies.

And that, despite VMware's amazing success in this space, sooner or later the market will demand alternatives.  Nature abhors a vacuum; markets abhor a near-monopoly.

It's fair to say that lots of folks are watching very closely what the VMware competitors are doing: Xen, XenSource, Citrix, Virtual Iron, Microsoft, et. al.

There are other players as well that are important.  Server vendors, for example, can help or hurt alternative virtualization schemes.  The availability of supporting infrastructure (e.g. management, security, backup, etc.) is also important.

And the application vendors are important as well.

Why Application Vendors Are Important

No IT user wants support hassles. 

So it's routine to ask any OS vendor, "does application vendor X support your operating environment?".  True for all the UNIXes.  And true for server/desktop virtualization.

Now, it doesn't matter if the application in question works perfectly; the real question is that -- if there's a support problem -- will the ISV in question will step up and fix the problem?

I think there are two reasons why the answer will usually be a rather unclear "well ...", especially for things like VMware.

First, it costs more for the ISV to support an additional environment, and unless they can see clear incremental revenue (e.g. licenses, maintenance, etc.) it's hard to make an internal case. 

Of course, customers beating you over the head helps a bit to move that discussion along.

Second, I think no major ISV really wants interloping vendors below them in the stack.  Look at Oracle's repeated efforts to commoditize everything below the database: remember "bare metal" Oracle?  Or SAP's preference for Linux?  Even Microsoft doesn't have a strong incentive to let others in here.

Ask any ISV of any size a direct question: do you support VMware? -- and you'll get a very contorted answer as a result.  It has absolutely nothing to do with technical realities and everything to do with business models and strategic advantage.

Application Vendors and Virtualization

Broadly speaking, there are four potential advantages of running an application in a virtualized environment. 

One, you might be able to save money via resource consolidation.  Two, you might be able to manage operational aspects of IT more efficiently (think provisioning, capacity planning, etc.).  Three, you might be able to run a better test and dev process with virtual machines rather than physica ones.  And finally, you may find the environment is easier to secure when virtualized.

But, please note, all of these benefits accrue to the customer, and not directly to the application vendor.

So, What Did Oracle Announce?

Nothing really earth-shaking, I'd offer.  At a fundamental level, they offered to sell support subscriptions for Xen, the hypervisor that comes as part of most Linux distros.  Oh yes, there was some mention of a web tool to help manage virtual machines, but I wouldn't get too excited about that piece.

That's about it.

No real "virtualization strategy" here, folks. 

And, of course, a lot of partners who were encouraged to say nice things about what Oracle was doing.

So Why Was I So Disappointed?

First, there was so much more that could have been talked about. 

There's the opportunity for tools to assess environments, and figure out the impact of virtualization. 

Maybe ntegration with Oracle's management tools, such as OEM, that could make Oracle environments easier to run.  Maybe even solutions targeted at test and dev, or security.

None of these were hinted at in Oracle's announcement.  Not even a hint.

By the way, all of these are available to some extent with VMware.

Second, it's pretty clear it's not about customer demand.  I'd invite you to mentally make one list of Oracle customers who are using VMware.  Now make a second mental list of Oracle customers who are using Xen. 

I think the lists would be very different, wouldn't you?  I actually had a representative from a major ISV tell me "well, we're not seeing much demand for VMware support".  Luckily, I wasn't drinking coffee at the time, otherwise I might have sprayed the poor guy.

Third, it smacks of buzzword marketing.  Yes, virtualization is a hot topic.  And, yes, every vendor is figuring out how to claim a stake here.  But, given the timing of Oracle Open World, it feels like someone in marketing was looking for something cool to say, rather than an organized, thoughtful initiative being discussed.

Fourth, there's a not-so-subtle hint of head-in-the-sand here.  As if by ignoring VMware, somehow it will all go away. 

I Guess I Shouldn't be Surprised

Maybe I was expecting too much.  Given what I've seen in the press and on the message boards, I don't think I was the only one who was disappointed.

Broadly speaking, I think many parts of the IT industry has figured out what they're going to do with virtualization.  They realize it's a big deal that means a lot to their customers, and changes all manner of things going forward.

Unfortunately, I don't think Oracle is one of them. 

Yet.

And that's too bad.

Comments

[Tarry Singh]

Indeed, Chuck!

An alternative to VMware does not mean that VMware has to perish. An alternative to VMware means that the customers get to choose depending on their requirements. We are seeing some movement from Citrix's end and some good promising movements from open source management perspectives (like that of openQRM) but I keep yelling : "Stop this battle of the hypervisors!"

Customers are not interested in virtualization, as you point out, their focus is on more serious issues: Resources (both human and non-human) allocation and optimization, Efficiency and obviously security. Adding another layer, be it VMware, Xen, KVM, xVM (which Jonathan will talk about in OOW today) or any other, will simply mean that (1) It has to be managed (2) since its another software stack, it needs to be secured.

Extreme caution and creativity is required to combat that and also try to keep those costs down while the data centers grow as content, in many cases, quadruples (test, development, staging and production).

My worry is that the customers have a much bigger burden to shoulder than the rest of the vendors. Good, inter-operable management platform will/should eventually help them get out daedal nebulae of the renewed multi-vendor lock. Responsible vendors, whether open source or proprietary, will shine out eventually and that is how their businesses will grow.

[Marc Farley]

Oracle's announcement does seem to lack meat. Mostly a statement that some of their applications will run on Xen and that they will attempt to support them. What's the point? Does this mean they don't support their apps on VMware? I certainly know customers of ours that run Oracle on VMware and everything works fine.

[Eugene Sergejew]

I couldn't agree with your comments more, Chuck. Oracle's innane and reprehensible per-core licensing model tells the world all they need to know about Oracle's
'commitment' to server virtualisation.
I spent last week working for a customer who has a very sound and well-advanced server virtualisation strategy - which excludes all their Oracle hosts - simply because of the ridiculous costs Oracle imposes on shared hardware. I see a big future for Oracle alternatives....

[Ernie Oporto]

Xen. Ugh.

[Mark]

I, for one, as an EMC and Oracle customer, am thrilled to hear that they will support Xen, and welcomed their announcement without reservation. I think it sad to find an EMC executive crying and whining because the poor vendor's point of view isn't represented and that their favorite virtualization platform (VMware) isn't getting all of the attention. I think EMC could learn something from Oracle in this instance. Yes it is true that "all of these benefits [virtualization] accrue to the customer, and not directly to the application vendor", but is also true that a benefit for a customer is ultimately a benefit for the application vendor. Chuck may be disappointed in Oracle but I'm disappointed in EMC.

[Chuck Hollis]

I thought I'd post your rather unique point of view, and invite others to comment on your perspectives.

I'd also offer that it's possible to disagree without being disagreeable ...

Thanks!

[Mark]

I don't mean to be disagreeable but I do disagree and I don’t mean to offend. EMC should be working to leverage Oracle's announcement for the good of its (EMCs) customers. In short, instead of complaining about Oracle's strategy take the initiative and give your customers some value-added that they would not have had otherwise.

For example, EMC could offer Oracle’s VMs pre-configured to run EMC products (such as Documentum). Fill in the blanks where they apply to EMC products and leave the rest to Oracle or other application vendors to implement.

I understand that if a customer has a large investment in VMware (or some other product) then it might not be such great news but it is still better than nothing and you can’t fault Oracle for not meeting every expectation or at least you should hope that your customers don’t have such high expectations of EMC.

[Chuck Hollis]

So, I'm not really clear as to what you might be asking for here.

Do you want to see increased support for Xen in terms of EMC products, e.g. storage, replication, backup, management, security, etc. much as we have done for VMware?

Or, are you asking that we offer EMC software products in Xen-based software containers, much as we're starting to do for VMware?

Or, is more specifically about EMC endorsing and/or supporting Oracle's specific distro/branding of the same Xen found in RHEL, SuSE and others?

I'd also be curious as to your reasons for wanting Xen (regardless of distro) rather than VMware support.

The reason being is that supporting each and every flavor of hypervisor virtualization will entail some cost -- of course, we'd want to understand why someone sees an advantage one way or another.

Thanks!

[Chuck Hollis]

I'm sorry, Tarry, somehow I didn't see your comment and was neglectful in posting it here.

I think it's inevitable that we'll see many hypervisors in the market before too long, and -- eventually -- they become near-commodities.

Differentiation will come with the "everything else": management, integration, compatibility, performance, etc.

I'm not knocking alternative hypervisors; I'm knocking Oracle slapping their name on an open-source offering and calling it their "virtualization strategy".

Certainly, I think their customers deserve better.

[Chuck Hollis]

Hi Marc -- your comment was lost in the same pile as Tarry's, and I overlooked it -- my apologies to you both.

Oracle runs *perfectly* in VMware (we too have many, many customers who are doing it), ditto for the Microsoft products, and the SAP products, and ... well, none of these software vendors have said 'we support it' for reasons I tried to explain.

Having it work, and having a vendor stand behind it are two different things in this world.