Paul Maritz

  • April 28, 2010
  • The VMForce announcement leaves us all with one big question — ok, many questions like when and how much.  But the question I am most interested in at the moment is whether or not this is a single or multi-tenant thing and what it means for the industry and Salesforce’s multi-tenant chops and possibly, how this all plays in the discussion of public and private clouds.

    Let me start by digressing.  My understanding of VMForce is that it offers a way to move Java code to the cloud while enabling it to access data from Salesforce Cloud applications.  This happens automagically when a developer selects VMForce as the server for an application.  VMForce.com provides a virtual Java server and voila the application is available to users of the cloud instance along with other Force.com applications with proper security.

    That all looks good but just because the Java code is running on the VMForce platform it does not mean the Java code is suddenly multi-tenant.  Does it matter?  Hmmm.  The VMForce virtual server is a multi-tenant device conjured up from multi-tenant resources, but the Java code is separate so it’s operating as a single tenant instance.

    Someone send me mail if this is not the way it works.

    The net effect to the customer is a familiar application running native in Salesforce and able to access all of the user’s data and other applications so the difference is nugatory to the user.

    How is this different from running the Java application on a hosted server in the sky?  Well, first off, the server in the sky runs like the server in the data center which is to say it is walled off from the rest of the world and integration with a cloud (or any other application) requires a more tedious and conventional integration process (read time and money here).  So you get much simpler and less costly integration and the ability to run concurrently in one cloud environment.

    Is this a private cloud?  I guess so but only to the extent that by providing an instance of Force.com to any user, that user has a “private” cloud that just happens to be integrated to the rest of the world.

    Is this a departure for Salesforce?  That’s debatable but I lean towards saying yes here so I am calling it smulti-tenant.  I’m not very concerned about doctrinal purity here.  The facts as I see them are that this approach merges legacy applications into the future of cloud computing.  The alternative, moving your data center to the sky or using infrastructure as a service, does nothing to move legacy code into the future.  It just changes the location of the private data center.

    As an engineering proposition this is elegant, like making the strongest and lightest airframe or bridge.  If Cole Porter were a Java programmer I am sure he’d say ‘swonderful, ‘smarvelous.

    Published: 14 years ago


    I got this long comment on yesterday’s VMForce post and I disagree with some, but not all of it, and rather than just posting it and letting it run, thought I would comment using it as the Q part of a Q&A.  Here it is.

    “Great analysis on vmforce announcement.”

    Ok, you didn’t expect me to disagree with everything did you?

    “At the outset vmforce will benefit vmware by providing their Java developers instant access to cloud services.  Also vmforce will benefit salesforce.com by increasing adoption of Force.com platform among Java community.  From developers perspective I believe today’s announcement is a ground breaking one that is a win-win for both salesforce.com and vmware.

    This is true though as I told John Pallatto at eWeek, I think the early adoption will be among Java developers employed at companies already using Force.com and possibly ISVs and SIs who have catalogues of Java code they’re just itching to deploy in the cloud.  There are six million Java developers and even one percent of them represents a lot of code.

    “However today’s announcement is short on details around business deployment scenarios and pricing models.  I hope today’s announcement is just an incremental step towards a much larger strategy on open cloud infrastructure that addresses some concerns around cloud computing like security and scalability for enterprises.

    Salesforce has a well-documented and time tested approach to its announcements.  Like a good baseball team, that looks fresh and crisp on the field, it’s because they execute well on the fundamentals.  For Salesforce, this means the old rule of three — tell them what you are going to do, do it and then tell them what you did.  The company has gone through a year many times hitting on this formula for every major advance and VMForce is no different.  This was round one.

    Not sure where we get the security and scalability for enterprises issue that the writer is talking about.  These were issues in the middle of the last decade that haven’t been raised recently because they literally went away.  I noticed at the NetSuite partner meeting a couple of weeks ago that CEO Zach Nelson was dealing with these issues in ERP but they’re about as valid in ERP as they were in CRM.  Salesforce has many enterprise customers with thousands or tens of thousands of users, what do they need to do to put the scalability issue to rest?  And when was the last time your credit card information was stolen from a Salesforce application?  I’ve got a lifetime subscription to the credit rating agencies from all of the enterprises with conventional IT who lost my credit card information.  And there are dozens of companies in Silicon Valley that have lost valuable IP to state sponsored IT pirates from China.  They had conventional security too.  Sorry, but I just can’t buy the argument that SaaS is not scalable or secure.

    “Currently salesforce.com lacks support for private clouds for enterprises and only supports public clouds through its hosting services.  Majority of enterprise customers will be reluctant to migrate to public clouds until concerns around security are addressed.  Industry market trends point to migration towards “private/managed” clouds by enterprise customers in the next 3-5 years time-frame.  I hope salesforce.com will announce products or partner with companies like vmware to fill this gap.

    Show me your 3-5 year data please.  I see 1.4 million users and 70,000+ companies using Salesforce.  Let’s call them the early adopters if we must.  I am sure some enterprises will not migrate to the cloud no matter what.  My prediction is that they will all eventually be headquartered in some inland Rocky Mountain state, dig bomb shelters and join the NRA.

    But I digress.  The definition of public and private clouds is something invented by the part of the industry that thinks a cloud is infrastructure as a service or IaaS.  I don’t agree.  IaaS is a data center in the sky, but it’s not cloud computing.  To be a legitimate cloud you really need platform or PaaS and software or SaaS.

    Then there’s that security bogie again.  A data center in the sky will still lose my credit card just as well as one on land unless enterprises devote many more resources to the job — something they won’t spend sufficient money on.  SaaS and Cloud providers spend on this as a matter of course.  The private/managed clouds idea took a big hit yesterday when VMWare and Salesforce showed how to bring a Java application to the security of a scalable VMForce virtual server.

    Finally, Force.com has always offered private server capability, that’s what you get when you run your enterprise on an instance of Force.com.  What the writer is alluding to when he says “private” is more about single tenant, dedicated hardware in the data center in the sky.  Of course this approach destroys much of the rational and cost savings of cloud computing so why go there?

    “In addition current public cloud deployments models have some limitations on scalability and performance front.  While multi-tenancy is good from h/w utilization perspective, due to the inherent sharing model it is not ideal for compute intensive and complex data processing applications.  Until this limitation is addressed few enterprises will be willing to migrate to public clouds.

    Now its scalability and performance.  Ok, you’ve got me, the cloud is probably not the place to do molecular modeling or tornado tracking — though the human genome made great and creative use of spare PC cycles unraveling DNA.  So, all of you molecular modelers and complex calculation buffs are excused from class.  The rest of us will be just fine running our business applications in the cloud, I suspect.

    “I hope trailblazers like salesforce.com and vmware will address these limitations around public clouds soon.  On a final note we seem to have a new “— as a service” acronym pop up every other day.  Unless each and every one of these services is tied to “customer value proposition” we will just end up with technical jargon that only confuses the end customer.

    I think we did all that yesterday, at least at the announcement I attended.

    Published: 14 years ago