Main content

Understanding the Salesforce-HP relationship

Den Howlett Profile picture for user gonzodaddy November 20, 2013
Digging into the HP-Salesforce relationship to understand why any copmany would wish to burden themselves with a problem that Salesforce takes care of today.

We have already talked about the Salesforce-HP relationship. I believe it represents the opening up of a major front against Oracle and SAP, Phil calls it the Vanity Cloud. Regardless of your view on the marketing side, Salesforce was very clear about what it is not. Stuart Lauchlan quotes Marc Benioff:

“It’s not single tenant. It is an innovation on our model.

“What we have inside our data centres is multi-instance, multi-tenant.

“That architecture has worked incredibly well for us, but have had large customers come to us – HP being the first – who have said ‘we love what you do, we love the pods, but we want our own. We want to use multi-tenancy but we want to buy a dedicated instance of it.

“So if you walked into our data centre today you’d see 15 racks. Now what we’re providing is an opportunity for a customer to come in and say ‘I want to put my rack next to those racks, put my logo on it and run my own brand’.

“But when we update our software, all the pods are updated at the same time. The most important thing is that we have the same software and the same model. But instead of 15 pods, we have 16.”

But as we all know in software, what is said one minute is not necessarily the same as what happens in the real world. In an effort to put the marketing aside and get to the technical nuts and bolts, I spoke with Steven Tamm, CTO Salesforce.

I posited that in the case of a say a GE, the businesses are so complex and different that the hopes of running the same software on a single instance would be harder to sustain. In that case I could envisage a GE coming to Salesforce offering the argument that consumption patterns preclude everyone from consuming everything at the same time but that they might well pay a significant premium to have flexibility in when they make switches.

His response was blunt and unequivocal:

"We wouldn't take the money - we'd refuse. We update the API three times a year and we simply cannot have different versions running. However, we can certainly let them run in the sandbox for testing and for some customers we will actively monitor for up to 75% bug fix."

So if the multi-tenancy isn't broken then why would anyone seek to have their own Superpod? I've already made the argument as it applies to the Global 2000 but there is another reason.

As Salesforce starts to get serious about integrating all its (active) acquisitions as it is with Salesforce1 as a first step, the portfolio becomes much more complex. It's what happens with software. The architecture on which is sits is now 15 years old - a lifetime and more in software terms. Salesforce is working to update that but even so, there is a degree of brittleness in the architecture that leaves it vulnerable to breaking down. There was an outage just before Dreamforce got going. The registration system fell over on Sunday. You get what I mean.

Running a GE or ANO global company inside their own managed infrastructure acts as a shield against such potential failures. Buyers will pay handsomely for the pleasure but the balance against risk will be considered worthwhile. Whether it works is another matter.

A grey colored placeholder image