Why argue about bits of cloud?

It is curious that, against the background of last week’s VMworld conference where the focus of attention was on Software Defined Datacentres (SDDCs), a significant discussion sprang up over whether the world should go public cloud or hybrid cloud.

That discussion was hinged around a debate between Marc Andreessen, founder of VC firm, Andreessen Horowitz and the founder of early web browser pioneer, Netscape, and Pat Gelsinger, CEO of VMware. Andreessen was firmly in the private cloud camp, suggesting that this was the only way to go. Gelsinger, on the other hand, proposed that the only way is hybrid cloud.

Of course, they are both wrong – not least because they are both right. In other words this is a nonsense argument that is irrelevant to what any business needs to be thinking about these days.

This is especially the case with SDDC coming to the fore. Put at its crudest, this creates an environment in which each user can build the type of cloud environment that suit their needs best. It is therefore not the case that there is any one solution that represents `best solution’, for it will be different for each company.

To be fair, it can be said that VMware’s Gelsinger is `more right’ by suggesting that the future is hybrid cloud. But that is arguably only the case in the near to mid-term future - and for established businesses. And to be fair to Andreessen, he was directing much of his argument towards the needs of the start-up business. His observation that the total Capital Expenditure required for a start-up these days is the cost of a few laptops to connect to a cloud service is broadly accurate.

But even here, to propose that the public cloud is the only answer is stretching things. While it will be the case for many that services like Google Apps and Microsoft Office 365 will provide all the resources required, there will be many exceptions. Even small companies can have processes that require a faster, guaranteed performance than is reliably possible in the cloud, and once trading overseas becomes part of the business picture it is quite possible that Governance and compliance issues could inhibit moving onto public cloud services.

There are, it is true, local service providers that can help with some of these problems. Companies like Xanadu and Carenza, for example, can provide localised services based on Microsoft’s cloud offerings, going from a version of Andreessen’s start-up zero-Capex model through to full, in the cloud, do-everything-for-you managed services. But the latter, it could be argued, is only a variation on a private cloud model.

And that highlights the fundamental argument: businesses can – and will – use applications and services that are either and/or both on-premise or off-premise. After that, the rest is a matter of appropriateness. There are now a decreasing number of reasons that justify running IT on premise.

These include situations where an application has to be run with extremely stringent latency issues. Here, running on-premise may be necessary because using cloud-delivered alternatives involves too much risk.

A second example is where there is a constant business process that requires no scaling, up or down, to meet fluctuating service requirements, and where the existing on-premise resources are not amortised. In addition, the deployment requires no upgrades to meet the stable performance level the task requires. In the short term, maintaining the status quo probably makes business sense.

But even here there is the increasing option of moving such jobs to bare metal services with a third party – essentially a managed services private cloud - where at least the management and maintenance costs can be defrayed.

There are also still arguments that real security is best maintained by running an application or service on-premise and under `permanent guard’ of some kind, though there plenty of arguments to the contrary. One such comes from Paige Leidig, senior vice president at CipherCloud.

“There is a key misperception that the public cloud is less secure than the internal enterprise environment. As Marc (Andreessen) noted, internal environments are riddled with faults, defects, holes, malware and hackers.  This makes the cloud a safer place for data, given recent innovations and continued new technologies.  As Marc pointed out, CipherCloud encrypts data in real time before it is sent to the cloud, which turns the data into gibberish for any cloud provider or third party attempting access.

“By handing the encryption and decryption keys directly to the organisation, we give customers tighter control over their data.  This creates better security and satisfies privacy regulations that require organisations to be aware of and prevent all third party access.  Our cloud information protection platform also offers cloud data loss prevention, malware detection and user activity monitoring to provide enterprises with additional layers of security and compliance controls.” 

There are also issues where governance and compliance with legislated regulations or best practice requirements create issues most easily solved by running services on-premise. And even where off-premise operation is allowed there can be restrictions such as data sovereignty laws which determine where data may be stored and services run.

The issue, therefore, is remarkably simple, and one which arguments about private, public or hybrid cloud `architectures’ are at best confusing and at worst arguably deliberate obfuscation. Should the IT resources required to run `Business Process X’ be located on-premise for some definable, substantiated reason, or can it be run off-premise using the resources of a service provider?

If the latter, it is a cloud service and that is the end of the strategy argument. One can then discuss tactics, such as should it be on a bare metal `private’ service, right through to the other end of the spectrum, such a deciding there is a really good SaaS provider that is better all round.

And the decision for every business will be different. Even three companies using the same on-premise application – for the sake of argument, SAP – will find four different options to choose from: remain on-premise, go bare metal private cloud services, move to SAP’s own cloud service provision (or one of its partners), or a move away to SaaS delivery with Netsuite.

So I just wish that members of the IT vendor community, and especially the ones deeply embedded in the provision of cloud services, would forget trying to dissect `cloud’ into constituent parts that serve no purpose from the users’ point of view. There are much bigger marketing `fish’ to fry getting users to appreciate what they can achieve using whatever `cloud’ is best for them.

I fear, however, that they have too many vested interests to make that likely, at least in the short term.