Cloud starts to get federated

 

The same goes for many cloud service providers, these days. The trick now is the need for them to generate skills in federating cloud services in order to meet the full complexity of their customers’ service requirements.

 

If there is to be one `vendor’ of cloud services, or at least a primary vendor for any cloud deployment, it is going to come from the businesses that have an understanding of what is involved in aggregating, building and managing federated cloud services. These are likely to come from disparate backgrounds, ranging from the old systems integrator community through to industry-sector based organisations such as Trade Associations or Certification Bodies. It is less likely, however, that they will come from the existing ranks of the mainstream vendors.

 

One company pitching itself at just such a role is Computacenter. In its past it has been better known as a PC and systems distributor, but has for a long time had a second string as systems integrator. Now the distribution side of the business is gone, and the SI element is growing into the cloud service provision business off the back of its role as a provider of certified systems implementations and services to a significant number of major businesses, from the financial services industry outwards.

 

Matthew Yeager, the company’s Project Leader on Data Storage and Protection, has found an interesting way to demonstrate how federated cloud services work – a version of a Rubic’s Cube with each square labelled as a product, technology or service. This represents the package of `stuff’ that can make up a full systems deployment process. And getting all those elements in their right places and integrated together correctly is something that has rarely been achieved. It is with the coming of the cloud that he feels it might at last become possible.

 

“We as an industry, we as Computacenter, are guilty of delivering you that (the cube), we have promised to deliver that. On there are all of the technologies and things that were promised would deliver value. But none in the industry have ever delivered fully on the promise,” he said.

 

“Cloud service providers won’t tell you about the cube – about what’s inside the datacentre”, he continued. “They just want you to connect to some APIs. But in the end customer CEOs and CIOs really want to know what it costs to deliver IT per user/per annum. They are aware that they are now spending a lot of money on IT, and the cloud provider message of `I can solve that’ is tempting.”

 

The trouble then, he suggested, is that the IT department becomes involved. The CEO may like the cloud delivery model but once the techies get involved their message tends to be based on stalling the process: `we’ll get back to you in a year or so’.

 

“To be fair to them,” Yeager said, “it is difficult to continue to provision the existing business, and keep the lights on, and figure out how to consume APIs and how to do things in the cloud.”

 

One of the factors here is that the cloud moves the important criteria away from the established domain of the IT department – the technology stack. In practice the cloud has already made the stack irrelevant. The whole point of the cloud is that it becomes possible to put different technology stacks together as a mix-n-match exercise.

 

The problem, however, is that there will still be a large number of users who feel it is important to know the brand name on the box being used to run their service. And if the box hasn’t got the right brand name they do not want to use it. In addition, there is going to the more practical problem of users that have recently purchased a Brand X system, which is an asset they want to sweat for the next four or five years. So they are not going to give it up.

 

What is important in federating services, therefore, is the ability to integrate those disparate systems, so that workloads can be migrated, and federated service provision models built and applied.

 

To this end Yeager doesn’t see Google or Amazon as competitors, more as potential partners. Much more of a threat are the likes of Japanese telecoms company, NTT, which recently acquired IT services company Dimension Data, which already has a partnership with Cisco. This is aimed at creating a federated service provider, but in the process raises another question for customers: when seeking a cloud services federator, they will need the widest catchment of possible components, technology stacks and applications. In this context, Computacenter’s model as an integrator with a wide assortment of vendor partners could well give greater flexibility and choice than a package of services built on the limitations of acquisition.

 

“The big vendors with the big systems integrators as part of their portfolio have this covered, and have their own stacks pretty much established. But the mid-sized vendors have the problem of creating their own stacks by acquisition. The other way is to form partnerships, but that does require the individual companies in a partnership to fully understand their role,” he said.

 

 

In essence, as cloud service delivery develops the trick for many systems and components vendors will be to understand their place in the stack – and ideally many different stacks – and then stick to their knitting.

 

Ideally, many of those vendors will need to get used to white-labelling their products to others that have the brand names that play best in specific marketplaces. In fact the aggregators and federators themselves could well become white labels in their own right, providing all the service provision, infrastructure integration and service federation for organisations with the appropriate, trusted brand. The brand of the stack shouldn’t matter and won’t matter.

 

As a builder of federated cloud services one of the key technologies Computacenter has to apply is automated provisioning using a web-based service catalogue. This, Yeager indicated, has been an easier job than expected. 

 

“We gave our consultants three days to develop a storage systems catalogue and it took them two hours. They kept testing it because they didn’t believe the results. These were guys for whom storage was not their core discipline, but they had provisioned a storage system with zero help,” he said.

 

“So we realised that it could be used to provision storage environments not only within the business, but could be used directly with customers. It is called Virtualised DataCentre (VDC). Within very short order most of the other major vendors had come up with similar offerings – EMC/Cisco/VMware, then HP, IBM, NetApp, and HDS.”

 

All have the same objective – to make the stack work as a fungible unit – and all of these can run several thousand virtual machines.

 

What this means is that businesses can now get a real number for what such a system costs, how much the depreciation is over four or five years, and how many users they can run per virtual machine. This gives them a figure for the cost per user/per annum. “Hey presto, the business managers have what they have always wanted. And the technologists have what they want, for all just works together.”

 

So the various technology stacks are in fact irrelevant, though many in the industry seem keen on having a `stack war’. What Yeager sees, however, is a stack battle within an API war. “Once I’ve got the Federated Service Provider model it transforms the internal IT department into a service provider, so the business is happy,” he said. “But there is still going to be a need to consume services externally, so it doesn’t really matter what the individual stacks are like. The key will be the APIs, the policy based engines. That is where the real wars will be.

 

“Irrespective of the stack our objective is `Virtualise, Containerise and Mobilise’. We want to virtualise it so that we can containerise it, and we want to containerise it so we can mobilise it, either internally or externally. That is the key to unlocking the federated service provider model. Today we look like a Visa because we are a certification authority for complete implementations regardless of individual vendors,” he said.

 

But the aim now is to look more like Vodafone, delivering managed services built around a service catalogue, though the company can still deliver on-premise if that is what the user wants.

 

Part of the package is then analysing the cost benefits involved, which is something he feels the cloud service providers have still yet to master. “They say, I’ll build this thing and offer it out – do you want it or don’t you? They sell broad spectrum. We sell everything that is on the rubic’s cube, so we already have a good idea of what its costs to deploy a system, and we can wrap that up into a cost benefit model. Then look at compound annual growth rates, where we know that doing `this’ over five years is going to cost this much, and doing `that’ over five years is going to cost that much,” he said.

 

This gives the company an important advantage in the marketplace that few others have so far attempted – a guarantee. “We are then able to create a Delta that I’ll underwrite, and if I’m wrong in any one year I’ll write the customer a cheque. It’s as simple as that.”

 

Most other guarantees, partly because they are from global suppliers offering global suppliers, usually involve offers of free disk capacity, or free professional services. For an Financial Director that just means more stuff that they have to write down. A cheque is in language FD’s understand, and they are more likely to unlock the additional investment needed to make it really work well.

 

“We have never had to write a cheque and I don’t think we ever will. We can certify systems more efficiently and more quickly than vendor partners, and we can underwrite the benefit.”