More than just hot air

Trying to pin down exactly what cloud computing is, and what it is not, is about as easy as trying to pin down an actual cloud. At its broadest, “the cloud” is synonymous with “the Internet”, but that non-definition is not much help to those struggling to sort the hot air and the hype from the reality.

A more useful approach is to think about cloud computing as “everything as a service”. Providing for your IT needs is increasingly not about buying stuff, it's about buying the ability to do things. We all want, for example, to back up our data, to do it securely and reliably and to pay as little as possible for doing so. Exactly where those backups are stored, and how they are made is immaterial as long as they exist and we can get them when we need to. Nowadays, getting someone else to take care of it for us is a real alternative to going out and buying a set of tapes or a spare server.

So: Storage as a service. Likewise software as a service, hardware as a service, infrastructure as a service, data as a service, platforms as a service, disaster recovery as a service and so on. Or, to give them their jargon names, the friendly family of StaaS, SaaS, HaaS, IaaS, DaaS, PaaS, DRaaS and their inevitable cousins yet to be named.

“Cloud computing is about turning IT into a utility,” says RSAWeb MD Rob Gilmour. “With infrastructure, for example, you don't actually need a server - rather, you need access to processing power, system memory, data storage and transport. Those are the services we provide.”

For technology and financial managers alike, the benefits of making the shift away from seeing IT infrastructure in terms of physical hardware are legion.

Thanks to technology that allows multiple virtual servers to run independently on a single piece of hardware, infrastructure service providers can reap economies of scale that are simply unavailable to anyone else except the very largest enterprises.

“You reduce the resources needed to run your physical environment, because you can utilise your physical assets far better,” says regional director of VMware Southern Africa Chris Norton. “Without virtualisation, it's quite common to get an organisation running, say, 1 000 servers at 8% capacity. With virtualisation, you can reduce that to 10 servers running at 70% to 80% capacity. You get significant consolidation.”

Virtualisation - and the move to the cloud that it enables - can make services more reliable as well.

“We now have the ability to move a live, running virtual server from one physical machine to another without interrupting operations,” says Norton. “That removes planned downtime from the equation.”

There is no financial downside to the cloud.

Todd Schoeman, head of MEA portfolio and marketing, BT Internet

In addition to this increased efficiency and reliability, scalability and flexibility are the other most frequently cited advantages of the move to buying infrastructure as a service.

It's a relatively common experience, for example, to discover that when you upgrade the software your business runs on, your existing hardware suddenly can't cope with new demands.

“If your server suddenly needs double the memory, in a cloud computing situation that's a matter of a few clicks,” says Roman Hogh, head of technology and product strategy at MWeb Business.

“We can give you a new server in five minutes,” confirms RSAWeb's Gilmour. “This on-demand scalability is one of cloud infrastructure's biggest selling points when you compare it to the traditional hardware purchase cycle where you need to buy the server, configure it and host it. “Under that scenario you're lucky to get it running in days.”

Adding a new server need not even involve a phone call, says Richard Vester, executive head of hosted services at Vodacom Business. “We're moving to the point where customers can provision their own resources remotely. We can hand over the capability for them to add machines to their resource pool as they need them, based on predefined templates. It can all be done from their desktops in the office.”

“The advantages are all about capacity and scalability,” says Accenture's cloud computing lead Willem Thompson.

Just as attractive as the ability to increase capacity quickly is the ability to reduce it again - businesses only pay for what they need, when they need it.

“Our customers like the idea that they don't have to pay for resources that are only ever used at peak times. If there's a demand spike at month-end, it's the vendor's problem to make sure that demand is met,” he says.

There are other advantages too. When you move to the cloud, maintaining your IT infrastructure becomes someone else's problem.

“By moving applications, systems and services into the cloud, you are removing your requirement to have skills in-house or contracted,” says Hogh. “You can outsource all the complexity; your service provider has to deliver according to your service level agreement.”

Similar arguments apply to the idea of buying software as a service (let's surrender and call it SaaS).

“The whole idea of SaaS is to spread the cost of your system over time based on usage,” says Rony Ross, CTO of business intelligence company Panorama Software. “One of the big issues has always been adding new users to your application: you need to buy new licences, ensure the users have compatible machines, install the software, then keep it upgraded and maintained.”

In the cloud, that all goes away.

The boundaries of the organisation are now logical, not physical.

Chris Norton, regional director, VMware

The most commonly touted advantage of SaaS is doing away with lump sum payments for software.

“Software becomes an opex cost that's predictable month by month,” says Todd Schoeman, head of MEA portfolio and marketing for BT Global Services. “The service provider takes the risk of investing in the massive shared platforms that are needed, and the customer is able to add and remove capacity flexibly as demand fluctuates.”

So, SaaS is financially attractive and promises far greater ease of management than the alternative of buying your software on CDs the old-fashioned way. But there is a risk: when that software is being delivered over the Internet, the quality and reliability of your bandwidth becomes critical.

“The whole premise of cloud computing is that your services are remote from your actual location,” says MWeb's Hogh. “That means your communications, as the foundation of it all, need to be solid, reliable and fast. It's no good putting stuff in the cloud if the underlying foundation can't offer you performance and reliability that matches what you would have had on site.”

Has South Africa reached the point where the speed and reliability of its bandwidth makes a move to the cloud feasible? The consensus appears to be “almost”.

“The bandwidth issue is being solved,” says RSAWeb's Gilmour. “I'd say we're at the point where anything that isn't mission-critical can and should be run in the cloud.”

But with international bandwidth still reliant on a handful of cables that are subject to outages - as Seacom users discovered to their cost in July - Gilmour says there is an important proviso: the cloud should be local.

“There are some global SaaS vendors that don't offer the option yet, but if I could choose, I would definitely host locally,” says Gilmour. “The link between the server and the desktop should be as short as possible. People used to prefer hosting overseas because of price, but that doesn't hold true anymore. The only reason to host offshore is to serve offshore clients.”

Much also depends on the application you're trying to run from the cloud. “We definitely have enough bandwidth to run applications like salesforce.com,” says Accenture's Thompson. “But if you're going to move lots of digital content, you need to make arrangements with your telco. There's so much being invested in communications infrastructure right now, though, that things are getting better all the time.”

“The more competition we have in the market, the better things are going to get,” adds Yaron Assabi, CEO of the Digital Solutions Group. “What I expect we will begin to see is the ability to pay for quality of service. Right now, we're paying, but without the service. We'll soon see people willing to pay more for guaranteed service levels.”

Until then, customers looking at basing their applications in the cloud need to ensure they have at least one backup connection in place - and preferably more, says Hogh of MWeb Business, which focuses primarily on small to medium-sized businesses. “Between fibre, ADSL, WiMax, iBurst and the various 3G connectivity options, there's a viable failover option for every customer,” he says. “It all depends very much on their individual needs and how long they can afford for their connections to be down. In some businesses, half an hour is fine; in others, that would be catastrophic.”

This ability to rely on service providers to advise on and help design appropriate solutions, without needing to bring in expensive consultants, is one more reason why cloud solutions are proving especially popular among small and mid-sized companies.

“These are the companies that don't need or want to run their own IT infrastructure,” says VMware's Norton. “Large enterprises like banks, on the other hand, don't trust anyone else to run it.”

This is where the distinction between private and public clouds becomes useful. Running a private cloud doesn't necessarily mean building your own data centre: it's entirely possible to build one using someone else's physical infrastructure. What is critical is keeping everything inside your own firewalls, under your own control.

“So long as the resource is within the customer's domain and control, and limited to their use, it's a private cloud,” says Norton. “Physical location is irrelevant. It would be entirely possible, for example, to create a private cloud for the government, linking locations across the country through the Sita backbone.”

To make matters more confusing, it's equally possible to have equipment that's housed on your own premises, but is not part of your own private cloud. This is a solution some organisations have used to get the benefits of cloud computing without the hazards of bandwidth trouble.

“We have a few cases where we have built cloud-like infrastructure at the client's site,” says Richard Firth, CEO of MIP. “We own and manage the equipment and use a cloud billing model, but the customer doesn't have any issues with bandwidth or latency.”

This is one variant of an increasingly common phenomenon, the hybrid cloud: a variable mix of public and private, on-site and remote infrastructure and services. “The cloud” is not an all-or-nothing affair.

There are many different examples of hybrid solutions. One of the most common is the increasing trend for small and medium businesses to give up on owning and managing their own mail servers, opting to have these hosted by a professional provider.

“The peripheral services are the first things to move into the cloud,” says Hogh. “Smaller businesses in particular are keen to outsource things like e-mail, security, virus protection, archiving and so on. It's important, but not core to their business.”

“We have 50 000 Exchange mailboxes hosted in South Africa,” confirms Microsoft's Bester. “It's an ideal solution for the smaller guys who can't afford to retain the best in-house skills and run their own hardware.”

Another hybrid strategy is to “provision for the average and outsource the peaks”, as Norton puts it.

“You only need to have enough infrastructure in-house to cater for your average demand; then you can burst into the cloud during peak times. That whole process can be completely automated so that the business will never even know it's happened.”

Panorama Software's Ross offers a third strategy, of separating applications from data.

“It's great to have an application running in the cloud, but people want to keep data on their own premises for security and speed reasons. Putting just the application layer in the cloud allows people to enjoy the power and cost benefits of cloud computing, while maintaining control over their data.”

Whatever you do, says Schoeman, “don't move without an audit. Most cloud service providers have consultancy practices as well to help clients establish a baseline: what they have, what is under-used, what is over-used, what applications are suitable for the cloud and what a sensible migration path would be.”

Also, says Accenture's Thompson: “Cost savings are not automatic; you need to do the sums. Flexibility is often a stronger motivator than cost.”

Results from Mimecast's May/June 2010 Cloud Adoption Survey show that 51% of respondents are already using cloud computing in some capacity for their IT solutions.

The Mimecast Cloud Adoption Survey was conducted by Loudhouse Research on over 500 business and strategy respondents from the UK and US. All respondents enjoyed some level of business or strategy visibility, and were from companies with a minimum of 50 employees.

Among the cloud users surveyed, 62% employ cloud solutions for e-mail, 52% for security, and 50% for storage. Mimecast calls these applications 'gateways to the cloud', as they are the most popular applications for first-time cloud users.

I firmly believe security can be better in the cloud.

Richard Vester, executive head of hosted services, Vodacom Business

Of the benefits identified by cloud users, reduced pressure on internal resources, improved IT, enhanced end-user experience and lower infrastructure costs are reported by over 70% of respondents. More than 60% have experienced augmented performance, sustainability and continuity after adopting cloud solutions. Smooth integration with legacy systems was noted by 56% of users.

Sixty-six percent of those surveyed are planning to invest in cloud computing in 2011. Twenty-nine percent of respondents plan to incorporate cloud solutions for the first time next year.

Gartner's Global IT Council, consisting of IT leaders from numerous large organisations and Gartner analysts, recently published the following list of rights and responsibilities of cloud service customers, aimed at facilitating successful business relationships between vendors and consumers.

* The right to retain ownership, use and control of one's own data: cloud providers should specify the extent of their access to and control over the customer's data. The original contract or service-level agreement (SLA) should also clearly define the process of data disposition in the event of the provider becoming unable to continue service.
* The right to SLAs that address liabilities, remediation and business outcomes: in the event of service slowdowns or failures, the provider contract must define recovery procedures and timeframes, capacity increase procedures and timeframes, and other forms of rectification.
* The right to notification and choice about changes that affect the consumer's business processes: customers must receive advanced notification of system upgrades, service interruptions and significant changes, and should be allowed some control over when to move over to updated systems.
* The right to understand the technical limitations or requirements of the service upfront: consumers and providers should both be transparent about the technical limitations of their own systems, particularly for complex projects or architectures, in order to avoid costly upgrades mid-contract.
* The right to understand the legal requirements of jurisdictions in which the provider operates: providers may not violate rules for which the customer may be held accountable, eg, when transporting data through a foreign country.
* The right to know what security processes the provider follows: customers should understand their provider's security and business continuity plans and the legal compliance required from them, so as to avoid inadvertent violations.
* The responsibility to understand and adhere to software licence agreements: both the customer and provider must take responsibility for their own licensing compliance, and neither must be held accountable for violations committed by the other.
* Compiled by Tamsin Cracknell