A cloud on the horizon

Clouds on the horizon Tim Knudsen, Director of Product Specialists for Akamai's Application Performance Solutions recently chatted to Justin Richards about all things cloud. 

Do you think the use of cloud computing is essential for the growth of many businesses particularly during this current economic downturn?

Yes, I do. I think it's going to be critical for the enablement of small companies. There are a growing number of SaaS providers out there who actually have savvy solutions but need the flexibility of not having to build multiple data centres or even a single data centre in order to establish their service. The value of their service is not the hardware upon which it sits, it's the capabilities it provides, the business processes.

So having rentable infrastructure, rentable platforms and having delivery mechanisms that also complete the cloud, between the web infrastructure and the users, is vital because it allows them to grow without having to make massive investments up front.

For existing companies it's also a benefit because as companies continue to grow their own application sets, they're trying to reduce the cost of business and increase reach. They're able to leverage cloud infrastructure to offload applications and infrastructure that otherwise they would have had to have made upfront investments for.

In today's economic environment there's general hesitation towards larger scale capital investments to enable the business. So if an IT department can launch a capability where they do not have to spend a lot of money upfront then it's more digestible to the company.

Tim Knudsen How much, in general terms, does it cost to use cloud computing as opposed to investing in one's own infrastructure? What's the degree of cost saving?

To build one data centre within, say, the Asian-Pacific region would cost between $500,000 and $1.2 million, depending on size, which is an entrepreneurial upfront investment. Furthermore the ongoing annual maintenance costs can be anywhere from $200,000 to $500,000. Of course I could find the struggling SaaS provider, who'd quote numbers of half that, but that's a small footprint compared with some of the more established players.

The key point is that it's a non trivial matter of investment in order for companies to establish a presence internationally outside of their current data centres. Even though you may have a data centre in a region, we've found, through our research, that performance problems don't go away just because you have a data centre presence in a region.

For example, we actually provide our acceleration solutions services to a number of SaaS providers whose end users are just within North America. So you'd think that, logically, if the end user region data centre solution was viable then we would have no market within North America.

However, we do have a market, in terms of ensuring performance and reliability and in scaling infrastructure based on what we can do in offloading, whether it be requests or connections, through intelligence that we've built into our platform. The win-win for organisations, big and small, from a cloud perspective, is they can find the on-demand infrastructure as well as the on demand acceleration by leveraging the complete portfolio of cloud offerings.

How would you define cloud computing and how would you define virtualisation, from your perspective?

From my perspective I've chosen to focus more on existing definitions that have already been published rather than explain what exists and where the gaps are. The version I think works for most people is the one promoted by Gartner, which covers many of the core areas that are now known as the building blocks of the cloud computing space.

They divide it into three areas: infrastructure as the service, which are the rentable bare-metal resources, storage, computational capacity; platforms as a service, which are actual platforms which you can push your code to and run there and then the last area which is software as a service which include, for example the various CRM solutions, the sales forces of the world, where you don't store any software on the premises, you basically pay a subscription to those capabilities. So those are the three stacks or components to the cloud as I see them.

Regardless of whether companies are leveraging one or all three of those elements, the delivery component and completing it with the acceleration is critical to ensure performance for the end users - availability and reliability. As a reference point, if you look at all the different SaaS providers for leveraging our capabilities today, like the bullhorns, the autodesks; all these solutions are leveraging our acceleration platform so they can solve this problem.

If the performance of your SaaS offering is not fast and high you're not going to be able to acquire those customers, and even if you do acquire those customers, if you don't have fast and reliable performance the users won't adopt the solution and if users are not adopting the solution then the company's not getting that return on their investment, or even the productivity or the cost savings, or whatever it was that drove them towards that investment in the first place.

Solving the problems in the cloud, between whatever that structure may be and the end-users, is critical across that particular stack.

What would be your top tips to a company thinking of going down the cloud computing route?

The first point would be don't forget the user. No matter how slick your application may be, without ensuring the user has the experience and reliability that’s necessary to have a successful user application, it's all for nothing.

The second point builds on that somewhat - and that is to right size your investment in cloud computing. Which means put the things in the cloud that you think it makes sense to, based on what your needs are, who you're trying to get to and what you're trying to achieve from a business standpoint.

For example, some people may still find they can offload everything to a reputable infrastructure in a cloud but, in other cases, they may find, given the nature of the application that, for example, there's a lot of replication and compliance requirements.

For instance, it may make sense that that application stays within their data centre but they ensure performance through delivery. Or what they do is they host the logical tiers in their data centre and they offload storage to a core provider. It's not an 'all or nothing' type of strategy.

What are the main obstacles for people making the best use of the cloud?

There's a lot of noise in the system right now. A while back XML was the over-hyped new technology on the scene - the number of XML articles that were published was amazing in volume, and it wasn't until people had educated themselves on the capabilities and limitations of the technology that they understood how it could best be applied in their context. It didn't really start to achieve a really high degree of adoption. I call that the education cycle.

I think we're in an education cycle right now. And as people start to realise what it is that they can and cannot do, what the limitations are, and the benefits out of the back end of that, the by-products will be where they’ll start to apply and spend.

Do you think more people would make use of the cloud if they understood in clearer terms what the benefits were?

Yes definitely.

What do you think of the security of the cloud at the moment?

I definitely think that security is going to be a core requirement and a topic of concern. The way I categorise security is a factor which either accelerates or limits the total size of applications put into the cloud, but by itself is not going to stop it happening at all. If security was such a show-stopper right now, for the cloud initiative as a whole, you wouldn't see the massive proliferation of fast services today, which include things that support CRM and self source automation.

The interesting thing is that you can track this. As you talk to an enterprise, ask them if they're considering putting any of your enterprise applications in the cloud and at the same time ask them if they're using any SaaS providers or offerings today which include CRM or self-source automation, or compensation management, or the like. And I guarantee this dichotomy - the business will have accepted that risk while the IT side will be reluctant - the question I have is 'Why is that?'

Security will continue to be a topic, but I don't see how it will actually stop the cloud trend. It will definitely either slow or accelerate it depending on what the enterprises' positions are, but there's already enough adoption of services over the public internet to say that the security applications, which already exist, have already engendered a degree of security and of trust on behalf of those enterprises using the system, to not stop them from using them.

In which direction do you see cloud computing going, where are we going to be in say three years' time?

I think that's a hard question to answer right now. I think what we will see in three years' time is the infrastructure and services will be more readily available. I think the core stacks I've already mentioned are part of cloud today are definitely going to exist three years from today. With these there will be a stronger emphasis on the acceleration and delivery components which will be required for that.

People are going to realise that the cloud strategy is really an online data centre strategy by design. If performance and scale weren’t important then we wouldn't be doing the level of business that we're doing right now.

Can you provide a brief potted history of yourself and your activities?

I came to Akami through an acquisition of a company called Netley Technologies, but prior to that I was Director of Product Management for a company called TeaLeaf Technology.

With the acquisition of Netley we marriaged the technology assets of Netley, which was purpose built for non-cachable applications, with the global footprint that Akami had in the 'cloud'. When it came to an infrastructure platform, we could leverage the Akami footprint, the Akami technology database, plus the Netley technology assets together to create a very compelling acceleration solution and service set for applications which are both web-based and non web-based as well.

That's important to point out, not only for this story but for the importance of cloud because the enterprise web based applications get the most air-time from the perspective of people thinking about the problem, but the reality is when we engage with a company, whether it's in high tech or oil and gas, or communications or even in the retail space, they're also looking at applications that are non web based.

They could be real time collaborative applications or any other sort of application but generally they will be trying to reach a wide spectrum of users from a central location and they need to do so while ensuring fast performance, high reliability and high availability as well. So that's really the problem that we're focused on from a solution set but if you add the cloud discussion in there it adds a new dimension to everything.

Right now there's obviously a lot of attention on the cloud, and rightly so; from a purely IT strategy stand point it makes a lot of sense. You get agility, web ability, to be able to access your rental resources at will; you do not have to maintain your infrastructure.

The expertise of a company is not about maintaining its infrastructure; it's more about intellectual property and applications that drive their business. Although if you can offload that, to a rentable resource, whether it's the infrastructure itself, the server, the application platform or, in some cases, the application itself, in the case of staff, it makes a lot of sense from an IT strategy standpoint, in today's macro economic climate, and from a business standpoint as well.

The point that gets a bit lost in all the discussion right now is that clearly just putting things in a cloud is great, but just putting something there, whether it's SaaS, platform or infrastructure doesn't mean you're solving the cloud problem of distance induced latency and that's really what we're focused on in terms of our solutions set, when it comes to the performance and delivery component, when it comes to these enterprise applications.

Final points about cloud computing and virtualisation in general...

Over the next 12 to 18 months there's going to be stabilisation on what the cloud's going to mean for different sectors and industries, in terms of what they will leverage and what they won't need to leverage. I think we're entering a period of settlement and with that will see a set of emerging trends in terms of what are the best investments to make in the cloud.

I think whether it's the cloud in general, or virtualisation particularly, one of the leading things people are going to continue to invest in is ensuring the success that the end user is part of. Outside of that you have lots of bits and pieces that aren't really serving a purpose.

So the key takeaway is that you've got to think holistically when you think about your cloud strategy and you can't forget the user. So in order to complete the cloud you've got to leverage acceleration and delivery technologies.

Read more from Tim Knudsen

April 2009