genericcloud
Cloud Computing

Paddy Falls (US) - Am I Available in the Cloud?

The cloud continues to be a hot topic in the world of IT, and yet potential cloud customers who remain wary of the pitfalls are often met by stories of infrastructure failing, giving them reason for apprehension. Most recently, cloud platforms; from Amazon, Google and Microsoft, have all experienced downtime, leaving customers without access to business critical applications such as e-mail. With stories like this frequenting the press, it is unsurprising that concerns around ‘the cloud' still linger, and yet lessons are very quickly being learned in order to make cloud offerings more robust.

At Neverfail, one of the biggest questions from customers is how do they take advantage of the lower costs models of using the cloud, without risking key parts of my business on a cloud provider that I cannot guarantee will be available.

One answer is to host business critical applications in-house, and host the Disaster Recovery (DR) aspect remotely with a cloud provider. This gives a great mix of high availability in the form of in-house primary infrastructure, as well as hosted DR which is more cost effective than using a second data center with matching physical architecture. One word of warning is that it is important that the cloud provider chosen is not local so that local disasters, such as the recent power failures in southern California, do not bring down both your local data center and your hosted DR.

The primary advantage of any cloud solution is to drive down costs. The initial cloud DR cost saving comes from converting capital equipment costs to monthly operational expenses. The next fundamental cloud saving is taking advantage of the economies of scale of sharing your hardware, buildings, energy and people costs with all the other customers of the cloud provider. Cloud DR is a good example of an ‘elastic cloud application' which means you can pay for just the machine resources you use on an intermittent basis. With most cloud providers you should only pay the full cost of switching to applications on the DR site during those days in the year when you have a disaster.

Having setup a cloud DR site, you will find that it has additional uses. First you can use it for planned maintenance. So if you needed to upgrade your data center networks, you can switch to the DR site. Another common use of a DR site is to test your applications. For example when you need to upgrade them, and don't want to test on the live system. These are further examples of elastic cloud computing and simply working from a pay-as-you-go model. While hosting your DR site in the cloud will be accompanied by a certain level of risk if that cloud hosting provider fails, this risk is much less than having your live applications running in the cloud. In practice, the cloud provider is likely to have higher availability than at an internal DR site, but at a fraction of the cost. So cloud DR is a win-win from a cost and availability perspective.

Another fundamental concern about cloud deployments is how secure your data is. While security remains a major issue for many potential cloud customers today, we don't see this as being a long term concern. The security technologies exist to secure data across the public networks and on hosted environments. The issue is trust, which will need to be earned. A good example is Saleforce.com, which customers have grown to trust to securely hold their sales information.

Once organisations see the cost advantage of hosting their DR in the cloud, they will want to look at further reducing cost by hosting their primary application in the clouds. This raises a number of issues. First how do I protect against the cloud provider failing? The best solution is not to put all your cloud eggs in one basket, and choose a different cloud provider to host the DR function - that way if the primary cloud provider experiences failure, applications will remain protected. In this primary and secondary cloud solution, it's possible to minimize costs on the second cloud because it doesn't need the same level of availability or performance as the primary cloud.

Although the cloud has been lauded as the next ‘big thing' in IT, the reality is that, in terms of applications, moving the application to the cloud does not mean that the application will be any more reliable. Analyst surveys have been reporting for some time that the main risk to application failure is no longer hardware, but are instead application related risks. So whether you run your applications in house or on clouds, it's important to have application level protection that will detect failure in the applications and take remedial action such as restarting the application, or switching it to another machine where it can have increased memory, processing or disk IO capacity.

By guest blogger Paddy Falls, Chief Technology Officer, Neverfail

PREVIOUS ARTICLE

« Bruno Teuber (UK) - Getting Under the Skin of the Social Customer part I

NEXT ARTICLE

Calum M. MacLeod (Europe) - One Bad Apple »