We are rushing headlong into the cloud, and for the right reasons. We can see the future and are enthusiastically building it into our IT strategy – it’s all starting to make sense.
We’re able to design and operate complex global platforms with little up-front cost and pay for them only as we consume them. We can integrate platforms with our legacy physical environments and add in SaaS solutions as we like – we have power, flexibility and the commercial models that we only could have dreamt of 5 years ago. However, there are still pitfalls – some very simple and obvious – that we need to be wary of: easy to avoid if planned for; painful if you’re caught out.
Imagine what would happen if your cloud provider were to go bust. At best, as happened with 2e2 in early 2013, assets and customers would be bought by a competitor that would continue to take your money to run your service. In the case of the demise of Nirvanix Inc in October 2013, customers were notified on 16th September that they had to remove all data by the end of the month – which is a tall order when you have 10s of terabytes of storage.
At worst, your platform would cease to operate and your data disappear. Even if the data is backed up, unless it is to a separate, unaffected provider, without painting too bleak a picture, you could lose everything. The best mitigation for this of course would be to build a DR (or tertiary replica) environment in a different vendor’s cloud, but that’s an expensive option. Make sure that this eventuality is covered in your contract, so you’ve the possibility to retrieve your data if this worst case scenario were to occur.
Security & compliance
If you sign up with a cloud provider, do you know if that provider has the compliance framework in place to satisfy the regulatory requirements of your industry and marketplace? For instance, where will your data be stored? Will the cloud environment you’re using be in a site that’s sufficiently secure? 99% of the time, the solution you need can be found in the cloud, but don’t assume – if there are functions, certifications, compliances that are important to you, ask your potential provider for unambiguous statements and evidence.
Managing cloud sprawl
It is easy to keep building platforms and pumping data into the cloud – and unless you’re careful and delete obsolete or unused systems and data, they just keep on building up. Very often, a business will order and pay for cloud capacity from a number of sources; after all you just need a credit card number and this can lead to uncontrolled or excessive spending. And what would happen if one of the people paying for cloud services leaves your business – potentially no-one would know what’s deployed in their name, or why….and quite possibly they’ll be too scared to delete it/ stop paying the bills in case something important stops working! Not always easy, but in order to use the cloud efficiently and realise the promised savings, cloud usage needs to be properly managed and tightly governed according to a strategy.
Although clever tools are evolving that promise to make things more flexible and convenient, there are still no common standards for cloud providers in terms of data and machine formats. This makes it difficult to change providers or even to get data back once it’s in certain providers’ clouds. Of course this is convenient, beneficial and lucrative to the cloud provider – less so for the consumer. As part of any cloud provider selection process, I recommend planning the eventual exit from that provider.
It is difficult to compare the performance of an external cloud-based infrastructure with one that you are hosting yourself. There are significant variables – the network connectivity and latency, the actual speed of the vCPU, vRAM and shared disk on the cloud platform vs that which is currently used. When transitioning applications to the cloud, the important thing from a user perspective is that performance should be better than it is now. It’s difficult to guarantee application performance in the new cloud environment before go-live. Although at the very least you should have detailed metrics showing current performance of OS and apps that can be used to prove performance increase or degradation. With the correct metrics, you at least have the evidence to identify if performance is poor, pointers to possible areas of investigation and potential improvements.
Key questions when transitioning an existing application stack into the cloud are of course ‘will it work, will it perform and how much infrastructure capacity will it need?’ Also don’t forget the thorny ‘will the vendor support it?’ – and ‘will the licenses I have bought be transferrable to a cloud model?’ You may need to retain non-cloud infrastructure just to deal with compliance with software vendor licensing issues.
The way forward
The cloud represents a shift in the way businesses consume and pay for IT services. Although the vast majority of this is positive, there are some new rules. Taking advantage of the expertise and experience of a cloud integrator will smooth and accelerate the path to full realisation of cloud’s benefits – and ensure you don’t trip up on these simple gotchas.