Cloud computing may have been hyped but it can offer real benefits provided you address certain issues up front, says Clive Longbottom.
Cloud computing has been hailed as the cure for all IT ills. Fortunately, a measure of reality is now returning to people's view of the technology and the business value it can provide.
Nevertheless, some organisations are still not thinking through the issues of cloud computing, which may cause them problems down the line. If these issues aren't accommodated for, the cloud can become a barrel over which providers can stretch client organisations.
So, what are those issues and how can we minimise their impact?
1. Data ownership
The first issue is making sure you own the data and that this ownership is built into the agreement with the provider.
The application, the hardware, the operating system and everything else can be owned by the cloud provider. But the data is what your intellectual property is predicated upon and it has to be acknowledged that you can take that data away with you as you see fit.
2. Data access
Data ownership does not amount to much if you are denied access to that data to migrate it away from the cloud provider.
Your cloud subscription gives you access to the functionality of the application or function that you use. If that access is removed, can you still access the data so that you can take it away with you?
Make sure the contract allows for access to the back-end data, either directly or via the provider offering an export capability, even after the contract has finished.
3. Data volumes
Cloud is great for off-site elastic computing, where extra resources can be applied in the form of more compute power, or more storage. However, as that storage capability grows, so does a specific problem.
Migrating 1GB of data across a wide-area network is pretty simple but how about 1TB? That migration can take a long time, and if you need to work against that data in real-time, you'll have to plan for a degree of downtime while the data is pulled from the cloud and reinstalled against a replacement application or function.
Even if you can agree to set up a mirrored cloud or on-premise data topology, look out for clauses in the agreement that charge for data volumes.
Also, look at data cleansing and deduplication options, which can minimise overall volumes. If left unmanaged, mirroring data can rapidly become a major cost if data transfers are being charged for.
4. Data usability
Most cloud-based systems are built on a standardised database but that does not mean you can, for example, take a copy of the database from Salesforce.com and use it on a NetSuite system.
For an on-premise system, you have to understand the database architecture. For a cloud-based system, that understanding is just as important. Data has to be...
Clive Longbottom is the founder of user-facing analyst house Quocirca. As an industry analyst, his primary coverage area is business process facilitation.