The information sought from each provider we looked at, with an explanation, follows.
The name under which the company is trading.
2. Infrastructure-as-a-service (IaaS) offering name
The name that the provider is giving its IaaS product.
3. Online sign-up
Can businesses sign up for the cloud service online in a self-service fashion?
The ability to sign up and transact, without interacting with a salesperson or an account manager, is viewed by many as a critical element of cloud computing. Technical people have generally researched what they want, so they just want to get on with the job. They prefer less manual interactions, and more means to automate activities. This can be in stark contrast, though, with less-technical management or management of enterprise, which may want to engage in a more traditional sales process.
4. Location of datacentre
In some circumstances, even though you have a contract with an Australian entity, that entity could be using an offshore facility. Some customers have mandatory requirements to keep data in Australia. Communication latency for certain types of application scenarios can also be too high with offshore cloud providers. Using Secure Shell to access an Amazon EC2 instance, for example, can be very frustrating when latency is poor. However, this is not true in other circumstances or application scenarios.
5. Contract jurisdiction
In the event of a legal dispute, which laws apply to the data? Which courts will rule or arbitrate?
Having a contract with an Australian organisation hosting data in a datacentre located in Australia makes it likely that Australian law would be applied to the data. Scott Stewart, Longhaus' cloud-computing analyst, said that Australian organisations can still be subjected to the US Patriot Act if they have operations in the United States, due to treaties signed between the US and Australia. It would have to be exceptional circumstances, however, for such powers to be invoked.
6. Pay by credit card
Can you pay for the service with a credit card? Doing so provides a confirmed identity to the provider, and also shortens the time to bring a new customer on board.
7. API for external automation
Cloud computing is about on-demand access, with minimal management effort or service-provider interaction. Without an API, manual intervention is always required. A good API allows for tasks to be automated, and for metrics about usage to be retrieved. It also opens the door for innovative uses of the cloud provider's services. An example would be retrieving spot prices for virtual machine instances. Based on the pricing and availability, a determination could be made to start a job with that cloud provider.
8. Minimum commitment
What is the shortest amount of time that you can use the service for? A short commitment provides customers with flexibility. It is not uncommon to see it as 1 hour for compute if you have already set up an account with the cloud provider. If you only use an hour at a time, then that is all that you will be charged for on the next billing run. However, if you're asking the cloud provider for dedicated infrastructure in a private cloud configuration, the minimum commitment will be longer. Some providers will use the enterprise tag to bump up the minimum commitment.
9. Compute cost (CPU per hour)
How much will a CPU cost per hour? Depending on your transactional needs, different CPU configuration and memory combinations are normally available. If you have a low-usage website, you can ask for an instance type that is appropriate. The cost will increase with more compute capacity, with the higher costs for more cores or CPUs and memory.
10. Storage cost (GB per month)
How much does it cost to store 1 gigabyte per month? Rates for storage will vary, with some cloud providers having an initial allocation that can be extended. The more you store, the cheaper per GB it can be.
11. Data transfer (cost per GB)
Many providers don't charge you for internet traffic coming in, but will charge for internet traffic going out. There may be a rate table applicable for the provider. Internal traffic within a datacentre is normally not charged. However, if a cloud provider has multiple regions, then a charge to transfer between the regions may apply.
If there is a significant amount of data, it may not be viable or cost effective to transfer it over the internet. In that case, a service may be available where the data from posted hard drives can be loaded into the cloud provider's storage.
12. Load balancing
If a new server is provisioned, it also needs to be added into the load balancer, and, if de-provisioned, removed from the load balancer. The load balancer's job is to distribute the workload between available resources.
How many customers does the provider have, and how big are they? This can be a marker for reliability, but it may also affect bandwidth and latency.
When you're relying on someone else to keep your data safe for you, security is paramount. Unfortunately, it's very hard to compare security in a table, but we've taken a look at what security measures are possible for cloud providers and what customers should look for.