Agreed with the points Martin had commented, think other differences between a cloud data server / VM and traditional file data server include:
1) Cloud data server can support rapid elasticity / flexing i.e. it can accommodate high workload access during peak period by rapid provisioning of more resources & vice-versa compare to traditional file data server which is subscribed with certain resources (i.e. CPU, RAM and HDD) agreed in advance.
2) Cloud data server usage can be billed / charge-back / show-back based on actual resource usage e.g. CPU, RAM, HDD of the specific cloud data server compare to traditional file data server which is based on $$$ per server per month agreed in advance (of course there can be ladder pricing i.e. the more months you subscribe, lower will be the unit price).
3) Cloud data server can support bursting i.e. when the availability of pooled resources of a cloud is finished / used up, one can burst its consumption of other resources e.g. CPU, RAM, HDD, VM from other external clouds. This is generally not available from traditional file data server.
4) Provisioning / de-provisioning of a cloud data server / change request can be very fast due to extensive orchestration & automation are used compare to traditional file data server which will take longer time due to manual effort and more human error in provisioning / de-provisioning.
5) Cloud data server can be hosted at any DC in different geography if you didn't specify during initial registration whereas traditional file data server usually hosted at a specific DC location known in advance.
Agreed with the points Martin had commented, think other differences between a cloud data server / VM and traditional file data server include:
1) Cloud data server can support rapid elasticity / flexing i.e. it can accommodate high workload access during peak period by rapid provisioning of more resources & vice-versa compare to traditional file data server which is subscribed with certain resources (i.e. CPU, RAM and HDD) agreed in advance.
2) Cloud data server usage can be billed / charge-back / show-back based on actual resource usage e.g. CPU, RAM, HDD of the specific cloud data server compare to traditional file data server which is based on $$$ per server per month agreed in advance (of course there can be ladder pricing i.e. the more months you subscribe, lower will be the unit price).
3) Cloud data server can support bursting i.e. when the availability of pooled resources of a cloud is finished / used up, one can burst its consumption of other resources e.g. CPU, RAM, HDD, VM from other external clouds. This is generally not available from traditional file data server.
4) Provisioning / de-provisioning of a cloud data server / change request can be very fast due to extensive orchestration & automation are used compare to traditional file data server which will take longer time due to manual effort and more human error in provisioning / de-provisioning.
5) Cloud data server can be hosted at any DC in different geography if you didn't specify during initial registration whereas traditional file data server usually hosted at a specific DC location known in advance.
No, there is no difference. Why? Because all the properties claimed in previous responses to be specific to the cloud can be implemented by an FTP interface. FTP is, after all, just a protocol - it's absurd to claim it can't be traced or billed, or can't utilize elastic infrastructure, or be multi-DC. After all, none of those properties somehow comes automatically (cloud simply provides infrastructure for these properties...). One could certainly implement an elaborate FTP interface using cloud building blocks such as those provided by Amazon.