– Carl Meadows, senior product manager of Cloud Services at The Planet (www.theplanet.com), says:
Enterprises looking for high-performance storage should look elsewhere: The cloud isn’t it. In order to achieve high-performance, the storage platform must be well connected to the compute resources accessing it. With the Internet acting as a bottleneck between the compute resource and the storage, high performance becomes impossible due to both bandwidth and latency constraints.
The only exception to this is if your compute resources are well connected to the cloud storage infrastructure. For example, The Planet’s Storage Cloud platform is located in our Dallas data center, which customers can access without going through the Internet. The platform is only limited by the 100Mbps or 1Gbps uplink of their hosted server.
Assuming bandwidth is not an issue, there’s still the issue of latency. Most enterprise-class storage platforms are targeting 1 to 5ms latency, while most cloud storage platforms would do well to deliver 30ms, which is still largely dependent on your proximity to the cloud storage provider.
In general, the best solution to address payloads that truly require high-performance storage is still placing a DAS, SAN or NAS storage array next to your server(s). In these instances, compute capacity is directly connected, which yields more reliable results.
Will Cloud Storage 2.0 eliminate storage silos with a single, multipurpose platform? The multipurpose utility of the cloud hasn’t yet been proven. Although cloud technologies scale seamlessly, they do not currently offer choice of file systems or performance profiles that would be required for application optimization. In addition, because most cloud storage is accessed over the Internet, performance is limited.
Cloud Storage 2.0 will offer enterprises access to scalable storage without the expense of a large-scale dedicated platform. However, all these benefits are currently available to SMB and SME customers with Cloud Storage 1.0.