Getting the Most Out of Your Cloud

Article by Gopan Joshi



For most enterprises, getting on the cloud is seen as the key to greater data efficiency and improved operational performance.

But the fact is that the cloud does not magically deliver all these benefits on its own. After all, the cloud is built using largely the same underlying technologies available to most datacenters. Rather, it is the way cloud environments are built, provisioned and managed, that delivers the improved flexibility and performance that most organizations strive for. In that regard, then, there are a number of key elements to consider when building a cloud infrastructure intended to meet the performance metrics of increasingly complex data environments.

One of the most fundamental elements is a robust service orchestration platform. Most cloud setups are designed to run multiple services, but without robust orchestration, they continuously vie for available resources, leading to performance lags and resource over-provisioning. An effective orchestration platform allows you to de-provision as well as provision, plus provide upgrade/downgrade capability of virtual compute, storage and network resources.

All cloud providers should be able to provide a detailed Service Level Agreement as well. This should spell out exactly what level of performance is to be maintained and how it is to be measured. Cloud providers should be able to guarantee virtual machine uptime of 99.95 percent, particularly for mission-critical applications. And it never hurts to go over the provider's support policies as well. Do they offer live support, 24x7? The last thing you need in a service disruption is a support staff that only works during the day.

In many ways, however, performance is improved by deploying a hybrid cloud infrastructure, which combines the resource capabilities of a public service with cloud-based elements of existing private architectures. In this way, enterprises can scale up resources as needed while maintaining the same management and policy regimes that have been in practice at the user end.

For example, a hybrid cloud built on a virtual private data environment within a public cloud service allows you to create private, dedicated instances that have the same network topology and policy settings as existing infrastructure. This makes it easier to create both web-facing transactional environments and mission-critical back-end services on a single, integrated platform without allowing sensitive data to cross from one to the other.

Of course, a private cloud does not have to reside within the enterprise's internal infrastructure. A private cloud on a hosted platform offers a number of performance advantages, but, again, only if it is properly configured. Naturally, a hosted private cloud must provide single-tenant, dedicated infrastructure, but it should also provide full control over those resources without burdening the user with maintenance and trouble-shooting issues. It should also provide full resource usage tracking and granular visibility into performance patterns and other parameters to ensure optimal resource allocation. And most importantly, it should provide solid security through physical and network separation, with full support for security and compliance standards like PCI - DSS and HIPAA.

The fact is, cloud computing is not a one-size-fits-all proposition. Basic cloud services can be had with a phone call and a simple contract. But in order to derive the highest performance from both the cloud and existing infrastructure, enterprises need to take a hands-on approach, beginning with a clear-cut understanding of exactly what they hope to gain from the cloud and how it can best be integrated into an overall, high-performance data environment, taking into confidence the cloud service provider not as a vendor, but as a partner to this setup.