How do you get a true picture of performance in cloud environments?

Cloud computing is latest rage these days and while there are varying opinions as to the rate of adoption as more businesses turn to the cloud, it is definitely a model that is here to stay. Like any new computing model cloud computing comes with its own set of challenges. You’ve got to be able to adequately plan for a migration (private cloud). For the private cloud, there is an infrastructure investment in terms of additional hardware and virtualization software and for those considering the public cloud there is the great unknown of accountability, measurability and control.

How does one get a true picture of performance in cloud environments?

Let’s face it, private clouds have lots of moving parts; between the server hardware, network devices, virtualization software, and the application software itself you’ve got quite a challenge. Invariably migration to a private cloud involves some sort of physical to virtual migration, so you’ve got to be able measure before/after capacity and KPIs. Once a private cloud is in production and you’ve sunk all the time effort and money into the environment, it better be faster/better and more efficient. Performance monitoring is key to 1) proving out any success criteria that you’ve used to justify a private cloud initiative and 2) ensuring continued success.

Public clouds present their own challenges. Some would argue that public cloud applications are usually “less critical” than those deployed in the private cloud. Well I’m seeing lots of email and storage on the public cloud. Since when is email not that critical? Lots of vendors are vying for public cloud business and making all kinds of promises about reliability and availability, the SLA acronym is bantered about quite a bit, however definition and expectations vary. Public cloud providers have to be held accountable and evaluated. More often than not you’re going to be the one that’ll want to measure performance availability independent of the provider. After all it is in the provider’s interests to tell you that you’re getting optimal performance and availability; whether their metrics are unbiased can be a bone of contention.

Recently announced Longitude Cloud Edition provides a mechanism to measure and quantify performance and availability of cloud environments both public and private.

For private clouds Longitude helps with both in pre/post virtualization deployments by providing comprehensive reporting and alerting capabilities, while leveraging a built in knowledge base. Once a private cloud is fully deployed Longitude helps keep the environment functioning by identifying potential bottlenecks and impending operational failures.

For public clouds Longitude provides a mechanism to measure, quantify, and report on end user experience and availability. It isn’t enough for the provider to notify you of availability and performance, you’ve got to be able to measure this independently. Those being serviced by public clouds need to keep their cloud providers honest.
I invite people to comment on my blog.

• What kind of challenges are you facing with public cloud providers, especially when it comes to measuring availability and performance? Are you able to measure what was promised to you?
• For those engaged in private cloud initiatives, what are you using as success criteria when you look for funding? How do you plan to prove the initiative was worth the investment and how do you ensure continued success??