What’s another sign to me that cloud services are going mainstream? I was watching an episode of 30 Rock recently and it had a commercial from Microsoft with home users talking about going “to the cloud.” As more businesses and consumers use cloud services, the networks to access the cloud and the processing, storage and virtualization resources within it invariably could get bogged down. So performance will become a critical issue.
Think of this way – I live in Atlanta, a city with notoriously bad commuter traffic. If I took a NASCAR car out on interstate 85 at midnight, I could freely race up the road with probably no one around me. However at 8 a.m., even though the car has 750 horsepower and capacity to go hundreds of miles an hour, I’d be crawling in bumper to bumper traffic just like the Toyota Corolla next to me with all that automotive technology going to waste.
So, what are the high-level considerations for cloud performance?
In a public cloud, by definition, many users or tenants share access to the cloud’s available set of resources. The appeal is obvious—the ability to easily consume and configure infrastructure on demand with a usage-based pricing model. But for businesses, this best effort approach comes with questions about performance guarantees and the ability to monitor, measure and assess performance.
Another option is to dedicate capacity and allow users to specify and guarantee capacity to meet their requirements in a ‘private’ cloud that they can tune to their needs. So to extend my NASCAR example, a private cloud would provide a separate lane on the highway just for me to speed past all the traffic and put that high-performance engine to good use!
Clearly whether a user decides between a public or private cloud, will be a delicate balance of understanding a user’s cloud services needs, specific applications and performance expectations. Service level agreements will play a huge role here and is an area I’ll be watching in 2011.