What is the difference between a public cloud and an enterprise data center?

If companies want to get the agility of public cloud providers like Amazon, Facebook and Google, they have to look at their IT architecture and operational management from beginning to end.

In a speech at the Gartner Data Center conference, vice president Ray Paquet summed up several key differences between cloud computing providers and corporate it, as well as how companies can bring themselves closer to the cloud.

"By 2017, the main public cloud architecture will become a common architecture in the enterprise." Paquet said, but cloud computing "is based on the different from the current enterprise." ”

Public cloud is not like enterprise it, is designed according to scale level. They take a shared-nothing architecture and use asynchronous communication. Applications can tolerate hardware errors, such as software layers that provide protection against independent hardware failures, and absolutely everything can be monitored. When it comes to cost, the cloud service operator is "looking for the cheapest hardware, not the best, and not the overhead."

"They focus on scale, smart software, and cost, and the result is a level of agility." ”

He also contrasts the public cloud and the way the enterprise works with a variety of data center originals, including servers, storage, networks, virtualization, data center facilities, and applications.

Server platform

The primary server platform for the public cloud is not a mainframe, high-end UNIX box, or a pre-fusion architecture platform. It's a low-cost 1U-free server or a server with a pizza box size purchased on a machine rack. The public cloud service provider does not buy the pre Fusion Blade server.

"They bought the model last year because the cost is already available, and inexpensive." ”

Storage

When it comes to storage architectures, public cloud providers do not buy products that appear to be pillars in the Enterprise data center. The primary storage architecture is direct-link storage (DAS), plus a distributed or object-based file system that provides a global namespace.

Public cloud providers generally do not use sans or other common arrays, Paquet, perhaps the most shocking thing for enterprise storage managers is that public cloud providers do not use RAID. This reduces storage costs by 25%. "But they replicate data from one site to two sites." "This is responsible for usability.

When it comes to suppliers, Paquet says you don't see the Tier 1 storage vendor in the public cloud.

"These major suppliers have not cracked the hyper-scale market, preferring to place x86 servers wherever possible." ”

Application

The conceptual and architectural applications of all public cloud providers are profoundly different.

"The real change in cloud computing is the application layer, which can make us clouds and become more agile." "Paquet said.

Cloud applications are based on differences from traditional enterprise workloads. They extend outward, do not share, have no nationality restrictions, and communicate asynchronously. For improved performance, cloud designers can use application tiering and application storage without using SSD or flash memory to improve performance.

But the idea is to use resources wisely and not to overdo the environment.

Continuous delivery

Finally, companies and cloud providers have different speeds in delivering and upgrading applications.

For developers, the goal is "faster," Paquet said: "But in traditional management we need to slow things down because we know that if no one brings change, things might keep running." ”

Public cloud service providers adopt DevOps as the basic contradiction of the two sides of repairing development and management. DevOps has connectivity tools in practice that make it possible to accelerate changes with the extended use of automation, monitoring, and configuration management.

(Responsible editor: The good of the Legacy)