Let me tell you how to avoid cloud service interruption

The disruption of cloud services is like an earthquake, a tornado disaster, so we can't find a way to prevent him from happening. Although there is no way to avoid, but we can take effective measures to prevent, as far as possible to protect themselves. April, Amazon's infamous cloud service outage has disrupted some popular websites, including Foursquare and Reddit. Many of Amazon's customers design their own systems with flexibility, use more than one available area, warm up in a traditional data center, or have a backup in the cloud provider settings and prepare to start running in the event of a failure. the was safely through the storm without a fault.

SmugMug, a photo-sharing company in Silicon Valley, is not using Amazon's resilient block storage when Amazon is facing disruption. so that the connection is still guaranteed during this outage while its competitor is disconnected from the network. But SmugMug also scattered its systems in several data centers called "usable areas" by Amazon.

Chris MacAskill, chief executive of SmugMug, said other companies would also be able to stay connected if their applications were dispersed. MacAskill also recommends that businesses use multiple Amazon areas, which are more isolated than available areas. Of course, Amazon has to charge an extra fee for using multiple areas. Therefore, these needs need to be considered.

SmugMug relies heavily on Amazon to store users ' photos and videos using Amazon's cloud-based S3 (Simple storage service) service. SmugMug also uses many examples of EC2 (elastic compute clouds). But instead of using Amazon's elastic block storage, SmugMug still uses traditional data centers. Amazon Resilient block storage is attached to a single EC2 instance and is typically used to store operational data.

That has its own weakness, MacAskill said. For example, in the week of the Amazon outage, the company lost a core router, its backup and a core primary database server. I hope I don't have to deal with routers or database hardware problems again. This is why we are still moving towards the cloud.

He said that despite the disruption, the cloud-based services he got from Amazon were still better than the SmugMug, and better than other cloud service providers. We are fully convinced of these services. Kitely, a start-up company in Israel, uses only one Amazon area available. Fortunately, however, no interruption occurred in this area. However, kitely plans to learn from this experience. Kitely, vice president of research and Development, Oren Hurvitz, said we are planning to spread all our services across a range of available areas. Kitely runs a cloud-based virtual conferencing and collaboration environment based on the OpenSim platform. Kitely continuously checks to ensure the normal operation of its services. Hurvitz says our system is designed to be based on the assumption that any service can stop working at any time. If we find a server that is not responding, we shut down this server and start a new server.

Traditional backup

Another company is not affected by an outage because it uses more than one available area. The company is named Mashery. Mashery provides application programming interfaces to more than 100 companies such as Best Buy, Hoovers and the New York Times. However, Mashery also has another backup plan, a traditional datacenter.

Oren Michels, chief executive of Mashery, said it was very early on that we realized that there might be service problems that would make Amazon completely unusable. We're sure we need a fault-tolerant infrastructure. We have special hardware for Internap company. Internap network Services, located in Atlanta, not only provides hot backup sites for Mashery, but also provides a production environment for those who need services faster than cloud services, or services in areas where Amazon does not provide services. ' We have a lot of infrastructure on both sides to handle peak workloads, ' says Michels.

When Mashery started building its own cloud infrastructure two years ago, Amazon was the only real operator in town. Backing up to another cloud was not an option at the time. But now this may be a choice. ' We're really looking at this, Michels said. But if it's not bad, don't fix it. Amazon is working very well with us. Similarly, Internap has been a good partner and continues to provide us with the services we need.

Michels added that Internap even lowered prices to stay competitive. But Price is not the main factor in his decision. We have 100 major brand manufacturers customers. In the event of a failure, the loss of a large customer will be greater. Our customers pay us to solve their application programming interface problems. That includes the ability to keep the connection if there is an interruption.

Rob Enderle, an analyst at market research firm Enderle Group, says companies that have just begun to migrate to the cloud typically use traditional data centers as backup at the beginning. Enderle says you can have a secondary resource on standby. You can transfer fault tolerance to these resources. These resources are usually the devices you have before you transition to the cloud. You can be fault tolerant to lower performance technology and still have your customers.

However, Enderle warns against the belief that a cloud service is used as a backup of another cloud service in the same cloud service. He said a redundant service might use the same resources as the primary service. Care should be taken to ensure that the redundant service is redundant and not a different name for overlapping hardware and software.