Expose the nine lies about cloud computing

Today, no matter where you go, someone will tell you or sell you something related to cloud computing. However, you only have to strip the coat of its lies and restore its original face, so that you can understand whether cloud computing is really suitable for your enterprise. The following is a list of the nine essential lies about cloud computing.

Lie 1: There is only one "Cloud ". There are at least three formats of "Cloud", each of which has different interests and risks. They are 1) "infrastructure as a service" (IAAS provides on-demand virtual servers, such as Amazon EC2); 2) web Services, or "Platform as a service" (PAAs), provide APIs or development platforms for customers to create their own applications in the cloud; 3) software as a service (SAAS ), for example, the CRM software of salesforce.com.

What applications are you running and what data is generated will produce completely different results for you and how cloud computing is used.

Lie 2: All you need is a credit card. If you are a lonely developer and have some time, it is okay to configure the virtual server on the command line, but if you are in charge of an enterprise, so how can I install and configure the operating system, multiple applications, and database connections in order to generate a return? These are all issues you must consider. If your company is large enough, you have to develop many standards, such as security, data format, and data quality.

Don't trust what some vendors have said: "As long as you buy a development server from a vendor, it takes 3 or 4 days for the IT department to complete the configuration and now 15 minutes to complete the operation ." Michael Kollar, chief architect of Siemens IT Solutions and Services North America, said that about 2500 of its servers require virtualization to provide cloud services to internal and external users. However, cloud services are neither secure nor suitable for the company's various standards and cannot be integrated into a wider IT environment.

Many IAAs vendors cannot even meet the application requirements of enterprises. Phil Calvin, founder and CTO of sitemasher, tries to find a cloud service provider to manage the servers he manages. What are the results? "We cannot find a vendor that can scale our standard servers as needed," he said. No vendor can provide the low-latency performance that it needs, nor can it provide global Load Balancing across data centers.

Lie 3: cloud computing can reduce the workload. This may be the case in the long term, but at the beginning, you must first consider which mode of cloud computing is suitable for you and which applications or services are suitable for this type of cloud; how to ensure proper security, compliance with regulations, and normal operation. It also takes additional time to monitor the service performance of any vendor.

"There are a lot of issues to consider when you are running productive applications, such as redundancy, reliability, performance, and latency," said Thorsten von eicken, CTO and founder of rightscale. Before migrating an application to the cloud, the customer must ensure that the above requirements are met.

In addition, not all applications are suitable for cloud computing. For example, applications that rely on Server clusters are not suitable for sharing computing resources with others in the cloud environment, because they require that the configurations of each server cannot be changed at will and require dedicated bandwidth.

Lie 4: seamless integration of private and public clouds. Cloud Service evangelists present us with two wonderful worlds: You can have control provided by internal data centers, you can also enjoy the low cost and flexibility provided by the external cloud, and all kinds of applications, storage and servers can be freely dragged and used according to your needs.

However, it is quite difficult to achieve this. Currently, a large amount of errands, manual reconfiguration, and engineering design efforts are required to migrate applications to both public and private clouds. Seamless integration is easy to implement only when the customer runs the same platform in Public and Private clouds.

According to Kollar of Siemens, the key is to have a security infrastructure that can be deployed across two environments. There must be a secure and cost-effective way to replicate or access data across two clouds.

Lie 5: it is impossible to seamlessly integrate public and private clouds. This is also wrong. In fact, some vendors are already providing seamless integration. For example, Kollar is expected to provide seamless integration between the two clouds to its customers within 12 to 18 months. Rightscale's von Eiken recommends some standardized configurations, data models, and automatic deployment policies for both public and private clouds. When technologies, standards, and processing procedures are complete, it is possible to build a foundation that can share public and private resources.

Lie 6: How can cloud computing save costs. McKinsey recently published a controversial white paper claiming that it is possible to save costs only when running certain special platforms, such as Linux, on the cloud.

Google objected to this. Rajen sheth, Senior Product Manager of Google Apps, believes that McKinsey's report only takes into account the cost savings brought about by low-cost servers in highly redundant architectures, which is not comprehensive. He believes that customers can use the same eXtensible Application Server and database, without having to purchase, install, maintain, and expand their own databases and application servers, which can also save costs. McKinsey did not respond to this.

Some users also pointed out that, under the existing license and technical support model, the customer may have to pay a higher fee to deploy software in the cloud than to deploy software in the enterprise.

Lie 7: cloud vendors can ensure cloud security. Even if cloud vendors have multiple security authentication certificates, they cannot guarantee that your specific servers, applications, and networks are secure.

Taking Siemens as an example, in order to ensure the safety of the Siemens cloud environment, IT systems should be carefully inspected from the outside, ensure that the path for every user to access key information is safe, Kollar said. It is a huge challenge to ensure that every platform is secure. Although it is not a challenge, it is necessary to ensure that all technologies that require security can work together.

Lie 8: As long as a virtual machine is running, it is doing cloud computing. Virtualization is the creation of logical servers or logical storage between multiple physical devices. It is only a part of the elements required by cloud computing. However, even with virtual machines, cloud computing does not exist. To fully benefit from virtualization, IT departments or cloud vendors must also provide capacity scaling capabilities and pay-as-you-go pricing mechanisms, this allows users to easily configure new servers and storage devices as needed.

It is a key cost-saving goal for cloud customers to allow users to order virtual servers (especially servers that are reconfigured for some special tasks) for some work. However, if you run VMware's Infrastructure 3, self-service will automatically happen, but not necessarily. Kollar said that Siemens has invested "a large sum of money" to develop standard service menus for virtual servers and related services. Users can order these services from their private clouds as needed.

Lie 9: cloud computing is only related to technology. Indeed, it is technology that makes cloud computing possible, but a correct process is required to achieve cost savings and flexibility. Virtualization is the foundation of cloud computing. When a customer wants to migrate data and applications between various physical devices, the former is extremely dynamic and highly changing.

Standardized processes can be used in the cloud to improve efficiency. Kollar said that by combining the ITIL management framework with virtualization technology, Siemens has reduced the IT management workload by 25% to 35%.

The truth about the cloud: cloud is not a takeout, nor a magical pleasure of carefree computing. It is a complex resource that requires in-depth understanding and efforts to correctly manage. There is no myth here.

Before migrating services to the cloud, enterprises should consider the following five issues:

1. How does your cloud service provider Define "good customer service? The advancement of the service itself is important, but you must find out which service a specific manufacturer thinks is a good service and what is included in the service, such as compensation policies and technical assistance.

2. Are you satisfied with the tangible facilities of the cloud service provider? Part of the due diligence includes inspection of the manufacturer's physical facilities, attention to matters related to the vendor's building maintenance, the manufacturer's equipment maintenance plan, and the number of employees in the building.

3. You should know which type of service interruption accident will occur. Sometimes, the vendor has to disable some update or upgrade devices, because those activities will seriously impede your cloud computing services. Determine how long the vendor plans to perform routine maintenance and what kind of advance notifications it will provide to users.

4. How fast is the development of cloud service providers? You must know whether your vendor is able to accept more new users without affecting the service level.

5. cloud service providers prepare the follow-upProgram? Once a service is interrupted, it is important that it can quickly identify the root cause of the problem.