Cloud service level agreement signed: Do not put eggs in a basket

Amazon's enterprise-class cloud service EC2 failed some time ago, which prompted many corporate CIOs to sign a cloud service-level agreement (SLA): "Do not put all your eggs in the same basket." . For CIOs at major enterprises, it is important to consider the extremely critical performance of the cloud services provided by the vendor in terms of security, performance, and redundancy when using external cloud applications.

"When choosing a supplier's cloud services, you have to understand that not all enterprise applications can be implemented on cloud services," said Ptak, partner and chief executive of Noel & Associates LLC in Langmont, Colorado, USA Analyst Audrey Rasmussen commented.

Rasmussen further recommends that before adopting cloud services provided by suppliers, enterprise CIOs should carefully investigate the infrastructure design and technical details of these cloud services, and must sign their cloud service-level agreements with their own needs The function is listed as a service requirement. Compared to the average business application, the important business system is obviously more dependent on the higher level of redundancy and failover capability. Of course, high-level service agreement overhead is relatively more expensive, so IT needs to make a smart choice between expenses, risks, and benefits.

Cloud service level agreement challenge

At least for IT executives, this challenge requires considerable courage.

Henry Mayorga, a New York-based financial services company and Baron Capital's network technology manager, says "finding 100% availability is as unrealistic as finding the legendary Holy Grail because" the more you add redundant devices to your system , Then your system complexity will be higher and higher.If all the equipment and configuration are indeed properly connected and set in place, then the theoretical high availability exists.However, once the error is set in the system of any A part or any one component does not work as expected, the difficulty of finding the cause of the error increases exponentially and the difficulty of restoring the system also multiplies the love equally. "

Even so, out of trouble with risk management, Tanya Forsheit, a partner at an information law firm in Los Angeles, and a cloud services expert, still maintains that IT executives must still stick to a higher level of protection, And give up considering those who can not really meet the needs of business programs. "The needs of businesses are not the same, so every single business needs needs to be tailor-made after negotiation and confirmation, not to accept the stereotyped service model that most cloud service providers provide."

Before signing a cloud service level agreement, you must be clear about the impact of the failure on your business

Phil Kramer, chief technology officer at a technology company focused on systems integration consulting solutions, believes an organization's CIO must make prior evaluations of the services provided by cloud service providers, the impact and loss in the event of an outage.

Kramer example to: "Unless the server also crashed at the same time, if only the user's server list directory was not backed up overnight, this is a very low-risk failure; but if it is the user's precious information? In the absence of backup The server under which the information is stored crashes, and the consequences can be imagined, which is definitely a bad choice if your provider's cloud service level agreement does not ensure your data access. "

"The high level of redundancy is less important if the cloud services you are using are for testing and prototyping purposes only," said President and Chief Operating Officer of Hurwitz & Associates LLC, a consulting firm based in Norfolk, Mass. Judith Hurwitz thinks so, "But if a company uses a cloud service from a vendor to support a customer-facing enterprise service, then the company must ensure the service has the right backup strategy and business continuity solution."

Once the company has identified each application's need for reliability levels, it can start looking for a solution. Vinoo Jacob, a data services product manager at Vector, a New Jersey-based network infrastructure service provider, warned: "There are a range of options to choose from, including traditional in-site backup and recovery to quick failover across multiple vendors To ensure business continuity. "

Amazon's cloud service incident forced customers to place higher demands on their service providers, Jacob said. "Although this is a short-term failure, it helps to clarify the definition and evaluation of cloud services in the future; it's not about whether or not Decided to adopt the issue of cloud services, but on how to assess and adopt the cloud architecture based on how to design system architecture issues. "

In fact, cloud services and service-level agreements have changed the role of CIOs, Ptak, Noel's Rasmussen said. "Instead of just managing the IT services that they provide internally, as they are in the past, CIOs now have to monitor both internal and external IT Service, and to ensure that the necessary functions required by the business and the corresponding service level are seamlessly docked. "

A few questions to consider when signing a cloud service level agreement

How do you get notifications when cloud services are not available? Can the service be restarted automatically?

How does your cloud service provider ensure quality of service to protect your business?

Does the supplier provide load balancing or reliability assurance?

What is your requirement for vendor RPOs and RTOs?

How can vendors provide you with the ability to manage your application's performance and troubleshoot?

Translator's Note: Amazon's EC2 (Elastic Compute Cloud) service is similar to Google's Instance Virtual Enterprise Cloud Service, a system that lets users rent applications for cloud computing operations. EC2 provides users with the flexibility to run their own Amazon machine image files by providing web services to run any software or application they want on this virtual machine.

Users can "flexibly" use the services provided by EC2 to create, run, terminate their virtual servers at any time, and charge on a per-use basis. EC2 gives users control over the host location on which the virtual server is running, which minimizes business latency-for example, users can run their own virtual server in each time zone for the shortest system maintenance time.