At night time around the world, we can keep low-level applications running, and the usage of these devices is more efficient. During the day, the email server is usually running, and at night it may be some data warehouses, rather than dedicated servers used to run certain types of workloads, "Rasit said.
Cloud computing also reduces the need for some servers to be enabled while they are idle, Rasit adds. This is because enterprises usually only need to run their workloads at full capacity during the day's work hours-if they want to complete special projects, they may require more capabilities, then they can seek help from cloud providers or traditional service providers to overdraw such capabilities.
Aman Neil Dokania, Vice President and General Manager of the Asia-Pacific region and Japan region for sales of infrastructure software and blade servers, Hewlett-Packard Enterprise Business Unit, pointed out in an email that, many servers (such as HP's ProLiant servers and Blade Systems) are energy star certified, making them more energy efficient.
Unlike shutting down a server, keeping the server in STANDBY state will "Save a lot of energy and it will not affect the flexibility of the server when needed", Dokania said.
"We believe that some users will set the server to the standby mode instead of shutting them down completely, because in this case, the speed is much faster when they need to reactivate the server, "Dokania says." Ultimately, this depends on the importance of the application workload and the service level required by the business ".
However, enterprise users may still choose to shut down their servers. Dokania adds and explains that this situation may include considerations for hardware upgrades.
Consider disabling non-virtualized workloads
Gartner's Rasit also admitted that enterprises still have room to shut down their servers. According to him, only about 20% of enterprises around the world deploy virtualization. At the top of virtualization software, only half of the enterprise environment is ready for virtualization and integration.
Therefore, enterprise users may consider shutting down servers that run non-virtualized applications or do not target virtualization. This is true both at night and on weekends, especially if they do not want applications to be used too frequently during these shutdown periods.
But Rasit warned that "these plans are risky. This is the focus of debate among some conservative end users-if the server starts and runs, it can continue. If you shut down the server, when there is an application software requirement, you must start the \ Server and ensure that the application software runs smoothly, so the response time will be slow ".
When the server is repeatedly turned on and off during the day or within a month or within a year, excessive server labor loss may occur, leading to hardware component faults, Rasit pointed out.
In addition, if the data center is built in a remote location with relatively low energy costs, the cost of electricity and cooling saved by shutting down the server may not be obvious.
Alex, Service Director, Asia Pacific Data Center, IBM. Taylor said in an email interview that enterprise users should focus on workload demands rather than wasting their daytime work time on compressing servers. This is to ensure that the enterprise infrastructure is capable of undertaking sudden or unexpected spikes in work load.
When talking about the dynamic infrastructure that enterprise users have the ability to shut down or enable servers, Taylor says, "Obviously, this infrastructure has to spend more money on automation and additional control ".
Please note that the concept of shutting down and enabling a server has existed for many years, Taylor adds, in the region under his jurisdiction, many IBM users disable and enable servers in their virtualization projects.
Yap Chee Yuen, senior vice president of Resorts World Sentosa's creation and Technology Division, said in an email interview: "If Server workloads can be dynamically managed, it makes sense to disable or keep the server in standby mode. This will save energy costs or release resources to other application software. This method is similar to grid computing or computing resource sharing between computers ".
However, Yuen stressed that this option is not realistic for RWS companies because they need to keep their workloads running during peak and off-peak hours.
"For RWS, most of our key applications must run around the clock, and the features of the application workload are basically the same. In other words, our applications do not have the so-called peak and valley time periods, so this method is not realistic for RWS ".