Common energy-saving mistakes in enterprises.
When the company constantly looking for ways to stop the increasing cost of electricity, and they are contributing to global climate change an own strength, the company will find themselves surrounded by are Mired in energy crisis. But how can you be sure that your company's current power-saving strategy is the best? In fact, there are a lot of ways and tips for saving electricity, but most of them are wrong. In this article, we present a few incorrect ideas about how to save energy and tell you the truth.
Myth 1: Losing power and power to a computer can damage the life of the machine. Peak temperature and current changes in the power supply cycle can cause damage to electronic devices (especially capacitors and diodes) in the machine.
Fact: the power cycle is not a problem for electronic devices. In the IT equipment used in the electronic device can also be used in complex equipment, while the latter is often used in power supply cycle and temperature extreme cases, such as in factory automation equipment, medical equipment, and your car.
This argument has its truth, namely: periodic power supply for an unstable system will cause problems, you need to pay attention to the potential of defective components, these components are often overlooked in operation. With a specific management controller, the power diagnosis is usually fairly strict and is performed on a remote server. The power cycle is also useless for saving energy, and it is meaningless to maximize the use of the server.
Myth 2: for a server that needs a cold boot, it takes a long time to get started in the face of frequent requirements. If customers have to wait for the process, they usually leave.
Fact: the idle server that is a hot standby for zero workloads is an amazing waste of energy, and it is also a management burden. When you're cold starting a server to cope with a sudden increase in workload, if the user needs to wait, tell them. For a WEB site, place a static WEB page to tell the user to wait if additional resources are needed. For this kind of waiting, if they know that they're going to get something back, they usually keep the page. Put power management in your service architecture and make it part of your message to users and customers.
You can also choose systems that have a short boot time. Currently, according to different models and brands, the time the server displays during startup is different. This process is usually not measured, but when you need to turn off the system power, it becomes relevant. It doesn't take long. Those running from a mirror, a RAM load on a disk, or a server or blade on a SAN can move from a power outage to a working state in less than a minute. Most of the most effective TO keep/disaster recovery server TO SUSPEND - TO - RAM (STR) state, rather than into a complete power of state, so wake up the machine does not need the BIOS query and categorize the self-test or equipment, and the two work in the process of start is the most time-consuming.
Myth 3: looking at CPU power utilization (watts) is the simplest way to measure system efficiency.
Fact: power efficiency is measured by the ratio of power conversion, which can range from 50 to 90 percent. Ac power that does not convert to direct current is lost in the form of heat energy, which increases the burden of heat dissipation and increases the loss of the whole energy. Unfortunately, it is hard to tell the exact answer to the efficiency of the power supply, and many device manufacturers will not release it. You can make your decision by looking at the system's published efficiency or measuring the power consumption of the actual different systems in idle and full loads.
Myth 4: it's best to build a large server with all the RAM, CPU, and peripherals, which can save more power than using multiple smaller servers.
Fact: this argument is true only when large servers are fully utilized, and this is dangerous for mission-critical applications. Multiple small servers, if not used, may be cut off or put in a suspended state, which is also safer from a redundant perspective.
Similarly, a system tied too much too much CPU cores and RAM, it will cause the system to use more power, rather than a basic dual-core CPU and a modest basic configuration system is composed of RAM. Depending on the software you're running, the server's configuration can save you more power without taking drastic measures.