t–sql pl–sql_SQL Server处理器性能指标–第3部分–指示硬件组件问题的指标

t–sql pl–sql

Part 1 and 第1部分和 Part 2 of the SQL Server processor performance metrics series, we described the most important processor performance metrics – 第2部分 ，我们描述了最重要的处理器性能指标- Processor: % Processor Time, Process: % Processor Time, Processor Queue Length, Processor: % User Time, Processor: %Privilege Time, as well as various total times and thread metrics 处理器：处理器时间百分比，进程：处理器时间百分比，处理器队列长度，处理器：用户时间百分比，处理器：特权时间 ，以及各种总时间和线程指标

In this part, we will present processor performance metrics that indicate performance issues caused by hardware component problems

在这一部分中，我们将介绍处理器性能指标，这些指标指示由硬件组件问题引起的性能问题。

处理器：中断/秒 (Processor: Interrupts/sec)

Hardware interrupts are among the most common performance problem causes. They can induce processor bottlenecks, the same as other, more obvious causes such as insufficient memory

硬件中断是最常见的性能问题原因之一。 与其他更明显的原因（例如内存不足）相同，它们可能导致处理器瓶颈

An interrupt is an event generated on a hardware or software component, that sends a signal to the processor requesting immediate attention. When the processor is notified that there is a high priority event, it stops the execution of the currently processed thread, suspends all current activities, saves its state, and executes a specific procedure called the interrupt handler. After the interrupt handler finishes its tasks, the saved state is read and the processor continues where from the point where it was interrupted

中断是在硬件或软件组件上生成的事件，该事件将信号发送到处理器，要求立即关注。 当处理器被告知存在高优先级事件时，它将停止当前处理线程的执行，暂停所有当前活动，保存其状态，并执行称为中断处理程序的特定过程。 中断处理程序完成任务后，将读取已保存的状态，并且处理器将从被中断的位置继续

Hardware interrupts can be generated on a computer component (such as hard disk or a network controller) or on an external peripheral device. They are usually caused by errors in driver configuration, general hardware failure, or excessive component usage

硬件中断可以在计算机组件（例如硬盘或网络控制器）上或在外部外围设备上生成。 它们通常是由驱动程序配置错误，常规硬件故障或过多的组件使用引起的

Interrupts/sec is a processor metric that shows “The numbers of interrupts the processor was asked to respond to.” [1]

中断/秒是处理器指标，显示“要求处理器响应的中断数”。 [1]

An entity is shown for each processor on the machine, e.g. on a four-processor machine, four entities are shown

为机器上的每个处理器显示一个实体，例如，在四处理器机器上，显示了四个实体

The only device that is constantly creating interrupts is a processor hardware timer that generates an interrupt every 10 to 15 milliseconds, which results in up to 100 interrupts per second

唯一不断产生中断的设备是处理器硬件计时器，该计时器每10到15毫秒产生一次中断，每秒最多产生100次中断

The recommended value of Interrupts/sec depends on the processor, rate of disk I/O operations, and the number of network packets. However, for modern processors, up to 1,000 interrupts per second can be considered as normal

建议的“ 中断/秒”值取决于处理器，磁盘I / O操作的速率以及网络数据包的数量。 但是，对于现代处理器，每秒最多1,000个中断可以视为正常情况

High values of the metric can be encountered after adding a large number of new applications to the machine and they can indicate hardware problems and conflicts. Before coming to a conclusion, they should be investigated

在将大量新应用程序添加到计算机后，可能会遇到较高的度量标准值，并且它们可能表示硬件问题和冲突。 在得出结论之前，应对它们进行调查

“A dramatic increase in this counter value without a corresponding increase in system activity indicates a hardware problem. Identify the network adapter or disk controller card causing the interrupts. You might need to install an additional adapter or controller card.” [2]

“计数器值的急剧增加而系统活动没有相应增加，则表明存在硬件问题。 确定引起中断的网络适配器或磁盘控制器卡。 您可能需要安装其他适配器或控制器卡。” [2]

To check the disk controller and network adapter are working properly, use Windows Device Manager and check the device status for both

要检查磁盘控制器和网络适配器是否正常工作，请使用Windows设备管理器并检查两个设备的状态

Peaks can be encountered and they don’t clearly indicate a serious performance issue. If the peaks are less than 30 percent of the normal values, this is not alarming. If they exceed this value, deeper investigation is needed and the final solution would be upgrading to a faster processor

可能会遇到峰值，但并不能清楚地表明存在严重的性能问题。 如果峰值小于正常值的30％，这并不令人震惊。 如果超过该值，则需要更深入的研究，最终的解决方案将是升级到更快的处理器

Another metric that should be checked when troubleshooting high Interrupts/sec values is the Systems Calls/sec (explained below). Higher values are encountered in systems with excessive Windows system activity

对高“ 中断/秒”值进行故障排除时，应检查的另一个指标是“ 系统调用/秒” （下面说明）。 Windows系统活动过多的系统中遇到较高的值

处理器：％中断时间 (Processor: % Interrupt Time)

The Processor % Interrupt Time metric also shows the information about the interrupts, but unlike the Processor Interrupts/sec metric that shows the number of the interrupts, this metric shows the time the processor spends processing interrupts. To be precise, it shows the percentage of total processor time spent handling interrupts

“ 处理器百分比中断时间”度量标准还显示有关中断的信息，但与“ 处理器中断/秒”度量标准显示中断数量不同，该度量标准显示处理器花费在处理中断上的时间。 确切地说，它显示了处理中断所花费的总处理器时间的百分比

The recommended value is less than 50%. According to some DBAs, the maximum values should be up to 30 % per processor. Higher values might indicate hardware problems. The same as with Processor Interrupts/sec, in these situations it’s recommended to check the Systems Calls/sec value. If it’s higher than 20,000, it indicates that the system generates more interrupts than the processor can handle. Upgrading hardware components will most probably solve the issue

推荐值小于50％。 根据某些DBA的说法，每个处理器的最大值应高达30％。 较高的值可能表示硬件有问题。 与“ 处理器中断/秒 ”相同，在这种情况下，建议检查“ 系统调用/秒”值。 如果大于20,000，则表明系统生成的中断多于处理器的处理能力。 升级硬件组件最有可能解决问题

“Some components on the computer can force this issue and not really be a problem. For example a programmable I/O card like an old disk controller card, can take up to 40% of the CPU time. A NIC on a busy IIS server can likewise generate a large percentage of processor activity.” [1]

“计算机上的某些组件可能会导致此问题，但这并不是真正的问题。 例如，可编程I / O卡（如旧的磁盘控制器卡）最多可能占用40％的CPU时间。 繁忙的IIS服务器上的NIC同样可以产生很大比例的处理器活动。” [1]

总中断时间百分比 (% Total Interrupt Time)

In the Part 8, we described several total time metrics – Total Processor Time, % Total User Time, and % Total Privilege Time. The same logic is used when calculating the value for the % Total Interrupt Time metric – a sum of processor interrupt time percentages for each processor (Processor % Interrupt Time) divided by the number of processors

在第8部分中，我们描述了几个总时间指标- 总处理器时间，总用户时间 百分比和总特权时间百分比 。 计算“ 总中断时间百分比”度量值时使用相同的逻辑-每个处理器的处理器中断时间百分比之和（处理器中断时间百分比）除以处理器数量

The same as with other total times, for troubleshooting it’s necessary to monitor the interrupt time for each processor

与其他总时间相同，为进行故障排除，有必要监视每个处理器的中断时间

The threshold value for normal operation is the same as for Processor % Interrupt Time, i.e. individual processors, less than 30 %

正常操作的阈值与“ 处理器百分比中断时间 ”相同，即单个处理器小于30％

处理器：％DPC时间 (Processor: %DPC Time)

The Processor: %DPC Time metric shows the percentage of the time the processor spends on executing Deferred Procedure Calls (DPC). When a hardware interrupt happens, the interrupt handler can chose to put off some of the work for later by using a Deferred Procedure Call. These are also high priority tasks executed in the privileged (kernel) mode, but their priority is lower than the priority of interrupt handlers

处理器：％DPC时间度量标准显示处理器花费在执行延迟过程调用（DPC）上的时间百分比。 发生硬件中断时，中断处理程序可以选择使用Deferred Procedure Call推迟一些工作，以便稍后进行。 这些也是在特权（内核）模式下执行的高优先级任务，但其优先级低于中断处理程序的优先级

The time used by Deferred Procedure Calls therefore is included in the Processor: % Privileged Time metric. If the Processor: % Privileged Time value is high, check the Processor: %DPC Time to determine the cause

因此，“延迟程序调用”所使用的时间包括在“ 处理器：特权时间百分比”度量标准中。 如果“ 处理器：特权时间百分比”值很高，请检查“ 处理器：％DPC时间”以确定原因

Also have in mind that the sum of Process: % Processor Time: Total + Processor: % Interrupt Time + Processor: % DPC Time is equal to 100%. When working, the processor handles productive threads, service requests, or executes instructions (Process: % Processor Time: Total), processes hardware interrupts (Processor: % Interrupt Time), or processes Deferred Procedure Calls (Processor: %DPC Time)

还请记住， 进程：％处理器时间：总计+处理器：％中断时间+处理器：％DPC时间的总和等于100％。 在工作时，处理器处理生产线程，服务请求或执行指令（进程：％处理器时间：总计） ，处理硬件中断（处理器：％中断时间）或处理延迟过程调用（处理器：％DPC时间）

For high values of the Processor: %DPC Time metric, monitor the interrupt rate for each processor

对于较高的处理器值：％DPC时间度量，请监视每个处理器的中断率

The source of the Deferred Procedure Call requests is usually a network adapter. Disk can also be the source of excessive Deferred Procedure Call requests, but not so often. Replacing these components will solve the problem

延迟过程调用请求的来源通常是网络适配器。 磁盘也可能是过多的“延迟过程调用”请求的来源，但并非如此。 更换这些组件将解决问题

The threshold value for the Processor: %DPC Time metric is the same as for Processor: % Interrupt Time – up to 30%

处理器：％DPC时间指标的阈值与处理器：％中断时间的阈值相同-最高30％

Processor: Interrupts/sec, Processor: % Interrupt Time, % Total Interrupt Time, and Processor: %DPC Time indicate hardware issues that can affect SQL Server performance. All metrics have a clear threshold that signals a problem and usually a need for further investigation and checking other performance parameter values. Replacing the existing hardware components or using more powerful ones is in most cases the right solution

处理器：中断/秒，处理器：中断时间百分比，总中断时间百分比， 处理器：DPC时间表示可能影响SQL Server性能的硬件问题。 所有度量标准都有明确的阈值，表示存在问题，通常需要进一步调查和检查其他性能参数值。 在大多数情况下，更换现有硬件组件或使用功能更强大的组件是正确的解决方案

翻译自: https://www.sqlshack.com/sql-server-processor-performance-metrics-part-3-measures-indicate-hardware-component-problems/

t–sql pl–sql