Hello everyone, in the future, we will launch a discussion of the performance counters and LR scene design design of LR monitors in a topic, welcome everyone to speak. Today, I will put some of the counter and threshold requirements for me, these counters are some counters for me for the Windows operating system, C / S structure SQL Server database and web platform .NET product test; everyone can Continue additional, have made friends on the UNIX platform and J2EE architecture and WebLogic test, and I hope that the counter you use is posted and let everyone share. Ok, let's talk about it first, I hope to pass this topic, I will ultimately let everyone analyze their test results. Memory: Memory usage may be the most important factor in system performance. If the system "page exchange" is frequent, there is insufficient memory. "Page Switch" is the process of using a unit called "page", which moves the fixed size code and data block from the RAM to the disk, the purpose is to release memory space. Although some page exchange makes Windows 2000 to use more memory than actual memory, it is also acceptable, but frequent page exchange will reduce system performance. Reducing page exchange will significantly improve system response speed. To monitor the situation of insufficient memory, start from the following object counter: Available Mbytes: Available MBYTES: The number of physical memory can be used. If the value of Available Mbytes is small (4 MB or less), the total memory on the computer may be insufficient, or some The program did not release memory. Page / Sec: Indicates that the number of pages removed from the disk due to hardware page errors, or writes the number of pages of the workgroup space due to page errors to release the page number of workgroup spaces. Generally, if Pages / SEC continues above hundreds, then you should further study page exchange activities. It is likely to increase memory to reduce the demand for change page (you can multiply this number by 4K to get the hard disk data traffic caused by this). The value of Pages / sec does not necessarily indicate that there is a problem in memory, but may be caused by the program that runs using the memory map file. Page READ / Sec: The hard fault of the page, the subset of Page / Sec, for solving the memory reference, the number of page files must be read. The threshold is> 5. The lower the lower. The large value represents disk readings instead of cache read. Since too much page exchange To use a large number of hard disk space, it is possible to cause a disk bottle diameter that is insufficient to exchange the page swap memory. Therefore, when studying the reasons for page exchanges that are less apparent, you must track the following disk usage counter and memory counter: Physical Disk /% Disk Time Physical Disk / Avg.disk Queue Length, for example, including Page READS / SEC And% Disk Time and Avg.disk Queue Length. If the page read operation rate is very low, the value of% Disk time and avg.disk queuele length is high, there may be disk bottle diameters. However, if the page reading rate is not reduced while the queue length is increased, the memory is insufficient. To determine too many page swaps on disk activity, increase the value of Physical Disk / Avg.Disk Sec / Transfer and Memory / Pages / SEC counter. If the count result of these counters exceeds 0.1, the page exchange will take ten percent of disk access times. If this happens for a long time, you may need more memory. Page Faults / Sec: The number of flexible pages per second (including some can be met directly in memory, and some need to read from hard drives) Compared to Page / Sec to indicate that data cannot be used immediately in the designated work set of memory.
Cache Bytes: File System Cache, by default 50% available physical memory. If IIS5.0 is running in memory, it automatically organizes the cache. The trend change needs to be paid to this counter If you suspect that there is a memory leak, monitor Memory / Available Bytes and Memory / Committed Bytes to observe the memory behavior and monitor the process / private bytes you think may be in the process of leak memory. Set and Process / Handle Count. If you suspect that the kernel mode process has caused leaks, you should also monitor Memory / Pool NonPaged Bytes, Memory / Pool NonPaged Allocs, and Process (Process_Name) / Pool NonPaged Bytes. Pages Per Second: The number of pages per second. This number should be less than one page per second. Process:% Processor Time: The number of processor times consumed by the processor. If the server is dedicated to SQL Server, the acceptable maximum upper limit is 80-85% Page Faults / Sec: Compare the page failure generated by the process to determine the impact of this process to system page failure. Work Set: Processing the most recently used memory page in thread reflects the number of memory pages used in each process. If the server has enough idle memory, the page will remain in the work, when free memory is less than a specific threshold, the page will be cleared. InetInfo: private bytes: The current byte of this process cannot share with other processes. If system performance is lowered over time, this counter can be the best indicator of memory leakage. Processor: Monitoring Processors and Systems Objects The counter provides valuable information about the processor to help you determine if there is a bottleneck. % Processor Time: If this value continues to exceed 95%, it indicates that the bottleneck is CPU. You can consider adding a processor or for a faster processor. % User Time: Indicates database operations that spend a CPU, such as sorting, executing aggregate functions, etc. If this value is very high, consider adding an index, try to use a simple table connection, horizontal splitting large form, etc. to reduce this value. % Privileged Time: (CPU kernel time) is a percentage of time spent in the privileged mode. If the parameter value and "Physical Disk" parameter value have been high, I / O has problems. Consider replacing the faster hard disk system. In addition, TEMPDB IN RAM is set to reduce the value of "Max async IO", "Max Lazy Writer IO", and the value will be reduced. In addition, the server's server work queue that tracks your computer The current length Server Work Queues / Queue Length The processor bottleneck is displayed. The queue length continues to be greater than 4 indicates that the processor congestion may occur. This counter is the value of a specific time rather than a period of time. % DPC Time: The lower the lower. In multiprocessor systems, if this value is greater than 50% and Processor:% Processor Time is very high, adding a network card may increase performance, and the network provided is not saturated. Thread ContextSwitches / Sec: (Instantian INetInfo and DLLHOST Process) If you decide to increase the size of the threaded bypass pool, you should monitor the three counters (including one of them). Increasing the number of threads may increase the number of context exchanges, such a performance will not rise but will fall.
If the upper and lower text switching value of ten instances is very high, the size of the threaded bypass pool should be reduced. Physical Disk:% Disk Time%: Refers to the percentage of the time for the selected disk drive to provide services for the service to read or write requests. If the three counters are large, the hard drive is not a bottleneck. If only% of the% Disk Time is relatively large, the other two are more moderate, the hard disk may be a bottleneck. Before logging the counter, run DiskPerf -yd in the command line window of Windows 2000. If the value lasts more than 80%, it may be a memory leak. Avg.disk Queue Length: Refers to the average of the read and write requests (for the selected disk in the instance interval). This value should not exceed 1.5 to 2 times the number of disks. To improve performance, you can increase the disk. Note: A RAID DISK actually has multiple disks. Average Disk Read / Write Queue Length: Refers to the average of the read (write) request (column). Disk reads / s: The number of disk reads per second on the physical disk, the number of times written. The two are added, which should be less than the maximum capacity of the disk device. Average Disksec / read: Refers to the average time required to read data on this disc in seconds. Average Disk Sec / Transfer: Refers to the average time required to write data on this disk in seconds. Network interface: BYTES TOTAL / Sec: To send and receive bytes, including frame characters. Determine if the network connection speed is a bottleneck, and can use the value of the counter and the bandwidth comparison of the current network: Access Methods (Access Method) to monitor the method of accessing the logic page in the database. Full scans / sec (full table scan / second) unlimited full scan per second. Can be a basic table scan or a full index scan. If the value displayed by this counter is higher than 1 or 2, you should analyze your query to determine if a full table scan is required, and whether the S Q L query can be optimized. Page SPLITS / Sec (Page Split / Second) Due to the number of pages per second due to data update operations. Buffer Manager: Monitor Microsoft® SQL Server? How to Use: Memory Storage Data Page, Internal Data Structure, and Process Cache; Counters Read Database Pages from Disks in SQL Server and Write Database Pages to Disk Physical I / O. Monitoring SQL Server The memory and counter helps to determine: whether the bottleneck is caused due to the lack of data that can be accessed in the cache in the cache. If so, SQL Server must retrieve data from disk. Whether you can improve query performance by adding more memory or making more memory available for data cache or SQL Server internal structure. SQL Server needs to read the frequency of data from the disk. Compared to other operations, such as memory access, physical I / O will spend a lot of time. Minimize physical I / O to improve query performance. .Page ified reads / sec: The number of physical database pages issued per second. This statistic shows the total number of physical pages between all databases. Due to the large number of physical I / O overhead, the overhead is minimized by using a larger data cache, intelligent index, more efficient query or change database design. .Page write / sec (written page / second) The number of pages written by the physical database per second. .Buffer cache hit ratio. There is no page that is not read in "Buffer Cache / Buffer Pool) accounts for all pages in the entire buffer pool. You can find the percentage of the page read from the disk in the cache. This ratio is the total number of high-speed cache hits by the SQL Server instance to start the cache. After a long time, this ratio has changed very small. Since read data from the cache is much smaller than the overhead of reading data from the disk, it is generally desirable to high.
Typically, the cache hit rate can be improved by increasing the number of memory available to SQL Server. The counter value is determined according to the application, but the ratio is preferably 90% or higher. Increase the memory until this value continues above 90%, indicating that 90% of data requests can be obtained from the data buffer. Lazy Writes / sec (inert write / second) inert write process The number of buffers written per second. The value is preferably 0. Cache Manager (Cache Manager) Object provides a counter to monitor Microsoft® SQL Server? How to use memory stores such as stored procedures, special, and ready-to-TRANSACT-SQL statements, and triggers. The hit rate of Cache Hit Ratio (Cache His Motion, All Cache ". In SQL Server, Cache can include log cache, buffer cache, and procedure cache, a general ratio.) The ratio of the number of cache hits and the number of findings This is a very good counter for your system. If this value is low, you need to add more memory. Latches (latch) for monitoring To latch the internal SQL Server resource lock. Monitor latches to clarify user activities and resource usage, help to identify performance bottlenecks. Average Latch Wait Ti Me (MS) (Average latch wait time (milliseconds)) A SQL The Server thread must wait a average time of a latch, in milliseconds. If this value is high, you may have experienced serious competition issues. Latch Waits / sec (latlar waiting / second) in the lattory waiting quantity. If this value is high, you are experiencing a lot of competition for resources. Locks (Lock) provides information about SQL Server locks on individual resource types. Lock is added to SQL Server resources (such as behavior in a transaction) Take or modify it) to prevent multiple transactions from using resources. For example, if a row it (X) lock is released by a transaction, other transactions can not be modified before this lock is released. This line. Use the lock as little as possible to improve concurrency, thus improving performance. Multiple instances of the Locks object can be monitored at the same time, each instance represents a lock on a resource type. Number of deadlocks / sec (Quantity of deadlocks) / Sec) lead to the number of lock requests for deadlocks. Average Wait Time (MS) (average waiting time (ms)) thread waits for average waiting time for some type of lock. Lock requests / sec (lock request / second) per second The number of types of locks, Memory Manager: Monitoring the overall server memory usage to estimate user activity and resource usage, help to identify performance bottlenecks. Monitoring the memory used by the SQL Server instance helps Determine: Whether the bottleneck is caused due to the lack of data that can be accessed in the cache in the cache, if so, SQL Server must retrieve data from disk. Can you use more memory or make more memory available for data cache Or SQL Server internal structure to improve query performance. Lock Blocks: Lock on the server The number of blocks, the lock is on the page, row, or resource like this. Don't want to see a growth value. Total Server Memory: SQL Server Server is currently using the total amount of dynamic memory. Some counters needed for IIS Internet Information Services Global: File Cache Hits%, File Cacheflushes, File Cache Hits File Cache Hits% is a full cache request Central The proportion of it reflects the work of IIS's file cache settings.
