System performance is a system to provide it a quasi-more performance to users of a mixture, including both hardware performance, including software performance.
Use according refinement: integers, floating point, response time, network bandwidth, stability, I / O throughput, SPEC-Int, SPEC-Fp, TPC, Gibson mix the like. SPEC --- authority to test CPU performance.
Two branches: 1 as a reference for future development of computer technology and planning; two pairs of existing systems to adjust the performance has been optimized.
Four aspects of system performance
- Performance indicators: describe the performance of the main indicators of the current popular systems involved;
- Performance calculation: calculation methods are described to the currently used key performance indicators
- Performance Design: Describes how to adjust the existing system performance optimization, and introduce a few mature design rules and solutions
- Performance Assessment: Describes how to evaluate and improve the performance indicators currently obtained.
Performance:
Performance indicators mentioned here, both software, but also hardware. In hardware, both computers, including various communication switching devices, and other network hardware; in software, including both the operating system and various communication protocols, but also participate in a variety of communication applications, such as database systems , Web servers, etc. Therefore, the system performance to be mentioned in this section is actually integrated performance of these hardware and software.
1. Computer Performance Performance indicators
performance computer performance : . 1. The clock frequency (frequency) cache, the operation speed 3, 4, operation Kyoto, 5, the storage capacity of the memory 6, a memory access cycle , 7 data processing rate, the response time of 8 9, RASIS average failure response time characteristic 10 11 compatibility
RASIS characteristic reliability (Reliability), availability (Availability), maintainability collectively (Serviceability), integrity (Integraity) and security (Security) without those.
Reliability is the probability that a computer system at specified operating conditions and requirements of working hours continued to run properly. Reliability MTBF generally used (Mean Time to Failure, MTTF), or MTBF (Mean Time Between Failure, MTBF) is measured.
Maintainability is the ability of a system failure after the repair as soon as possible, the general recovery time (Mean Time To Repair, MTTR) with a mean time to failure to represent.
2. Network performance indicators:
(1), equipment and performance indicators. (2) and network performance, (3) and application performance (the QOS) (. 4) user-level performance index (5) Throughput
3. The operation of the system performance:
Reliability (1) system, the throughput (2) system, (3) the system response time (turnaround time) [ the user to submit jobs from the results obtained this time] (5) transplantation
4. database management system
To ensure the database where the data is stored in a safe and consistent, DBMS is to manage this in
(1) Database Description function. Define global logical structure of the database, the local logical structure and various other database objects.
(2) database management functions. Including system configuration and administration, data access and update management, data management and data integrity security management.
(3) to query and manipulate database functions. This function does a database to retrieve and modify packages
(4) do a database maintenance functions. Including the introduction of database management, database structure maintenance, data recovery and performance testing, in order to improve the efficiency of database development, modern database systems in addition to DBMS, but also provides a variety of support application development tools. Database management system to measure key performance indicators database itself consists of two parts and management systems.
Performance metrics database and database management systems include the size of the size of the database, the single database file, the number of data representation, the size of a single table, running the table records (rows), the size of the individual rows.
Allowed number of indexes on a table, the database allowed number of indexes, the maximum concurrent transaction processing, load balancing capabilities. The maximum number of connections.
5. Web server
Web key performance indicators including the maximum number of concurrent server connection, a response delay, the throughput (the number of requests processed per second), the number of successful requests, failed requests, the number of hits per second, the number of successful hits per second, number of clicks per failures try connections, user connections and so on.
The main performance calculation method : method definitions, formula, detection procedures, detection equipment. Defined method is mainly acquired by definition directly over the data, formulas rules generally applicable to calculations derived in accordance with the basic definitions that meet the performance index, and the program detection and instrumental detection rule is to obtain its actual value is an actual test (due to the test the uncertain environment and conditions, the results may also be relatively large difference)
Article Number instruction MIPS = ÷ (execution time × 10 ^ 6) = Fz ÷ CPI = IPC × Fz Fz of processors working frequency, CPI is the average number of clock cycles required for each instruction, IPC per clock cycle the average number of instructions executed.
eg: To calculate Pentium IV / 2.4E processor operation speed, because the Pentium IV / 2.4E processor of the IPC = 2 (or the CPI = 0.5 ), Fz = 2400MHz , so MIPSP4 / = 2.4E the IPC ' Fz = 2 ' 2400 = 4800MIPS .