Application Performance Management. HPE Recipe
The main task of any IT department is to provide users with high-quality and uninterrupted operation of applications, as well as their further development in accordance with the company's strategy. For this, it is necessary to control not only the applications themselves, but also to be able to monitor the impact on their functioning of the state of the infrastructure (network and server equipment, data storage systems), and how seamlessly the interaction with external clients and service providers is established.
To solve this problem, Hewlett Packard Enterprise offers an Application Performance Manager (APM) solution. In addition to monitoring tools for all types of applications (“normal”, cloud, mobile), it also integrates components designed to monitor the state of the infrastructure, including heterogeneous and hybrid. These components make it possible to understand who and how loads information systems, with which there are problems that reduce the performance of applications where errors occur.
The solution includes software tools for consolidating event information, as well as operational analytics tools that provide intelligent detection and response to events based on log files, collected metrics, events, etc. All these data can be further analyzed in depth. to visualize what is happening in the IT system.
')
The solution is based on service-resource models (Real-time Service Model, RTSM), which can take information both from external configuration management databases (Configuration Management Database, CMDB), for example HPE Universal CMDB, and from the DBMS built into the solution itself .
In order to ensure the quality of software solutions, it is necessary to understand how a particular application is perceived by the user in terms of availability, speed, etc. HPE APM has two main components designed for this.
The first is Real User Monitor (RUM). It allows you to "listen" for traffic between the user and the application, as well as between applications included in a service, for example, between the application server and the database. By analyzing the traffic, we can understand who connects to the system, whether there are no parasitic connections, how long the response time of the applications and the server is, how large the amount of transmitted data is.
The second component, Business Process Monitor (BPM), allows you to emulate user activity using protocol and interface level scripts.
The RUM module as a whole works in the same way as a sniffer, but it does not need to collect all traffic. In the settings of the component, you can specify what kind of traffic interests us in each specific case - depending on what is being studied.
There are several scenarios for using RUM. Firstly, it can be placed on a specialized device that is placed in the network break and analyzes the traffic going through it. The second option is to mirror traffic to a device with RUM, in this case its effect on the work of the object of study is substantially less.
If these methods are not suitable for any reason, then there is a third option: a software module (Plug-in) is installed on the server platform that scans the traffic of interest.
Real User Monitor supports work with a wide range of client workstations: PCs, thin clients, mobile devices. The set of supported protocols is also great — various HTTP protocols, SOA protocols, mail protocols, application servers, databases, network protocols for a number of specialized devices, such as ATMs.
When studying applications using Business Process Monitor, the load from real users is emulated by means of scripts that are distributed by “probes” - software components installed on separate computers located both in internal network segments and “outside”. In the latter case, the passing can be estimated from the point of view of not only internal users, but also external service providers or customers who are granted access to the corporate network.
After the distribution of scripts, the monitoring process begins, the emulation of resource calls at the protocol or interface level is performed, and the results of these calls are written to the database. The latter can be visualized depending on the purpose of the study, for example, in the form of maps showing the location of monitored users, histograms showing the progress of transactions and describing the nature of problems, if any, pie charts with error statistics for a selected time interval.
Testing with the help of scripts is carried out within the framework of the DevOps concept, which implies close cooperation of developers and testers with the operation service, which makes it possible to find bugs and improve the software much faster.
The best way to monitor performance is to share both RUM and BPM, because comparing test results in two ways can provide important information about what is happening in the IT system and will eliminate negative phenomena even before they affect users.
In addition to user behavior, it is necessary, as already mentioned, to study and the processes occurring at the infrastructure level. For this, HPE offers Diagnostics - a solution that allows you to understand what activity on the application servers causes activity of users. From servers built on .Net, Java and Python technologies, information is collected through software agents that are embedded in the technology on which the application is running (no need to change something in the application itself), and from a number of other servers (SAP, SQL Server, IBM WebSphere MQ) - with the help of the agentless solution Diagnostics Collector.
Diagnostics visualization tools allow you to visually present a detailed topological diagram of information flows, information interaction processes between users and application or database servers, as well as between the systems themselves. This data can also be useful in monitoring - sometimes the application “slows down” through no fault of its own, but because the server on which it is located is too busy with other operations.
RUM, BPM and Diagnostics can work together with ALM functional and load testing tools. At the same time, feedback with testers appears, and the scripted developments of the latter are transmitted to the monitoring system for later use. Conversely, RUM is able to create a script that actually displays what is happening in the work environment. This makes it possible during testing to set the correct loads for the application that must be installed in the system.
In addition, RUM is an additional source of information for configuration management databases: it allows you to detect previously unaccounted protocols, open ports, etc., that applications use during their work.
Also, do not forget that all processes in the IT system go through the network. Therefore, to assess the performance of the application, it is necessary to pay sufficient attention to monitoring the operation of the network. Through data integration from HPE Network Node Manager with Real User Monitor or Business Process Monitor, you can see which network devices stand in the way of traffic from an application to a user - real or emulated, and in case of braking or unavailability of the application, see if on this network.
To get the most accurate information about the state of the IT infrastructure, HPE offers to collect all the information coming from applications, networks, servers and other devices into Vertica DBMS, designed to work with Big Data, and analyze it using HPE Operation Analitics. This analysis provides not only a complete picture of what is happening in the IT system, but also forecasts regarding the development of events, allows you to identify anomalies in the behavior of certain components of the system, and learn about previously unknown problems.
The resulting information can be visualized using the HPE Business Value Dashboard - a tool for creating individual consoles that allow you to customize them to suit the tasks performed by employees and the levels defined for them in the management hierarchy.
Recently released a new version of APM - 9.3. It has updated some interfaces, as well as added support for a number of protocols and new versions of browsers, in particular, Microsoft Edge (albeit incomplete due to the features of this program).
To solve this problem, Hewlett Packard Enterprise offers an Application Performance Manager (APM) solution. In addition to monitoring tools for all types of applications (“normal”, cloud, mobile), it also integrates components designed to monitor the state of the infrastructure, including heterogeneous and hybrid. These components make it possible to understand who and how loads information systems, with which there are problems that reduce the performance of applications where errors occur.
The solution includes software tools for consolidating event information, as well as operational analytics tools that provide intelligent detection and response to events based on log files, collected metrics, events, etc. All these data can be further analyzed in depth. to visualize what is happening in the IT system.
')
The solution is based on service-resource models (Real-time Service Model, RTSM), which can take information both from external configuration management databases (Configuration Management Database, CMDB), for example HPE Universal CMDB, and from the DBMS built into the solution itself .
Two views of applications: RUM and BPM
In order to ensure the quality of software solutions, it is necessary to understand how a particular application is perceived by the user in terms of availability, speed, etc. HPE APM has two main components designed for this.
The first is Real User Monitor (RUM). It allows you to "listen" for traffic between the user and the application, as well as between applications included in a service, for example, between the application server and the database. By analyzing the traffic, we can understand who connects to the system, whether there are no parasitic connections, how long the response time of the applications and the server is, how large the amount of transmitted data is.
The second component, Business Process Monitor (BPM), allows you to emulate user activity using protocol and interface level scripts.
System "the eyes of the user"
The RUM module as a whole works in the same way as a sniffer, but it does not need to collect all traffic. In the settings of the component, you can specify what kind of traffic interests us in each specific case - depending on what is being studied.
There are several scenarios for using RUM. Firstly, it can be placed on a specialized device that is placed in the network break and analyzes the traffic going through it. The second option is to mirror traffic to a device with RUM, in this case its effect on the work of the object of study is substantially less.
If these methods are not suitable for any reason, then there is a third option: a software module (Plug-in) is installed on the server platform that scans the traffic of interest.
Real User Monitor supports work with a wide range of client workstations: PCs, thin clients, mobile devices. The set of supported protocols is also great — various HTTP protocols, SOA protocols, mail protocols, application servers, databases, network protocols for a number of specialized devices, such as ATMs.
In emulation mode
When studying applications using Business Process Monitor, the load from real users is emulated by means of scripts that are distributed by “probes” - software components installed on separate computers located both in internal network segments and “outside”. In the latter case, the passing can be estimated from the point of view of not only internal users, but also external service providers or customers who are granted access to the corporate network.
After the distribution of scripts, the monitoring process begins, the emulation of resource calls at the protocol or interface level is performed, and the results of these calls are written to the database. The latter can be visualized depending on the purpose of the study, for example, in the form of maps showing the location of monitored users, histograms showing the progress of transactions and describing the nature of problems, if any, pie charts with error statistics for a selected time interval.
Testing with the help of scripts is carried out within the framework of the DevOps concept, which implies close cooperation of developers and testers with the operation service, which makes it possible to find bugs and improve the software much faster.
The best way to monitor performance is to share both RUM and BPM, because comparing test results in two ways can provide important information about what is happening in the IT system and will eliminate negative phenomena even before they affect users.
Servers and networks
In addition to user behavior, it is necessary, as already mentioned, to study and the processes occurring at the infrastructure level. For this, HPE offers Diagnostics - a solution that allows you to understand what activity on the application servers causes activity of users. From servers built on .Net, Java and Python technologies, information is collected through software agents that are embedded in the technology on which the application is running (no need to change something in the application itself), and from a number of other servers (SAP, SQL Server, IBM WebSphere MQ) - with the help of the agentless solution Diagnostics Collector.
Diagnostics visualization tools allow you to visually present a detailed topological diagram of information flows, information interaction processes between users and application or database servers, as well as between the systems themselves. This data can also be useful in monitoring - sometimes the application “slows down” through no fault of its own, but because the server on which it is located is too busy with other operations.
RUM, BPM and Diagnostics can work together with ALM functional and load testing tools. At the same time, feedback with testers appears, and the scripted developments of the latter are transmitted to the monitoring system for later use. Conversely, RUM is able to create a script that actually displays what is happening in the work environment. This makes it possible during testing to set the correct loads for the application that must be installed in the system.
In addition, RUM is an additional source of information for configuration management databases: it allows you to detect previously unaccounted protocols, open ports, etc., that applications use during their work.
Also, do not forget that all processes in the IT system go through the network. Therefore, to assess the performance of the application, it is necessary to pay sufficient attention to monitoring the operation of the network. Through data integration from HPE Network Node Manager with Real User Monitor or Business Process Monitor, you can see which network devices stand in the way of traffic from an application to a user - real or emulated, and in case of braking or unavailability of the application, see if on this network.
Analyze everything and always
To get the most accurate information about the state of the IT infrastructure, HPE offers to collect all the information coming from applications, networks, servers and other devices into Vertica DBMS, designed to work with Big Data, and analyze it using HPE Operation Analitics. This analysis provides not only a complete picture of what is happening in the IT system, but also forecasts regarding the development of events, allows you to identify anomalies in the behavior of certain components of the system, and learn about previously unknown problems.
The resulting information can be visualized using the HPE Business Value Dashboard - a tool for creating individual consoles that allow you to customize them to suit the tasks performed by employees and the levels defined for them in the management hierarchy.
Recently released a new version of APM - 9.3. It has updated some interfaces, as well as added support for a number of protocols and new versions of browsers, in particular, Microsoft Edge (albeit incomplete due to the features of this program).
Georgy Tokayev, Technical Consultant, Department of Software Solutions, HPE in Russia
Source: https://habr.com/ru/post/308762/
All Articles