Shulou(Shulou.com)06/01 Report--

One: monitoring components.

Sensor: collect data and detect data.

Data acquisition: CPU utilization disk space, memory bandwidth user online rate = =.

Data storage (time series data) data display (graphics) alarm

(the data collected exceeds the threshold)

IT system monitoring mechanism: we use a platform (network monitoring system NMS) to periodically send data requirements to the monitored computer, which collects the data and then feeds it back to the monitoring computer (periodically). After the collection, the storage is realized, and then the data is analyzed.

SNMP protocol: simple network management protocol Simple Network Management Protocol (composed of the monitoring side and the monitored side)

SNMP protocol version: v1, (no authentication function) v2 (NMS-- "agent shared secret key plaintext authentication), v3 implements authentication encryption and decryption functions

SNMP mode of operation:

First, the monitoring side (NMS) initiates a request to collect data from the monitored side.

Second, the monitored side (Agent) initiates to report the data to the monitoring side.

Third: once the monitoring end finds that the monitored end has an abnormal state, the management side initiates to modify the data to the monitored end.

Components of SNMP:

MIB:management information base (Management Information Base) any one of the monitored resources can describe a collection of unmonitored objects. Floating point, × ×, and so on. The interface between NMS and Agent. Through him to send requests, to achieve data interaction (thin pipeline so the port is powerful. The agreement is too simple, so stipulate the documents first. )

MIB view: define the execution node and draw a larger MIB tree structure (for example, ignoring B and other execution of one of the indicators of C through Agent)

From the bottom down, each object that can be monitored has its own OID. (not only the MIB system but also other monitoring objects)

SMI: representation symbols for MIB

SNMP protocol implements network management itself based on TCP/TP protocol.

Actions that can be initiated by NMS

Get, get data, GetNext. Get multiple parameters, Set modifies the configuration, and Trap captures the active report data sent by each monitored terminal.

Agent:Response, which returns one or more operation values.

UDP (port-based),

NMS:161

Agent:162

Linux uses snmp:net-snmp packages

Must be able to achieve: note that in case of a state transition of a monitored object, the soft state of the first time is sampled many times before it becomes hard before the alarm can be given.

Nagios: flexibly call a variety of ports for powerful alarm mechanism. When the hard state of the software occurs, call the police immediately. (email and SMS) customize the alarm frequency. The alarm is completed within a specified period of time, and the alarm is upgraded to a high level in turn. It is defined that there is no alarm during the maintenance time (within a window, the alarm operation is not performed when a data transformation occurs). Define the dependency relationship of the monitored node to avoid being affected by a node break and get a large number of alarms. It is not suitable for environments where many indicators are used on a large scale. The distributed monitoring system can not be well implemented. Do not collect specific data, only care about whether the abnormal state transition, and complete the data alarm.

Zabbix: combine the advantages of cacti and Nagios. Once an exception occurs in a certain index, the script will be called remotely to solve the problem, and then the alarm will be realized upwards. Complete the integrated function of data acquisition, storage collection, display and alarm.

Famous open source monitoring tools: zabbix,zennos,opennms,cacti,nagios (icinga), ganglia.

The realization of zabbix monitoring function: through a special zabbix process to obtain the super management rights of the monitored computer for data collection. (complete local command or local script execution for operational restart or other configuration. If you listen again and find that you are not up yet, then call the police.)

Mode support: dedicated agent,ssh,SNMP,IPMI

IPMI: smart platform management interface. Control the startup, shutdown and so on of the monitoring server.

Zabbix: has its own dedicated agent monitoring tool.

Monitoring host:

Linux/windows/freeBSD

Network devices:

SNMP,SSH (not all)

Who can be monitored:

Equipment / softwar

Equipment: server, router, switch, IO system

Software: OS, network, application

Occasional glitches:

Server down, service unavailable, host unreachable

Serious failure: disk is full, memory is out. Cpu. Wait.

Host performance index

Trend: time series data (an indicator within a certain period of time)

Data storage:

Cacti:rrd (round robin datebase ring database is not persisted and all data will be overwritten)

Zabbix:mysql.pgsql . (call the API interface of mysql to save drawing data for operation)

1998 started, 2001 officially launched. Version 2009 has become more mature. Version 2.4 is up to date.

Zabbix enterprise-class monitoring that can monitor up to 100000 devices.

Really open source, a variety of users up to more than 4W. Supported monitoring systems include mac,windows hp,unix aix reeBSD openBSD linux SOLARIS

Zabbix workflow: periodically collect and store data, define a trigger or multiple threshold indicators for each monitor to generate abnormal events, trigger action, execute programs, and send alarms. The drawing interface takes the data out and draws the trend chart to display, so as to realize the visualization of the data. It is handed over to a large amount of traffic, such as disk IO, network IO and so on, so distributed monitoring is carried out.

Distributed monitoring: form an agent to monitor the value from the agent to reduce the pressure of central control. To achieve monitoring in different regions.

Automatic monitoring system

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.