System monitoring program:
The current mainstream surveillance program has three, namely:
- Zabbix
- Prometheus
- Open-falon
I. Introduction
Zabbix
zabbix is a veteran monitoring systems, enterprise-class open source monitoring software, web-based interface. Zabbix server LAMP environment or need LNMP environment, providing a distributed system monitoring and network monitoring. Which includes a host of performance monitoring, network equipment performance monitoring, performance monitoring database, a variety of alarm mode, detailed reports, charting functions. Monitoring objects can be Windows or Linux server, it can be routers, switches and other network equipment, servers and other network to provide remote monitoring via SNMP, zabbix Agent, PING, port monitoring and other methods, data collection and other functions.
Features : can monitor servers, routers, switches,
Prometheus (Prometheus)
Prometheus is an open source version of the open source monitoring system developed by SoundCloud. 2016, Google sponsored by the Linux Foundation (Cloud Native Computing Foundation, CNCF) to Prometheus into its second largest open source projects. Prometheus also very active in the open source community. Prometheus is a fellow with the kubernetes relationship, CNCF first project is open source k8s, the second project is to prometheus.
Features : more suited to a cloud environment monitoring, especially in service monitoring system under k8s
Open-falcon
Open-Falcon out of the Internet is an open source enterprise-class monitoring system operation and maintenance department millet, currently include millet, Jinshan Yun, the US group, Jingdong finance, go to the market and so the use of Open-Falcon. Open-Falcon whole can be divided into two parts, i.e. the drawing assembly, the alarm assembly.
Is responsible for data collection "graphics component", the collection, storage, archiving, sampling, query, display (Dashboard / Screen) and other functions, can work alone, storing a display program as the time-series data.
"Alarm component" responsible for alerting policy configuration (portal), the alarm determination (judge), alarm handling (alarm / sender), user group management (uic), etc.,
Features : Domestic monitoring system, the wealth of information
to sum up:
Less Zabbix documentation, not many configuration items, feature rich enough, the advantage can be configured directly on the page;
Prometheus features more than Zabbix, less than the Open-falcon, configuration items and more but the documentation is not detailed enough; feature rich Open-falcon, official documents also detail.
Second, the difficulty of installation and deployment
Zabbix:
- Installation zabbix-server terminal
- Using web page configuration
- Increase in client-side monitoring is done on the web
Prometheus
- Installation go locale
- Installed on the monitored server prometheus
- Installed on the monitored environment client export
- Installation grafana (may not be installed)
Open-falcon
- Install redis
- Install mysql
- Installation go locale
- Install the back-end module
- Client Installation
Third, data collection
zabbix
Zabbix data acquisition through SNMP, Agent, ICMP, SSH, IPMI and other system. When the amount of data to show the need for direct reading the database, the card will be slow
Node Monitoring Scale: 1000+
prometheus
Prometheus is a basic principle of the HTTP protocol crawl status periodically monitored component, any component as long as the corresponding HTTP interface can monitor access. Without any SDK or other integration process. This is very suitable for a virtualized environment monitoring system, such as VM, Docker, Kubernetes and so on. Output monitored component information HTTP interface called exporter. At present the Internet company most commonly used components are exporter can be used directly, such as Varnish, Haproxy, Nginx, MySQL, Linux system information (including disk, memory, CPU, network, etc.).
Data collected include Prometheus host performance monitoring, performance monitoring vessel, Nginx access traffic performance, status, and the individual components Kubernetes internet. Prometheus refers to the current local data retention month, the preservation of historical data to M3DB in.
Node Monitoring Scale: 1000+
Open-falcon
Open-falcon tcp-based data transmission protocol. agent node can be automatically acquired based system monitoring index, and reported to the transfer, the transfer agent long TCP connection is established, data is sent once every 60 seconds to transfer.
open-falcon's agent components directly support CPU, Load, memory, disk, IO, network-related kernel parameters, ss statistical output ports collection, process core service of survival information gathering, critical business processes resource consumption, NTP offset acquisition, DNS resolve collection. There are third-party components can be extended to support more data collection.
Node Monitoring Scale: 1000+
Fourth, data storage
Zabbix
Storage: MySQL prostgresql
Zabbix is stored in MySQL, it can be stored in other database services. Zabbix the use of a relational data store time series data, it is often stretched in terms of the data stored in the monitor large-scale clusters. So start from Zabbix 4.2 version supports TimescaleDB timing database, but the current maturity is not high.
Prometheus
Storage: opentsdb
Prometheus research from a high-performance timing database, the V3 version can reach tens of millions per second level of data storage, through a third-party docking sequence database extended stored historical data.
In order to overcome the limitations of single-node storage, prometheus did not realize his clustered storage, but provides a remote interface to read and write, let the user choose the right timing to achieve scalability prometheus database storage.
Open-falcon
Storage: Archive with RRD storing mysql + redis + opentsdb
Open-Falcon RDD data storage are used, is also added to the Open-Falcon sliced data consistency hash algorithm, and may receive OpenTSDB.
RRD stands for Round Robin Database, that is, "Ring the database." As the name implies, it is a recycled database storage space for storing data and associated time series.
RRD database is created when you've defined size, when storage space is full, and overwrite the old data from the beginning, so the linear growth and other databases different, RRD of controlled size and no maintenance.
to sum up:
If only short-term storage and query, Prometheus, Open-falcon, zabbix can meet the existing requirements. However, if large amounts of data to be analyzed prediction procedure, all of which are required to use other databases accessible spark, mapreduce like. Open-falcon need for secondary development, Prometheus provides an interface to remote read and write. Zabbix background in C development, secondary development difficult.
Fifth, alarm monitoring
1.Zabbix can implement complex multi-alarm information
Phone alert, email alert, micro-channel alarm, SMS alarm, alarm upgrade mechanism (no matter what the alarm can be).
2.Prometheuss
Currently supported e-mail, Slack, micro-channel and webhook, if it is docked nails, you can send an alarm to trigger client-side nails through webhook way.
3.Open-falcon
SMS, micro-letters, mail, and telephone.
Alarm levels can be configured, such as P0 / P1 / P2, etc., and each of the other alarms will correspond to a different queue redis.
Six options
|
|
Installation complexity |
Data Acquisition Support |
Data storage difficulty |
Alarm support |
| Zabbix |
in |
low |
high |
high |
| Prometheuss |
low |
high |
high |
in |
| Open-falcon |
high |
in |
low |
in |
Compare various features of monitoring indicators from these data, it can be concluded open-falcon is not the most powerful, but it is the most simple and convenient. In particular, do not worry about data storage disk to increase the amount of data has been a problem that occurs, operation and maintenance cost savings. Taking into account the size of the current server openlab system, functional requirements are not complicated and alarm extended, open-falcon may need are able to meet the demand. So finally recommended continued use of open-falcon.