Introduction to the hottest VMware software defini

  • Detail

Introduction to VMware software definition data center analysis tools

in the next period of time, the author will share VMware analysis tools, vrops and loginsight. Due to the limited space, I will introduce vrops (it is expected that there will be three) in the form of serial, and then show that the wiring polarity is correct. I will introduce loginsight

analysis tools are used to analyze data. When it comes to the types of data, it can be divided into structured data and unstructured data. In the SDDC solution of VMware, we have corresponding analysis methods for both data. For structured data, it mainly refers to the performance, alarm and event information in the virtualized environment, which is collected through vrops and vCenter communication, analyzed and displayed to users in the form of reports. For unstructured data, it mainly refers to logs. Loginsight is used to communicate with vCenter, collect information, and then conduct intelligent analysis to facilitate users to view and trace. In this article, I will first introduce vrops

what is vrops? The full name of vrops is vrealizeoperationsmanager. When it comes to what vrops is, quite a few people first react that it is a monitoring tool. In fact, this is not very accurate

strictly speaking, vrops is an analysis tool. All data related to virtualization displayed in vrops come from vCenter. In fact, vCenter mainly focuses on configuration management and has simple monitoring functions. For relatively complex scenarios, it can not meet the requirements of monitoring, operation and maintenance. These tasks need vrops to complete. This is also a question that many customers have asked me: vCenter can monitor the total virtual machine, and what do you want vrops to do

for example, if the information monitored by vCenter is compared to the specific indicators and results of each physical examination item during our physical examination (in short, non-medical personnel cannot understand it), vrops is the summary of physical examination results and doctors' suggestions presented on the front page of the final physical examination report. Therefore, the true meaning of vrops is that it can visually show the operation of all virtual machines in the production environment to it administrators and give corresponding suggestions. Therefore, vrops saves operation and maintenance management costs and ensures that the analysis process is standardized

two vrops indicators

measure the operation of an object (such as a virtual machine). Vrops has three indicators: operation (or health), risk and efficiency. At the same time, different colors are used to indicate the status of this indicator. There are four colors: green, yellow, orange and red. The darker the color, the greater the problem

the three major indicators are not captured by vrops, but each indicator has its sub indicators. Several sub indicators jointly determine the status of this major indicator. There are 9 sub indicators in total. Plus the large indicators, there are 13 in total. It is these 12 indicators that jointly determine the value of the object in vrops by Xufei, director of the Department of industry and information technology of the Tibet Autonomous Region

sub indicators related to operation or health include:

load, abnormality and fault (the first three in the figure below)

sub indicators related to risk include:

remaining capacity, remaining time, pressure and compliance

remaining capacity

remaining time

sub indicators related to efficiency include:

recoverable capacity, density

recoverable capacity

three vrops architecture

vrops versions prior to PS6, Its name is vcenteroperationsmanager, or VCoPs for short. The architecture of VCoPs consists of two virtual machines, one for analysis and the other for UI. UI is the interface for us to actually log in to vcopsweb

the following figure shows two virtual machines of VCoPs in vCenter

vrops6 has undergone great changes in its architecture. It no longer uses two virtual machines, analysis and UI, and is merged into one

to ensure high availability of vrops, vrops can be configured in cluster mode. It should be noted that the cluster mode is not necessary. In the non cluster mode, a virtual machine can complete all the work. When installing vrops appliance, vrops virtual designers can easily build a measurement system and construct their own instruments like building blocks. The configuration of front panel computers (CPU, memory, disk) will vary according to the virtualization scale we choose

A cluster with

vrops can have up to four roles and can be configured with four virtual machines, which are:




where master and replica are the primary and standby nodes responsible for the main work of vrops. The information of the two nodes is synchronized. When the master node fails, replica will become the master and continue to provide services. After the master is repaired, the node that becomes the master will change back to replica again


the main work of the data node is to collect and process data. If there is a remotecollector, the data collection is completed by the remotecollector


remotecollector is mainly used to collect data and submit them to datanode

in the test, the author configured a vrops cluster with four nodes:

in the cluster, we can manually rebalance in the cluster (average load distribution):

the problem of four threshold values

we know that the alarm occurs after a certain target of the object triggers the threshold value. For the monitoring and system system, the baseline is very important. Most monitoring tools use static thresholds, while vrops uses dynamic thresholds. So what is the significance of using dynamic thresholds

simply give an example, and you will understand. Take the plot ratio of Beijing Metro Line 13 as an example: many people take the subway in the morning and evening rush hours. At that time, it was basically full. If the plot ratio was used to measure, it was assumed that 90% was normal, which was the baseline of the alarm. In the non peak period at noon, we take the subway, and there may be seats. At this time, it is normal to assume that the plot ratio is 40%. Therefore, if we build a monitoring system to monitor the plot ratio of line 13, the threshold values of morning and evening peak and noon must be different. Otherwise, if 90% is taken as the baseline, the monitoring system will not alarm basically. If 40% is taken as the baseline, the system will alarm frequently

briefly explain the principle of vrops dynamic threshold generation: after vrops is deployed, it will first collect the data of the object for one week, and then take three weeks to correct it, so as to obtain the baseline. Therefore, if customers want to put vrops into production, they must deploy them one month before the formal use and start monitoring the virtualized objects

five customization parts

vrops6's progress over VCoPs is that there are many places that customers can customize. We can customize symptoms, suggestions, alerts, reports, dashboards, etc

in fact, there are many parts that can be customized in vrops, but it should be noted that actions cannot be customized. But we can tie actions and suggestions together

in the following case, a complete record of creating an alarm will be shown. That is:

create two symptoms => create a suggestion and associate the suggestion with an action => create an alert containing the previously created symptoms and suggestions. = => Trigger and view alerts

in fact, the alert I created contains two symptoms: 1 When the CPU of the virtual machine demand is greater than 60%. Meanwhile, mandcpu is greater than 70%. When both symptoms are satisfied, the alarm will be triggered, and the suggestion will be to modify the virtual machine CPU limit and increase the number of virtual machine CPUs

customize a symptom:

it means that when the CPU of the virtual machine demand is greater than 60%, this condition is a symptom. It is called wxymediumcpudemand

after defining it, we find that this symptom will be triggered by a virtual machine

it should be noted that at this time, wxymediumcpudemand is only a symptom, and there is no alarm associated with it. Therefore, no relevant information can be seen in alerts

create a second symptom

create a symptom that is triggered when the hostdemandcpu is greater than 70%

create a suggestion

create a suggestion and associate it with existing actions. Here I have created two suggestions. The second one is to increase the number of CPUs in the virtual machine. The creation method is the same, so I won't paste it again. The latter step actually uses the second suggestion. Readers should not be confused

create an alert

an alert can contain multiple symptoms

object selection virtual machine:

define the level and location of alerts:

select the corresponding symptoms. The symptoms here are the two I created at the beginning:

select the suggestions here, which I created in advance

by this time, the alarm has been created

view alarms

we can see that alarms are triggered and appear under healthy items

in this way, a complete alert is created


vmware China

vmware China R & D center

vmware recruitment

v cloud technology alliance member

loveu lifeu andit

software defined storage and virtualization

yeyurui, VMware storage architect


container and cloud native applications

zhanghaining,the, Chief architect of R & D center

end user computing (chinaeuc)

desktop and application virtualization

wukonghui, VMware architect


infrastructure and application services

Zang Tiejun, VMware architect

David share

sddc and it management

weixinyu, VMware financial industry pre-sales

it review

you Guixian, Senior technical consultant

Copyright © 2011 JIN SHI