Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

...

  • The machine that recieves traps or sends queries to clients is called the SNMP Management Station.
  • There are 2 methods of exchanging data between the SNMP management station and the SNMP client - TRAP and POLLING (sometimes called discovery)
  • Data that can be retrieved from devices, applications and operating systems varies greatly between vendors but there is some standardization.  Most data is in the form of Attribute/Value Pairs or Tables that are made availble to SNMP by the vendor. 
  • A/V Pairs are defined by the Internet Assigned Numbers Authority (IANA) and are assigned to institutions for private use in what is defined as an ARC generally at an enterprise level.  Rice's defined ARC in OID format is 1.3.6.1.4.1.134 which translates to .iso.org.dod.internet.private.enterprises.Rice.
  • Management Information Base (Mibs) are a heirarchical representation of SNMP data, its data type definitions, OID and syntax.  MIBs provide the OID to Mib map and are provided by vendors or are based on IETF standards for standards based resources such as TCP/IP networking, printing, SMTP email, LDAP and other resources.
  • A open source client suite of software CLI tools is provided called net-snmp that is availbale for Windows, Mac and Linux at http://www.net-snmp.org/download.html
    • snmpget - used to querey hosts for information.
    • snmpwalk - used to traverse the SNMP OID tree that is supported by a client.
    • snmptable  - used to querey host for table information.
    • snmptranslate - used to translate between mibs and oids.

Intelligent Platform Management Interface (IPMI) is a standards based specification initiated by Intel to define a common set of interfaces for monitoring of computer hardware such as CPU, Fans, Power Supplies, Temperatures and other Iron related resources.

  • Similar to SNMP, IPMI if supported on the hardware communicates to a management station but can also be configured to communicate directly via email and SMS to managers.  Each vendor implementation may vary however.  As an example, SUN provides IPMI data via SNMP as well as IPMI directly. 
  • Zabbix supports the polling and trending of IPMI data in the same manner that SNMP is supported.  The only difference 

Where to Begin

  1. Install and/or configure an snmp client on the device that will be used as an SNMP client.  Most hardware and some software come with built in SNMP capabilities.
    1. Linux Installation - "yum install net-snmp" this will download and install all of the snmp client software, you will need to configure it yourself. We suggest a simple configuration with a read only community string of public.  If the machine is public facing, restrict SNMP to the monitoring station access and any equipment that you will test from.
    2. Windows Installations - Installation is from the Add or Remove Software/Hardware section of the Control Panel and choose Windows Components, Management and Monitoring Tools, SNMP with configuration done in Control Panel / Services.
    3. Once installed and configured you can use CLI tools to test your access controls and data.
  2. There are a number of resources used to find out what data is available from an operating system, a hardware component, a network resource or an application.
    1. Use snmpwalk to traverse the entire tree of available A/V Pairs.  "snmpwalk -v2c -cpublic <hostname or IP of client>
    2. Use snmpget to  querey for a specific oid for its value.  "snmpget -v2c -cpublic <hostname or IP of client> 1.3.6.1.4.1
    3. Vendor supplied mibs provide specific OID and ARC information regarding data that can be used for monitoring, trending and notification
    4. MIB repositories Mib Depotand Byte Sphereare online searchable databases where vendor mibs can be analyzed and downloaded.

The SNMP / IPMI Monitoring, Trending and Notification Process

  • Once clients are installed or configured to be accessible to and by the Zabbix server (monitor.rice.edu), the process of identifying, classifying and utilizing data for monitoring, trending and notification can begin. 
  • Begin by answering the following questions
    1. What do I need to know from the server or service(s) that will help me manage the service delivery?
      1. Do I need to track scalability?
      2. Do I want to be able to diagnose problems in the service delivery?
      3. What characterizes the service hardware and software requirements (disk space/performance, CPU cycles, Networking, memory)?
      4. How should the service or server be classified with regards to support (critical, business day only support, non-critical service)?
      5. When and how do I want to be notified if there is a problem?
      6. What should constitute a problem?
      7. Will baseline service data need to be gathered and if so, what data?
    • Determine if there is a way to get the data you need from the Zabbix agent, SNMP, IPMI or custom checks
    • Configure Zabbix to poll or trap data required to monitor, trend or send alerts.
    • Define the trigger thresholds that will satisfy your needs.
    • Create actions that fulfill your notification requirements
    • Test the notifications, graphs and track over time for tuning
    • Rinse and repeat as necessary.
    • Define the IT Service monitoring that you would like and include maintenance windows and KPI