It's not very comprehensive right now, but we hope to flush it out real soon now.
The SNMP specificatins define two things: a protocol and a data
The SNMP protocol defines a few simple operations: GET, GET-NEXT, SET and TRAP. Versions 2c and 3 define a few more: GET-BULK and INFORM. These are the requests that are sent between clients and servers.
The data definition language is the Structure of Management Information, or SMI. It is based on ASN.1, and is used to create text documents that define Managed Information Bases, or MIBs. A MIB defines a set of data objects and the format for those objects. Each data object is identified by a unique Object Identifier, or OID.
An OID is a
series of integers that uniquely identify a data object. OIDs can
be represented in two ways. The first is using raw numbers, like so:
While the raw numbers are easy for a computer to understand, it is
a little harder for us humans. A MIB can be used to convert the
number to the names used in the MIB. The above OID translates to:
A PDU is simply a data packet. The SNMP specifications define the format for each supported PDU type. SNMP PDUs contain various header fields (which depend on the protocol version being used and the PDU type), and variable bindings (varbinds).
A Variable binding, or varbind, is an OID for an object, and the data value associated with that object.
A GET request PDU ...
A notification is an unsolicted message from a SNMP agent or device that is sent to a notification receiver (aka Network Managment System (NMS) or Element Management System (EMS)). Notifications generally indicate a status change on the device. It could be information about the change, or a message to indicate an alarm condition. SNMP defines two types of notifications: traps and informs.
A SNMP trap is an unsolicited message sent to a notification receiver. Unlike messages sent to a SNMP agent, traps are not confirmed by the receiver. If it is critical that a Network Management System (NMS) be able to detect when a message is lost, then the device must provide a method for detecting lost messages, such as a log table or sequence numbers. If the device supports informs, critical notifications should be sent using informs.
A SNMP inform is an unsolicited message sent to a notification receiver. It was not introduced into the protocol until SNMP version 2c, so it is not availabe in devices that only support SNMP version 1. Unlike SNMP traps, informs are confirmed messages, and thus are more appropriate for critical notifications.
A SNMP subagent is a process which uses a protocol other than SNMP to communicate with a master agent. The subagent does not directly communicate with SNMP client, instead it relies on the master agent to handle the protocol details of SNMP. One of the primary advantages of subagents is that they can be added or removed dynamically, and new agents can be developed with any changes to the master agent. The two most commonly used protocols for subagents are AgentX and SMUX.
UDP, or User Datagram Protocol, is a network packet protocol, defined in RFC 768. UDP is a connectionless protocol, which means that the packet is sent, and no confirmation is received from the other side (like sending a letter in the mail). This makes it less resource intensive, but also less reliable, than TCP connections.
TCP, or Transmission Control Protocol, is a network packet protocol, defined in RFC 793. TCP is a connection oriented protocol, which means that the sender and receiver communicate to try and ensure the packet is delivered (like sending a fax). This makes it more resource intensive, but also more reliable than UDP packets.
Because the raw data is unsorted, to satisfy a particular request, the entire data set must be examined to find the apropriate index.
Last modified: Thursday, 26-May-2011 23:21:32 UTC
For questions regarding web content and site functionality, please write to the net-snmp-users mail list.