The attractiveness of SNMP is its simplicity and relative ease of implementation. With this comes a price: e.g., the more fine grained information that is need or required, such as the variance in interarrival time (jitter) of packets sent to a particular local address, the less likely it is that it will be available.
SNMPv1 uses the underlying User Datagram Protocol (UDP) for data delivery, which does not ensure reliability of data transfer. The loss of data may be a limitation to a network manager, depending on the criticality of the information being gathered and the frequency at which the polling is being performed.
SNMP is best suited for network monitoring and capacity planning. SNMP does
not provide even the basic troubleshooting information that can be obtained
from simple network troubleshooting tools
[Wellens 96]. SNMP agents do not analyze information, they just collect information and provide it to the network management application.
SNMPv1 has minimal security capability. Because SNMPv1 lacks the control of unauthorized access to critical network devices and systems, it may be necessary to restrict the use of SNMP management to non-critical networks. Lack of authentication in SNMPv1 has led many vendors to not include certain commands, thus reducing extensibility and consistency across managed devices. SNMPv2 addresses these security problems but is difficult and expensive to set up and administer (e.g., each MIB must be locally set up).
Vendors often include SNMP agents with their software and public domain
agents are available. Management applications are available from a
variety of vendors as well as the public domain, however they can differ
greatly in terms of functionality, plots and visual displays.
SNMP out-of-the-box can not be used to track information contained in
application/user level protocols (e.g., radar track message, http,
mail). However these might be accomplished through the use of a extensible
(customized) SNMP agent that has user defined MIB.5 It is important to note that a specialized or extensible network manager may be required for use with the customized agents.
There are also concerns about the use of SNMP in the real-time domain where bounded response, deadlines, and priorities are required.
SNMPv2 is intended to be able to coexist with existing SNMPv2, but in order to use SNMPv2 as the SNMP manager or to migrate from SNMPv1 to SNMPv2, all SNMPv1 compliant agents must be entirely replaced with SNMPv2 compliant agents-gateways or bilingual managers and proxy agents were not available to support the gradual migration as of early-1995. Since SNMPv1 and SNMPv2 are incompatible with each other and SNMPv2 is not stable, it is important when procuring a managed device to determine which network management protocol(s) is supported.
Common Management Information Protocol (CMIP) may be a better alternative for large, complex networks or security-critical networks.
CMIP is similar to SNMP and was developed to address SNMP's
shortcomings. However, CMIP takes significantly more system resources than
SNMP, is difficult to program, and is designed to run on the ISO protocol
stack
[X.700 96]. (However, the technology standard used today in most systems is TCP/IP.)
The biggest feature in CMIP is that an agent can perform tasks or trigger
events based upon the value of a variable or a specific condition. For
example, when a computer can not reach its network fileserver for a
predetermined number of times, an event can be generated to notify the
appropriate personnel
[Vallillee 96]. With SNMP, this task would have to be performed by a user, because an SNMP agent does not analyze information.
This technology is classified under the following categories. Select a
category for a list of related topics.
