LMIWBEM is a Python library that performs CIM operations using CIM-XML protocol. The library tries to mimic PyWBEM, but does things in different way:

• TOG-Pegasus client's library is used for communication
• lazy evaluation of CIM objects is used
• some minor API was added for performance reasons

Reasons why the library was created:

• Performance - Using TOG-Pegasus client's library, Boost::Python and C++ STL, CIM operations are executed faster (e.g. ~60% faster Instance enumeration compared to PyWBEM).

  NOTE: performance gain depends on CIM operation; there are CIM operations where the percentage is even higher.

• Memory - Using PyWBEM, the Python's interpreter can eat up to several GB of memory. In LMIWBEM, C++ allocator for CIM objects is used and unnecessary memory blocks are properly freed and returned to OS. Lazy evaluation (construction) of nested objects helps to perform CIM operations faster and Python does not use additional space for such objects unless it's necessary.

Latest sources of the software can be found at: lmiwbem

LMIWBEM has several dependencies:

• C++ compiler with support of C++98 standard
• tog-pegasus-libs (2.12.0)
• boost-python (1.50.0)
• python-devel
• build tools (autoconf, automake, make, ...)

LMIWBEM supports 2 build systems:

• Autotools (must precede distutils)
• Python's distutils.

To install the module using configure script, follow these steps:

$ ./autogen.sh # if you got sources from git repository
$ mkdir build && cd build
$ ../configure
$ make
# make install

To install the module using Python's distutils, follow these steps:

$ ./autogen.sh # if you got sources from git repository
$ mkdir build && cd build
$ ../configure
# python setup.py install

NOTE: It is necessary to run configure script, which generates Python's setup.py for build and installation.

These options apply for configure and setup.py will be generated based on the options selected:

--with-default-namespace=NAMESPACE; default: root/cimv2
--with-default-trust-store=DIR;     default: /etc/pki/ca-trust/source/anchors/
--with-listener=[yes/no];           default: yes
--with-slp=[yes/no];                default: yes

By default, LMIWBEM is configured to work with Python 2.6.x and 2.7.x. It is possible to build the project with Python 3.x.x compatibility by running configure script with:

--with-python3=yes

If the module is built using distutils, Python version is determined at runtime. Setup with Python 3.x.x:

$ python3 setup.py install

Currently supported intrinsic methods:

• Associators()
• AssociatorNames()
• CreateInstance()
• DeleteInstance()
• EnumerateClassNames()
• EnumerateClasses()
• EnumerateInstanceNames()
• EnumerateInstances()
• ExecQuery()
• GetClass()
• GetInstance()
• InvokeMethod()
• ModifyInstance()
• References()
• ReferenceNames()
• OpenAssociators()
• OpenAssociatorNames()
• OpenEnumerateInstances()
• OpenEnumerateInstanceNames()
• OpenExecQuery()
• OpenReferences()
• OpenReferenceNames()

NOTE: See the directory examples.

There is major difference when compared to PyWBEM: using LMIWBEM, it is possible to create and maintain a connection with a CIMOM:

import lmiwbem

conn = lmiwbem.WBEMConnection()
conn.connect("hostname", "username", "password")
...
conn.disconnect()

It is possible to create connections in PyWBEM's way:

import lmiwbem

conn = lmiwbem.WBEMConnection("hostname", ("username", "password"))
...

NOTE: It is not necessary to call .disconnect() when using such structure.

LMI Meta-Command uses "real" world constructions in LMIShell and was used for benchmarking purposes.

• Managed system:

  • openlmi-account-0.4.2-2.fc21
  • openlmi-logicalfile-0.4.2-2.fc21
  • openlmi-hardware-0.4.2-2.fc21
  • openlmi-networking-0.2.2-3.fc21
  • openlmi-service-0.4.2-2.fc21
  • openlmi-storage-0.7.1-2.fc21

• Client system:

  • openlmi-scripts-account-0.0.1-7.fc20
  • openlmi-scripts-logicalfile-0.0.3-7.fc20
  • openlmi-scripts-hardware-0.0.3-7.fc20
  • openlmi-scripts-networking-0.0.2-7.fc20
  • openlmi-scripts-service-0.1.2-7.fc20
  • openlmi-scripts-storage-0.0.4-7.fc20
  • lmishell with pywbem/lmiwbem

NOTE: LMIWBEM uses Boost::Python, which brings a little overhead with the module itself. This can be seen in memory results.

Report bugs to phatina@redhat.com or lmiwbem issues.