zweig@cs.uiuc.edu (Johnny Zweig) (12/21/90)
Being involved in the somewhat narrow domain of protocol processing software,
there are aspects of the Networking Scene(tm) I do not fully understand. Could
someone give me a succinct, nutshell definition of the term "Network
Management"? I mean, I can talk about the innards of TCP/IP all day long, but
I don't know whether NM means expert systems for doing routing, assigning
adresses, running YP, fixing cables, deciding which newsfeeds to use, or some
combination thereof.... It's kind of like some of the New Age terminology
("healing" is a fine example) that seems to be a set of terms I thought I knew
what they meant until they were used in a particular way...
-Johnny Unmanageablemjhammel@Kepler.dell.com (Michael J. Hammel) (12/22/90)
In article <1990Dec20.191404.1681@julius.cs.uiuc.edu>, zweig@cs.uiuc.edu (Johnny Zweig) writes: > Being involved in the somewhat narrow domain of protocol processing software, > there are aspects of the Networking Scene(tm) I do not fully understand. Could > someone give me a succinct, nutshell definition of the term "Network > Management"? My guess is that if they can, its opinion only. I worked on a project at Bell Northern Research (BNR, Inc, R&D arm of Northern Telecom) that was aimed at a sort of any-to-any connectivity system (plug your Goombox computer into the system and it could connect to an IBM mainframe or DEC or whatever). The project died (IMHO due to poor management). But the center of it was based on the ability to "see" the network and bring up or down individual pieces remotely. So if some modem connection was broken you would see it as a red icon in the network and could either re-route traffic or cycle the modem to try to bring it back up. Network management then was being able to see the network and correct "faults". "Faults" is weakly defined though. They were mostly hardware (modems, pc's, cluster controllers, etc) faults. There was little (that I remember) that dealt with wiring or software problems. Michael J. Hammel | mjhammel@{Kepler|socrates}.dell.com Dell Computer Corp. | {73377.3467|76424.3024}@compuserve.com #include <disclaim/std> | zzham@ttuvm1.bitnet | uunet!uudell!feynman!mjhammel #define CUTESAYING "Your cute quote here"
gregpf@momenta (Greg Presedo-Floyd) (12/29/90)
mjhammel@Kepler.dell.com (Michael J. Hammel) writes: >In article <1990Dec20.191404.1681@julius.cs.uiuc.edu>, zweig@cs.uiuc.edu >(Johnny Zweig) writes: >> someone give me a succinct, nutshell definition of the term "Network >> Management"? [...] >But the >center of it was based on the ability to "see" the network and bring up >or down individual pieces remotely. So if some modem connection was >broken you would see it as a red icon in the network and could either >re-route traffic or cycle the modem to try to bring it back up. >Network management then was being able to see the network and correct >"faults". "Faults" is weakly defined though. They were mostly hardware >(modems, pc's, cluster controllers, etc) faults. There was little (that >I remember) that dealt with wiring or software problems. My comments on network managment are from a consumer perspective. A lot of net h/w vendors (HP, NAT in Cupertino, David Systems in Mountian View, Synoptics in MV ) are trying to embed h/w in their components that (a) report various statistics and (b) can be controlled by a management station. There are currently 2 types of management schemes leading the field: S)imple N)etwork M)anagement P)rotocol and ?? (I can't remember) . Some vendors can actually demonstrate "management" from a central location. Many are not quite there yet. To stay abrest, try to goto/get_a_summary of the INTEROP network show that goes on every year. I would like to see "management" evolve into me being able to buy bridges, concentrators, various net h/w and to have 1 station query, control, tune, and shut down that multi-vendor h/w I purchased. There are a lot of aditional subtleties to this topic I will not address now :-) To the brave future of network computing ... -- Greg Presedo-Floyd "Representing myself" Momenta Corporation 415-969-3876 Mountian View, CA 94043
lstowell@pyrnova.pyramid.com (Lon Stowell) (01/03/91)
NONE of the following is to be construed as an official position
or opinion of my current employer...
Short answer....based loosely on OSI/Network Management forum...
Fault Management
Configuration Management
Performance Management
Accounting Management
Security Management
First a word about providers...
There are several Network Management providers....each has
differing views of what Network Management is...and is not.
This takes the broader view...of a more complete package..this
view eliminates Netmaster, etc, as it lacks the broad spectrum
of capabilities of a real network manager...
IBM's Netview (tm) is the granddaddy. Release 3 and later
versions are quite formidable in their capabilities( earlier
versions, IMHO, were glorified VTAM operator facilities..
Netview allows true two-way control and communications...network
elements with electronic identifiers may use Netview's
facilities to provide a full network configuration database of
ALL elements, links, hardware components, software components,
etc. A facility to automate microcode and code updates is
included. A graphic user interface is available.... the full
capabilities of Netview and it's "Common Repository" concept are
too great to detail here...
IBM has excellent documentation, including PC-based self study
tutorials on Netview....see any sales office.
AT&T has the UNMA, the implementing product is Accumaster
Integrator. This product makes somewhat (IMHO) better use of
the graphic user interface than does Netview, and has somewhat
better facilities and flexibility for connection of "foreign"
networks, managers, and facilities. (As opposed to Netview/PC)
Both Netview and Accumaster address OSI, although in somewhat
differing views. The major difference between Netview and
Accumaster is that Netview (personal opinion here) views a
network as a bunch of applications mysteriously glued together
with data links and hardware. Netviews tools tend to reflect
this somewhat--they are formidable at resolving application and
session layer difficulties--not too great at resolving
(accurately) telecom facility problems. Accumaster tends to
view a network as a bunch of telecom facilities with the hosts
just another chunk of hardware interfacing to the telecom
facilities....and the (current) tools provide formidable
facility management and not as elegant applications or session
tools... (views are mine--as a user of both)
Netview has one formidable advantage over competitors,
particularly UNIX based systems... Netview actually emerged
from earlier problem reporting tools which are actually
implemented in the Operating System's kernel. Unix has no
inherent kernel-implemented error logging and reporting facility
which compares with EREP...although it would not be rocket
science to modify the Unix kernel to include same
A dream Net Management package would include a combination of
Accumaster and Netview.
o There is the forthcoming OSI Network Management architecture
which is intended to replace the (largely 1-way) alert and
reporting mechanisms of SNMP based on TCP/IP. This OSI NM
architecture is intended to provide the five basics of
network management:
- Fault Management. This will range from detecting,
reporting, up through help desk, trouble ticket
management, even including fault isolation and recovery.
Transient and persistent faults are addressed.
The ability to filter, enable, and inhibit alarms is
included. Also included is the ability to request
self-testing or QOS testing upon schedule or command.
- Configuration Management. This includes ALL components
of the network... hardware, links, as well as software.
Configuration includes all components AND their status.
There are 4 subcategories..
Network State Information. This includes the network
element connectivity (topology), element operational
status, distribution AND configuration of equipment, and
availability (operational status) of network
resources...applications, ports, links, etc. Also
included would be performance measurements and accounting
info.
Modification of Network Attributes. This allows addition
of new features to an existing network or tuning of
network operations.
Configuration Modification. This allows new network
resources and capabilities to be added, allocation of
resources to network elements and/or users, as well as
dynamic reconfiguration for outages or performance.
Inventory Management. This allows control and management
of the physical pieces of the network (including
software). Facilities are provided to include such
information as type, vendor, location, revision levels,
and functional criticality. This should include
Detailed history such as Type, Serial Number, License
Number, Location, Vendor, etc.
Financial Information on ownership, warranty, contract
status, cross-billing charges, etc.
Historical Information on revision changes, prior
fault reports, location changes, previous recovery
action etc.
Directory Maintenance. This would include all hosts,
applications, network addresses, access priveleges,
security access privileges, service levels, etc.
- Performance Management. Provides tools and functions to
measure and report on the behavior of network
elements,,,hardware, links, software, etc.. This
includes the 3 basic features of response time, rejection
rates, and availability per cent. Also included are
aggregate throughput accounting, user useage accounting,
etc.
- Accounting Management. Provides user-oriented tools to
measure and provide accounting information based on
individual users or groups. This includes the ability to
set and alter accounting limits associated with the use
of network resources.
- Security Management. Provides for the addition,
deletion, and modification of security services
mechanisms and tools. Also provides for the
distribution and management of security related
information as well as the reporting of all security
related events.