OpenNMS Update, Volume 4, Issue 6

Gang:

Project Status (2 December 2002):
---------------------------------

Okay, it has been a long time since I have sent out an update. Sorry
for the delay.

OpenNMS 1.1.0 is now out. Although I haven't had much feedback, I am
hoping that you like it. I figure the poller response time data would
be a welcome addition to the product, as are the numerous UI and event
enhancements.

Speaking of events, there is a new How-To, Part 5, available at:

http://www.opennms.org/users/docs/docs/html/part5.html

New Projects and New Maintainers:
---------------------------------

We have a couple of new faces on the OpenNMS team.

Mark Murphy ([email protected]) and his team at Tyrell Software
(http://www.tyrell.com) are working on maintaining OpenNMS on the Apple
OSX operating system.

Earlier this year we had to make the move to the 1.4 Java SDK in order to
overcome some issues with sockets on Linux. Since 1.4 is not widely
available for the OSX platform, Mark and his group have built OpenNMS
with the 1.4 changes removed so that Apple uses can enjoy the product.

There is now also a MacOSX mailing list for discussing OpenNMS on OSX.
(http://lists.opennms.org/mailman/listinfo/macosx)

Also, Tony Simone ([email protected]) has joined the team as the
Debian package maintainer. His goal is to get OpenNMS into Debian "sid"
(development on "woody" has been frozen) but in the meantime packages
can be downloaded from ftp.opennms.org. Since Debian does not have an
official Java 1.4 package either, Tony has built an "installer" package
that will allow the OpenNMS package to be installed.

Bug of the Week:
----------------

There may be some issues with Tomcat, OpenNMS and the 1.4.1 version of the
SDK. People are reporting that everything pretty much dies after a certain
amount of time. It does not appear to be happening on 1.4.0.

For more information, check out Jalon's FAQ at:

http://faq.opennms.org/faq/fom-serve/cache/129.html

Roadmap:
--------

Many people are asking "What's Next?" with OpenNMS, so I thought I'd jot down
the roadmap I have in mind. Note that OpenNMS is a volunteer project, and so
the speed at which the product develops is a function of how many people
contribute to it, or decide to help pay for development of a particular
function. In other words, this roadmap may, and probably will, change.

<soapbox>
Not to get too "sales-y" here, but several of the new features in 1.1
were paid for or inspired by Sortova Consulting customers. Things like the
2750+ new trap definitions and the response time graphs might not have
happened if it weren't for them. I firmly believe that everyone can benefit
from supporting open-source development, especially with money that would
have gone to more expensive proprietary solutions.
</soapbox>

Okay, to review, the OpenNMS version numbering scheme is as follows: stable,
or production quality releases, have an even number after the first decimal.
Thus 1.0, 1.2, 1.4, etc. will all be stable releases.

Unstable, or development, releases, have an odd number after the first
decimal. Thus, 1.1, 1.3, 1.5, etc. will be development releases. Note that
the word "unstable" can be misleading. Development releases have not been
tested as thoroughly as production releases, but they are created to be as
stable as the current production release.

Once a development release has been in the field and tested for a certain
amount of time, it will become the next production release, so in the
following roadmap I have listed the unstable/stable releases in pairs.

Release 1.1/1.2:
----------------

This is the current development release. The plan is to work on a lot of
small features in the month of December and pretty much freeze the code in
the new year. The biggest feature on the books is the ability to associate
SNMP traps with nodes in the system, and there are a couple small features
concerning database fields and time ranges in the display of collected data
(to make OpenNMS a little more MRTG like). 1.2 should be available toward
the end of January.

Release 1.3/1.4:
----------------

This is the release that will really pose a challenge for the commercial
options out there. Outside of features to be voted on by the community, the
following are planned:

a) Path Outage: Currently, OpenNMS will correlate to the node level. If all
the services on an interface are down, the interface is marked down instead
of sending numerous "node lost service" messages. If all interfaces on a node
are down, the node is marked down instead of sending numerous "interface
down" messages.

This feature will test the network path from the OpenNMS system to a given
node that is down, and if the failure is due to a network element, that
failure will be reported instead of numerous node down messages. The system
will also be able to report all of the nodes affected by the outage.

For the first iteration, this will be Layer 3 only. Layer 2 path discovery
is very difficult and tends to involve vendor specific code. Commercial
products that do this tend to run on the order of US$100K for normal-sized
networks, so bear with us as we attempt this for free.  (grin)

b) Availability Report Improvements: While not necessarily easy, it is
currently possible to build fairly custom SNMP (and response time) reports
in OpenNMS. Availability reports, however, have not changed much. I would
like to see this change, and plan to build a similar report interface as
exists with performance reports (such as being able to determine
availability for a specific time range, etc.)

c) Nessus Integration: Nessus (http://www.nessus.org) is a remote security
scanner. It is our goal to be able to control Nessus and manage events it
produces via OpenNMS. Some of the work has already be started in the code
base.

d) Internationalization: If you go to http://www.opennms.org/company/, you
can see that OpenNMS is used in a number of different countries. Not all
of those countries use English as their primary language, so we want to
enable
OpenNMS to display its information in the user's language of choice.

Note that since I speak only English and poor Spanish, this feature will
enable localization, but it will be up to the community to volunteer to do
the translations. Do not take that lightly, as there will probably be over
4000 events by this time, all of which will require translation. And for all
you comedians out there, I will be running random parts of the translation
through Babelfish to insure that you didn't sneak some "dirty" words passed
me. (grin)

The time frame for this is to have a development release out by the end of
March, with a production release available by the end of June.

Release 1.5/1.6:
----------------

a) Remote Pollers: OpenNMS was designed to support remote pollers. This will
be the first step toward a true multi-tiered architecture. Remote machines
will be able to poll services and collect data, and then forward that
information to a central OpenNMS instance.

b) Event Modeling: One of my all-time favorite network management products
has to be NerveCenter. While commercial, NerveCenter lets you build state
models that can change based on polls and traps from the network.

The downside of the product is that those models can become very complex and
hard to maintain. Our plan is to take and code certain models that can then
be configured via an XML file.

For example: flapping events. A "down" event is quickly followed by an "up"
event, and this repeats. In NerveCenter, you could build a model that would
count the number of down/up transitions and if a certain number occurred
within a given amount of time, you could trigger a "flapping event" that
would remain until a certain degree of stability returned.

This could be coded into OpenNMS, and parameters such as the "up" event, the
"down" event, and the number of such transitions that must occur within a
given time could be defined via an XML file.

There are a number of such models that could be built into the system and
then easily configured by the end users.

The time frame for this is to have a development release out by the end of
September, with a production release available by the end of December.

Release 1.9/2.0:
----------------

OpenNMS 2.0 will be a true, multi-tier NMS. I am not sure what we will be
doing after 2.0 - perhaps branching more into asset management and security.

As I mentioned before, this schedule is subject to change and only includes
the "big" features. Other things we'd like to see: a syslog parser, better
outage management, etc. will be spread out across these releases.


Final Notes:
------------

In the United States, we just celebrated the Thanksgiving holiday. This is a
time to look back on the past year and give thanks for what it brought to us.

In the wake of the attacks in New York and Washington, DC on September 11th
of last year, 2002 has not been all that good. The world economy is in a
slump and tensions are rising everywhere. Having traveled to Asia, Europe
and the Middle East, it is hard for me sometimes to reconcile the actions of
governments (including my own) with the friendly people who met me wherever
I went.

But I also started working on OpenNMS on September 10th. Not only do I think
OpenNMS is a great product, but it allows me to work with people all over the
United States and the world without having to deal with politics. I am very
thankful to have the chance to do such work, and I love it when I get a
letter from, say, Vietnam or Croatia, that folks over there are using the
product and they like it. It is one reason I am eager to start working on
internationalization.

So, if you use and like OpenNMS, please drop me a note ([email protected])
and let me know. I'd also like to add you to our "About" page:
http://www.opennms.org/company/. Just send me your company name, country,
and URL.

If you *really* like OpenNMS, you can send us something on the wishlist
(http://www.amazon.com/o/registry/NO8FKWEK8753). (grin)

Seriously, thanks for supporting OpenNMS.

-T

Note: As the end of the year approaches and budgets for next year are being
written, consider checking out the commercial side of OpenNMS at
http://www.sortova.com. Think of it as being the mailing list on steroids:
rapid resolution to your OpenNMS issues.

-- 
Tarus Balog
Consultant
Sortova Consulting Group, http://www.sortova.com
+1-919-696-7625
[email protected]



_______________________________________________
announce mailing list ([email protected])
To subscribe, unsubscribe, or change your list options, go to:
http://lists.opennms.org/mailman/listinfo/announce