Network World just did an article on network management of IPv6 in which Netcordia was mentioned. (Where eight network management vendors stand on IPv6)
Of course, a short interview doesn’t provide much opportunity to say
much about a reasonably complex topic, so I’ll add some other thoughts
here. Also look back in this blog for an entry I wrote on “IPv6
Security”.
Another vendor and I both noted that displaying 128 bit addresses in
a GUI will be a problem. I anticipate that it will not be a major
problem, but one that will be like a small pebble in your shoe.
Different vendors will likely settle on different display methods, plus
the size of the field will consume big chunks of screen real-estate. As
several of us noted, using DNS will help.
Databases will grow because all the address fields will increase in
size. Expect to see bugs around failures in database functions that
don’t properly handle the new field sizes.
Along with IPv6, I see a migration to SNMP v3. While not
specifically related, adding support for SNMP v3 while working on IPv6
looks like a good opportunity. I expect to see a lot of organizations
take the leap to deploy both at the same time. Then we’ll have a fun
time debugging SNMP configurations (”was the config supposed to be
AuthPriv or AuthNoPriv, and do the keys match?”). If you’ve ever worked
on bringing up a number of VPN tunnels, you have a taste for what's
coming. A lot of details must match for it to work and troubleshooting
tools and techniques are often limited.
The real interesting thing that wasn’t mentioned in the
article was equipment vendor support for IPv6. Check the MIBs of the
vendors to see if the address fields have IPv4 and IPv6
representations. If the vendors don’t support the raw data, there's not
much that can be done by the management systems. Unfortunately, most
vendors treat SNMP support as secondary and we’re lucky to see
management support in the first versions of any new functionality.
Another place to look is the data types that the vendor exports in flow
data records (Netflow or sFlow or IPFIX).
As the probe vendors outlined in their responses, troubleshooting
will get a lot more challenging when the data streams are encrypted,
limiting application visibility. When will you know that an
application's poor performance is due to a server internal problem or a
network problem when all you can see is IP datagrams? Seeing TCP
retransmits may be essential to knowing that packets of the specific
application were being discarded (vs packets of another app).
There are going to be some interesting challenges and new techniques to learn for managing IPv6 networks.
-Terry