[kea-dev] Kea python hook?
David Hankins
dhankins at twitter.com
Mon Jan 22 17:56:30 UTC 2018
On Sat, Jan 20, 2018 at 2:13 AM, Christian Kratzer <ck at cksoft.de> wrote:
> Hi,
>
> On Fri, 19 Jan 2018, David Hankins wrote:
>
>> I'm looking at implementing this quarter a competing implementation of;
>>
>> https://lists.isc.org/pipermail/kea-dev/2017-December/000840.html
>>
>> Basically I'm after the same thing; I want ISC Kea to perform DHCPv4 and
>> DHCPv6 protocol implementations, and I just want to provide an interface
>> to
>> internal backend systems curating configuration intent in Python runtime.
>>
>> I was about to send an email to kea-dev announcing my intentions, but I
>> see there's already been some work in this area.
>>
>> Where's it at, what's the current status, and how can we best move forward
>> together?
>>
>
> instead of putting effort into implementing yet another binding for a
> language
> why not implement the hooks as configurable rest calls and then have
> something
> like python/flask to connect to or whatever else you prefer.
>
>
This is an excellent question that deserves some discussion; I think such
a hook would have merit in some deployment environments. I could see a
REST interface having traction in high-latency-tolerant and low-traffic
deployments, but I would prefer myself to use a gRPC or Thrift interface
with integrated service discovery (something like BNS or ZooKeeper).
There is always a "strange attractor" that software engineers analyze when
asking and answering the question, "in our micro-service architecture,
should we draw the box around these functions, or these other functions?"
Integrating a function inside one instance of a multi-service architecture
versus breaking it out into a separate micro service.
The tradeoffs are increased dependencies and potentially latency versus
service simplicity.
In this case, I think there are use cases for a native hook for which the
added dependency and latency aren't worthwhile tradeoffs. For example,
where all elements of configuration for a given host can be arrived at
algorithmically (mathematically) from packet contents alone. I think there
are also cases where host configuration is itself a seriously complicated
question, involving additional backend queries, and a micro-service
approach is appropriate.
> I was planning on on doing that to the previously announced hook library.
> Priorities
> are on other projects atm but that would be my way of proceding once I
> migrate
> my customer from classic isc dhcp to kea.
>
> Greetings
> Christian
>
> --
> Christian Kratzer CK Software GmbH
> Email: ck at cksoft.de Wildberger Weg 24/2
> Phone: +49 7032 893 997 - 0 D-71126 Gaeufelden
> Fax: +49 7032 893 997 - 9 HRB 245288, Amtsgericht Stuttgart
> Mobile: +49 171 1947 843 Geschaeftsfuehrer: Christian Kratzer
> Web: http://www.cksoft.de/
--
Infrastructure Management Services
Twitter, Inc.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.isc.org/pipermail/kea-dev/attachments/20180122/5a3c8c9e/attachment.html>
More information about the kea-dev
mailing list