On 03/19/2014 02:00 AM, Stan Lagun wrote:
Steven,
Agree with your opinion on HOT expansion. I see that inclusion of
imperative workflows and ALM would require major Heat redesign and
probably would be impossible without loosing compatibility with
previous HOT syntax. It would blur Heat mission, confuse current users
and rise a lot of questions what should and what should not be in
Heat. Thats why we chose to built a system on top of Heat rather then
expending HOT.
+1
Now I would like to clarify why have we chosen imperative approach
with DSL.
You see a DSL as an alternative to HOT but it is not. DSL is
alternative to Python-encoded
Not accurate. I see a HOT as one type of DSL. I see the MuranoPL as an
imperative language which is not a DSL. It may have domain specific
aspects, but it is more general purpose in nature.
resources in Heat (heat/engine/resources/*.py). Imagine how Heat would
look like if you let untrusted users to upload Python plugins to Heat
engine and load them on the fly. Heat resources are written in Python
which is imperative language. So that MuranoPL for the same reason.
agree this is bad idea.
We want application authors to be able to express application
deployment and maintenance logic of any complexity. This may involve
communication with 3rd party REST services (APIs of applications being
deployed, external services like DNS server API, application licensing
server API, billing systems, some hardware component APIs etc) and
internal OpenStack services like Trove, Sahara, Marconi and others
including those that are not incubated yet and those to come in the
future. You cannot have such things in HOT and when you required to
you need to develop custom resource in Python. Independence on custom
plugins is not good for Murano because they cannot be uploaded by end
users and thus he cannot write application definition that can be
imported to/run on any cloud and need to convince cloud administrator
to install his Python plugin (something that is unimaginable in real
life).
I understand. I am not critical of imperative approach for implementing
workflow/ALM. I believe it is mandatory. I hold no opinion on MuranoPL
specifically.
Because DSL is a way to write custom resources (in Heats terminology)
it has to be Turing-complete and have all the characteristics of
general-purpose language. It also has to have domain-specific features
because we cannot expect that DSL users would be as skilled as Heat
developers and could write such resources without knowledge on hosting
engine architecture and internals.
I understand your point of view, but DSLs are declarative in nature. I
think the problem is that the terminology being used is incorrect as
computer science has defined it :) Just because MuranoPL has custom
stuff for interacting with custom resources, doesn't mean its a DSL. I
can draw a parallel between implementing arbitrary length integers in a
general purpose imperative language. Said feature is not a DSL, it is a
language feature.
What is being discussed in the context of murano is a language feature,
rather then a DSL.
HOT DSL is declarative because all the imperative stuff is hardcoded
into Heat engine. Thus all is left for HOT is to define "state of the
world" - desired outcome. That is analogous to Object Model in Murano
(see [1]). It is Object Model that can be compared to HOT, not DSL. As
you can see it not more complex than HOT. Object Model is what
end-user produces in Murano. And he event don't need to write it cause
it can be composed in UI.
cool
Now because DSL provides not only a way to write sandboxed isolated
code but also a lot of declarations (classes, properties, parameters,
inheritance and contracts) that are mostly not present in Python we
don't need Parameters or Output sections in Object Model because all
of this can be inferred from resource (classes) DSL declaration.
Another consequence is that most of the things that can be written
wrong in HOT can be verified on client side by validating classes'
contracts without trying to deploy the stack and then go through error
log debugging. Because all resources' attributes types their
constraints are known in advance (note that resource attribute may be
a reference to another resource with constraints on that reference
like "I want any (regular, Galera etc) MySQL implementation") UI knows
how to correctly compose the environment and can point out your
mistakes at design time. This is similar to how statically typed
languages like C++/Java can do a lot of validation at compile time
rather then in runtime as in Python.
Personally I would love to see many of this features in HOT. What is
your vision on this? What of the mentioned above can be contributed to
Heat? We definitely would like to integrate more with HOT and
eliminate all duplications between projects. I think that Murano and
Heat are complimentary products that can effectively coexist. Murano
provides access to all HOT features and relies on Heat for most of its
activities. I believe that we need to find an optimal way to integrate
Heat, Murano, Mistral, Solum, Heater, TOSCA, do some integration
between ex-Thermal and Murano Dashboard, be united regarding Glance
usage for metadata and so on. We are okay with throwing MuranoPL out
if the issues it solves would be addressed by HOT.
I am not a fan of language features such as inheritance, classes,
properties, etc. I get that for a really general purpose language like
python they are useful. Python has a multi-year learning curve before
writing really "pythonic" code. Hot has a few day learning curve before
writing really "hotonic" code :)
If you have a vision on how HOT can address the same domain MuranoPL
does or any plans for such features in upcoming Heat releases I would
ask you to share it.
[1] https://wiki.openstack.org/wiki/Murano/DSL/Blueprint#Object_model
I don't think HOT can do these things and I don't think we want HOT to
do these things. I am ok with that, since I don't see the pushback on
having two languages for two different things in OpenStack. I got from
gokrove on iRC today that the rationale for the pushback was the TC
wanted Murano folks to explore how to integrate better with Heat and
possibly the orchestration program. I don't see HOT as a place where
there is an opportunity for scope expansion. I see instead Murano
creating HOT blobs and feeding them to Heat.
Regards
-steve
On Wed, Mar 19, 2014 at 8:06 AM, Steven Dake <sd...@redhat.com
<mailto:sd...@redhat.com>> wrote:
Ruslan,
Some of my thoughts on the evolution of the HOT DSL to date.
On 03/18/2014 05:32 PM, Ruslan Kamaldinov wrote:
Here is my 2 cents:
I personally think that evolving Heat/HOT to what Murano needs
for it's use
cases is the best way to make PaaS layer of OpenStack to look
and feel as a
complete and fully integrated solution.
Standardising these things in a project like TOSCA is another
direction we all
should follow. I think that TOSCA is the place where
developers (like us),
application developers and enterprises can collaborate to
produce a common
standard for application lifecycle management in the clouds.
But before Murano contributors jump into direction of
extending HOT to the goal
of application (or system) lifecycle management, we need an
agreement that this
is the right direction for Heat/HOT/DSL and the Orchestration
program. There are
a lot of use cases that current HOT doesn't seem to be the
right tool to solve.
As it was said before, it's not a problem to collaborate on
extending it those
use cases. I'm just unsure if Heat team would like these use
cases to be solved
with Heat/HOT/DSL. For instance:
- definition of an application which is already exposed via
REST API. Think of
something like Sahara (ex. Savanna) or Trove developed
in-house for internal
company needs. app publishers wouldn't be happy if they'll
be forced to
develop a new resource for Heat
- definition of billing rules for an application
If everyone agrees that this is the direction we all should
follow, that we
should expand HOT/DSL to that scope, that HOT should be the
answer on "can you
express it?", then awesome - we can start speaking about
implementation details.
If it's not the direction these projects should follow then at
least finding
where Heat ends and Murano starts to avoid any functionality
duplication would
be great.
The HOT DSL for the most part, either by design or subconscious
development choices, enables the application of Miller's Law[1] in
a positive way. HOT as a DSL takes less then a few hours to learn
and use effectively. Its relative simplicity is its *key*
advantage as a DSL. DSL's by their very nature declare a desired
state. It is the responsibility of the DSL processor to convert
that desired state into reality. On a fundamental level, this is
precisely what Heat does.
A DSL by its very definition is meant to express a desired outcome
without specifying the intermediate steps. To express the
intermediate steps would require recording state in variables and
offering conditional operations on those variables. This implies
individual steps in the processing of the input to the language.
If HOT were to add these sorts of features, it would no longer be
a DSL, but a general purpose language (perhaps less general
purpose then python or C). A DSL is by definition a declarative
language. I don't like the idea of expanding the scope of HOT to
add an imperative model of operation.
Learning imperative languages takes inordinately more time and
brainpower then learning declarative languages, especially those
which generally follow the advantages provided by languages
operating inside the constraints of Miller's Law. We want Heat to
be dead simple to explain and learn. Realistically I'd like folks
to be able to write a template in under an hour with 15 minutes of
explanation, and I think we have hit that mark.
The idea of expanding the scope of the Heat APIs and engine to
include ALM and Workflow don't make sense to me from an
engineering perspective. It over-complicates the code base. I
know we have already covered those thoughts in detail on the
mailing list previously and the Murano folks agree that is a bad idea.
I see a parallel between expanding the scope of HOT to support ALM
and Workflow and expanding the scope of the heat-engine in the
same fashion that is not appealing. What would make more sense is
to follow the general laws of Unix (do one thing, do it well) and
layer these other possibly imperative languages on top of Heat
using HOT and the Heat APIs to implement such imperative
programming models. Then if someone really wanted to invest in the
complexity of ALM or Workflow, they may be more willing to invest
in learning the complexity of a new imperative programming language.
My personal opinion is expanding the scope of HOT to include
imperative programming models is not desirable for Heat in
isolation. I understand such an outcome may be appealing as a
holistic approach to handling the entire orchestration space, but
feel the costs of learning an imperative model for HOT do not pay
for the advantages of having only one language to program all the
things.
I see no issue with HOT remaining simple and tidy focused entirely
on orchestration (taking a desired state and converting that into
reality) with some other imperative language layered on top to
handle workflow and ALM. I believe this separation of concerns is
best for OpenStack and should be the preferred development path.
Regards,
-steve
[1]
http://en.wikipedia.org/wiki/The_Magical_Number_Seven,_Plus_or_Minus_Two
Thanks,
Ruslan
On Wed, Mar 19, 2014 at 2:07 AM, Keith Bray
<keith.b...@rackspace.com <mailto:keith.b...@rackspace.com>>
wrote:
Georgy,
In consideration of the "can you express it" instead of
the "who will
generate it," I see Heat's HOT evolving to support the
expression of complex
multi-tier architectures and applications (I would argue
you can already do
this today, perhaps with some additional features desired,
e.g. Ability to
define cloud workflows and workflow execution rules which
could come when we
have a workflow service like Mistral). Therefore, I would
encourage Murano
contributors to consider whether they can help make Heat
sufficiently cover
desired use cases. I have never viewed Heat templates as
isolated
components of a multi-tier architecture. Instead, a
single template or a
combination of master/subordinate templates together
(using references,
nesting, or inclusion) could express the complete
architecture, both
infrastructure and applications.
If I've read your previous comments and threads correctly,
you desire a way
to express System Lifecycle Management across multiple
related applications
or components, whereby you view the System as a grouping
of independently
developed and/or deployed (but systematically related)
"components," whereby
you view Components as individual disconnected Heat
templates that
independently describe different application stacks of the
System. Did I
get that correct? If so, perhaps the discussion here is
one of "scope" of
what can or should be expressed in a Heat template. Is it
correct to state
that your argument is that a separate system (such as
Murano) should be used
to express System Lifecycle Management as I've defined it
here? If so, why
could we not use the Heat DSL to also define the System?
The System
definition could be logically separated out into its own
text file... But,
we'd have a common DSL syntax and semantics for both lower
level and higher
level component interaction (a building block effect of
sorts).
As for "who will generate it," ( with "it" being the Heat
multi-tier
application/infrastructure definition) I think that
question will go through
a lot more evolution and could be any number of sources:
e.g. Solum, Murano,
Horizon, Template Author with a text editor, etc.
Basically, I'm a +1 for as few DSLs as possible. I support
the position that
we should evolve HOT if needed vs. having two separate
DSLs that are both
related to expressing application and infrastructure
semantics.
Workflow is quite interesting ... Should we be able to
express imperative
workflow semantics in HOT? Or, should we only be able to
declare workflow
configurations that get configured in a service like
Mistral whereby
Mistral's execution of a workflow may need to invoke Heat
hooks or Stack
Updates? Or, some other solution?
I look forward to a design discussion on all this at the
summit... This is
fun stuff to think about!
-Keith
From: Georgy Okrokvertskhov <gokrokvertsk...@mirantis.com
<mailto:gokrokvertsk...@mirantis.com>>
Reply-To: "OpenStack Development Mailing List (not for
usage questions)"
<openstack-dev@lists.openstack.org
<mailto:openstack-dev@lists.openstack.org>>
Date: Tuesday, March 18, 2014 1:49 PM
To: "OpenStack Development Mailing List (not for usage
questions)"
<openstack-dev@lists.openstack.org
<mailto:openstack-dev@lists.openstack.org>>
Subject: Re: [openstack-dev] [Murano][Heat] MuranoPL
questions?
I see this in the following way - who will generate HOT
template for my
complex multi-tier applications when I have only templates
for components?
_______________________________________________
OpenStack-dev mailing list
OpenStack-dev@lists.openstack.org
<mailto:OpenStack-dev@lists.openstack.org>
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
_______________________________________________
OpenStack-dev mailing list
OpenStack-dev@lists.openstack.org
<mailto:OpenStack-dev@lists.openstack.org>
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
_______________________________________________
OpenStack-dev mailing list
OpenStack-dev@lists.openstack.org
<mailto:OpenStack-dev@lists.openstack.org>
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
--
Sincerely yours
Stanislav (Stan) Lagun
Senior Developer
Mirantis
35b/3, Vorontsovskaya St.
Moscow, Russia
Skype: stanlagun
www.mirantis.com <http://www.mirantis.com/>
sla...@mirantis.com <mailto:sla...@mirantis.com>
_______________________________________________
OpenStack-dev mailing list
OpenStack-dev@lists.openstack.org
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
_______________________________________________
OpenStack-dev mailing list
OpenStack-dev@lists.openstack.org
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
