Author Topic: Ubuntu as Zentyal Client  (Read 5374 times)

christian

  • Guest
Re: Ubuntu as Zentyal Client
« Reply #30 on: August 09, 2011, 09:29:13 am »
when i read the proposal to intergrate a dms  a lot of people was calling for.... dont bloat the server it will get to slow  etc etc etc.. 

I don't think you got such feedback...  ???
I mean "slow" is only matter of hardware, memory sizing and stuff like this. Nowadays, you can add almost as much CPU and/or memory as you need for almost peanuts.
Performance is definitely not an issue yet.

The real drawback while integrating additional components is that it increases complexity at the cost of less manageability (because of more constraint) and less reliability that is the key criteria for such box, far ahead hundreds of potential features, at least from my standpoint.

Furthermore, most of you guys are asking for products or technical implementations (in term of component, protocol...) while we should start with feature we want to cover.
This ends up with strange long thread mixing DMS, CMS and terminal server within same topic... Confusing isn't it?  :-[

So I will never tell developers "what to code" neither ask up front Zentyal to implement "product" but rather think about adding feature covering specific need.
Then up to Zentyal team to select component fitting best with their platform and develop, if needed, specific code.

This doesn't prevent to discuss low level technical stuff (e.g. I would like lighter HTTP server  ;) ) but this is not the same "dimension"  :)

ichat

  • Zen Hero
  • *****
  • Posts: 795
  • Karma: +28/-16
  • RTFM!
    • View Profile
Re: Ubuntu as Zentyal Client
« Reply #31 on: August 09, 2011, 10:38:24 am »
christian  - though i can emagine your view point, im not completely agreaing...

lets first set 2 type of users...

1st:  end users (this can be end users, or unskilled system administrators).
with these types of users  i would most definetly agrea, 

but there are also   power users...  these users ar allways trying to implement stuf in a non standard way...
these are the guy's who  would install  joomla on  a server and make it talk with  ldap...  these are the guys creating nifty winshell scripts to fix a requirement not met by zentyal yet... 

the problem with these kind of users is often that they can, or maybe even  allready did  fix a certain problem.. but they just dont because its to mutch work..    what i want is to find these users in the crowd and encourrage them to help me  creating documentation.. building quick fixes that could be later  intergrated into zentyal..   

take ltps for example...  we all seem to want it...   but no-one as of yet has posted a tread like...
"hey so if we go for ltps,  maybe we should  build it  based on....    %desktop%   $browser$  and  set_office-client  -   %package% could be configured to do  $action$  thus creating   &result;   

you might be reluctant to  tell programmers what to do  or not to do...  but here is the deal... if you dont
they'd have to start from scratch and it will take  %time%  to organise test and practice software... this is tast that anyone could have done...   

if you do five of these projects changes are  1 or 2 (or maybe 4)  will never be implemented for  reasons only known to the zentyal staff..  or maybe some other forum member had better luck with a different app..
but if you document the steps you took to set it up...  there will allways be so many people benefitting.
 
All tips hints and advices are based on my personal experience.
As I try my best to be as accurate as possible, following my advice is always at your own risk,
I claim absolutely NO responsibility in any way!

Sam Graf

  • Guest
Re: Ubuntu as Zentyal Client
« Reply #32 on: August 09, 2011, 02:49:48 pm »
... "slow" is only matter of hardware, memory sizing and stuff like this. Nowadays, you can add almost as much CPU and/or memory as you need for almost peanuts.
Performance is definitely not an issue yet.
I agree completely with you're point. I do think we need to keep in mind that some segment of the community runs Zentyal on older server hardware or on PCs, perhaps by necessity. That said, you're right; it's true that Zentyal runs easily within the resources of low-cost commodity servers.

The real drawback while integrating additional components is that it increases complexity at the cost of less manageability (because of more constraint) and less reliability that is the key criteria for such box, far ahead hundreds of potential features, at least from my standpoint.
This also is my chief concern. I think of the work involved in moving Zentyal to the next Ubuntu LTS release, for example. And the more complex and diverse the package set alone, the more likely we are to run into various kinds of problems for everybody. We already have run into that sort of problem (eGroupWare), so this isn't hypothetical even today. Such things are inevitable, but let's not compound the problem unnecessarily, I think, or get in over our heads.

the problem with these kind of users is often that they can, or maybe even  allready did  fix a certain problem.. but they just dont because its to mutch work..    what i want is to find these users in the crowd and encourrage them to help me  creating documentation.. building quick fixes that could be later  intergrated into zentyal..

... if you do five of these projects changes are  1 or 2 (or maybe 4)  will never be implemented for  reasons only known to the zentyal staff..  or maybe some other forum member had better luck with a different app..
but if you document the steps you took to set it up...  there will allways be so many people benefitting.
I think this is true. It would be nice to get more coders and system administrators interested in contributing to Zentyal. Everybody would win, it seems like. Zentyal is almost certainly a better product today because of past contributions by knowledgeable people, I think.

stuartiannaylor

  • Guest
Re: Ubuntu as Zentyal Client
« Reply #33 on: August 09, 2011, 03:01:56 pm »
I rather like the way the thread has gone. It demonstrates widely varying idea's on what a Zentyal server should be.

I think the UPS situation is sorted and disaster recovery is stuck slightly as the solution is there through subscriptions.

I also think that the combination of Zarafa, Alfresco and LTSP makes Zentyal a much better alternative to MS SBS.

The technologies to provide a CMS are already there and Joomla is a simple, Wordpress also and Drupal. They don't need to be intergrated in a similar way that the above do. I have been working with Joomla for a while and will create some documentation on the easiest way to install. If anybody would like to contribute then I would be happy to grab a brain.
Joomla can authenticate against an LDAP server as my current sites do. I just haven't worked out how to create users from Joomla and that is on my wish list.
With Apache, MySQL, PHP5, LDAP and the various libs already included most web applications will run with very little admin work.
It would be nice to be able to have Forum sub catagories for web applications / (LAMP) style admin, Zarafa and exchange replacements, Alfresco and sharepoint replacements and LTSP and terminal server replacements.
My server could run all these with no problem and cost £700 in fact the server isn't my problem. I have twenty clients that cost £20,000 which is my main fiscal problem.
I have mentioned this before as I like the way the Joomla repository works and I believe Zentyal could work in a similar way.
All Zentyal functionality is modular which allows us to choose how we load the server and what we run on a single box.
It is a falacy that this will make Zentyal cumbersome or unmanagable.
I would like to see a domain subscription system where a server can be registered with a core selection of services.
Then provide a cost for each module and a years subscription on that module. This allows the current blanket subcription to be split so that users with small narrow requirements can have small payments.
It gives the user a cost effective way to choose how he wants to run the server and pay.
This would also give Zentyal more feedback to what modules are worth supporting as payment would indicate this.
So basically I am saying I want more from Zentyal in choice. I want to be able to cherry pick what services I provide and pay a relatively small subscription charge on this.
Many will have relatively simple requirements and may only pay a couple of euro's. In my case I want a large selection and if this increases say what is a current subcription by 50% this is still excellent value for money.

Anyway this thread has been excellently diverse. Think it is about time to split it to Community headings of, Zarafa, Webservices, DMS, terminal services and possibly more options on subscription and module purchase.

Stuart  ;D

Sam Graf

  • Guest
Re: Ubuntu as Zentyal Client
« Reply #34 on: August 09, 2011, 03:42:21 pm »
All Zentyal functionality is modular which allows us to choose how we load the server and what we run on a single box.
It is a falacy that this will make Zentyal cumbersome or unmanagable.
Zentyal is modular but it is a managed, integrated solution. Just because a jigsaw puzzle is modular doesn't mean we can add and remove pieces at will and still have things come out right in the end. :)

We can agree to disagree on this part; I find Christian's argument pretty solid not only from a hypothetical point of view but also from my experience:

Without even getting into the idea of what a cumbersome Zentyal might look like, consider just the Quality Assured Updates service offering and its purpose. If the Ubuntu repositories inject some risk into maintaining a managed solution like Zentyal (and they do, because they have), then it seems to follow that the wider the range of modules (and their underlying packages and dependencies), the greater the potential risk to the whole. It can be quite serious for everybody if updates introduce regressions or conflicts, especially if they isolate the machine.

Then consider the work involved in maintaining Quality Assured Updates if they grow only in simple number and not in interactive complexity. I could go on, but I'm just trying to illustrate reasons why I think Christian's argument makes sense. This isn't just a geek-ish matter, but also a business and productivity matter. Obviously, that's just my own take on things and really means nothing to anybody but me--and those to whom I have to answer at work. :)

stuartiannaylor

  • Guest
Re: Ubuntu as Zentyal Client
« Reply #35 on: August 09, 2011, 04:22:34 pm »
Apache
PHP
Zarafa
Squid
Snort
Dans guardian
Postgres
Dovecot
Bind
Asterisk
Roundcube
samba
redis
z-push
ClamAV
Layer-7
OpenLdap
xjabberd

Just a list of current applications with dependancies from memory as there are lots more.
Your argument just doesn't make sense to me as zentyal is already complex but still works and is quality assured. Alfresco and LTSP are not going to make much more impact. Alfresco practically zero additional with LTSP prob needing more work.
The advantages of providing those two items make Zentyal a rounded small business server that can take on MSSBS.

I understand what you say about complexity but fail to see how you don't feel that it is already complex. Why do you not call for the removal of Zarafa for example?

 

christian

  • Guest
Re: Ubuntu as Zentyal Client
« Reply #36 on: August 09, 2011, 04:55:07 pm »
the problem with these kind of users is often that they can, or maybe even  already did  fix a certain problem.. but they just don't because its to much work..    .../...but if you document the steps you took to set it up...  there will allays be so many people benefiting.
I think this is true. It would be nice to get more coders and system administrators interested in contributing to Zentyal. Everybody would win, it seems like.
I'm convinced we all share this  8)  which means to promote and support initiatives from community. Doesn't mean to focus "only" on "adding one more component to Zentyal to extend the scope"

My experience, as IT guy having worked or still even working on rather large, world wide, deployment it that:
- the simpler, the more reliable
- do not club together "by design" services that are not tightly linked.

Zentyal must be somewhat different because target is SMB, therefore it must be capable to run on one single platform.
Try to think about this in term of corner stone. It has to be strong and reliable. Then if you can use it to build your customized house because it's also flexible, it doesn't mean that all houses must be the same neither that all components used to build your neighbour's house must come with Zentyal box the day you want to build your  ;)

This is where community can help, documenting "howto" to add DMS, CMS or whatever "on top" of Zentyal. This should really help Zentyal team to make the right decision.
In the meantime, they can improve what exists  ;)  we are not yet at the stage were every thing is rock stable and easy to deploy and implement, even if it's already great.

I will definitely open a new thread in the features request section to discuss this and not jeopardize any more this topic  :-[ because I'm confused with the way "SMB" market is perceived. I'm not saying that my view is the correct one (I work only for large to very large companies since years so your view is for better than mine) however this willingness to achieve the "all-in-one but simple and secure" holy-grail without any border makes me very uncomfortable. Let's discuss it elsewhere in the right section  :)

Sam Graf

  • Guest
Re: Ubuntu as Zentyal Client
« Reply #37 on: August 09, 2011, 05:05:37 pm »
Your argument just doesn't make sense to me as zentyal is already complex but still works and is quality assured....

I understand what you say about complexity but fail to see how you don't feel that it is already complex. Why do you not call for the removal of Zarafa for example?

In the past, I have called for a product with less emphasis on "Office" capability :) . In practice, I eventually ran eGroupWare separate from eBox, because it was too hard for me to properly maintain within eBox (hard to recover from disaster, and Ubuntu was a little too far behind the eGroupWare version curve).

The community version of updates isn't quality assured. It was a DansGuardian update that left several people with an eBox install that could no longer connect to the Internet, if I recall correctly. That's the benefit and value of the quality assured updates, but those are available via subscription only and would have to be maintained separately.

Zentyal is complex, but would you not agree that in a managed server product, there is a "must have" level of complexity and a "would be nice to have" level of complexity? And again, I'm not asking for simplicity, but reliability and sustainability. Those two things, and those two things only. If we can have those two things along with the kitchen sink, I'm all set. :D

Let's discuss it elsewhere in the right section  :)
Quite right.