Multi-site, single-signon topology

[TonyG]

I'm building a portal site with WordPress. From here users will jump to SMF forums. Each forum will be completely separate, with different mods. Is there already a mod or model to support that?

To bring a new forum online under the portal, I'll copy from a template of files and a database, then customize to suit the forum-specific requirements. This seems like a fairly cookie-cutter operation. Is there code to help with that? If not I'll probably create it over time.

There will be one super admin for the entire site but I'd like each forum to have a local administrator who only has access to specific features. Is there a mod where we can specify which admin functions are allowed for specific groups?

The structure I have in mind is that the domain will have WordPress at the root path, subdomains will be used as shortcuts to each forum, like mygroup.domain.tld. The actual SMF files will be under /home/user/forums/category1/group9/smf_files. With this structure I can consistently install under smf_files for every group and then in Administration I can set the URL to the subdomain. I'd like to put everything under /home/user/domain.tld/forums but I've heard that applications like SMF don't do well in the same folder with WordPress. Any comments on that structure?

I intend to keep all forums at the same SMF base but in practice that can be tough, so as you see I'm planning to consume a lot of disk and have a separate install for each forum. Is there a better single installation multi-site solution? I'm looking into mods now but I'd like real-world opinions.

Finally for now - I'm still struggling with an ID mechanism. SMF/WordPress User Profile integration is precarious these days. I'd rather keep them separate. But I'd like someone who is logged in to one SMF subdomain to be able to get into another one with the same credentials - if they've registered there. I'm hoping to find a Single-SignOn mechanism that works across SMF sites. The thing is though, that a single user might have a different display name in different forums. I'm thinking I would have users login with an email address and xref that to a different display name in each forum.

Thanks for your time and consideration.

[margarett]

You can use dougiefresh's Split Forum MOD, it should be able to simplify your setup

[Kindred]

I do have to ask.... Why Wordpress?

There is almost nothing I'm Wordpress that can not be done in smf...   And Wordpress is a HOG for system resources and never plays well with secondary systems.

[Hj Ahmad Rasyid Hj Ismail]

I'd rather suggest PortaMX Subforums mod instead. It is stable and have more features.

[TonyG]

I do have to ask.... Why Wordpress? There is almost nothing I'm Wordpress that can not be done in smf...   And Wordpress is a HOG for system resources and never plays well with secondary systems.

Well most people still think of WordPress as blog software but these days its as much a full-featured CMS. I used to use Drupal but then getting a decent blog in that CMS was tough, so I had Drupal plus WordPress for blog plus SMF for forum plus others. Drupal does a lot of things, badly. SMF is IMO the best forum software available but it's not a CMS at its core. It does content management by the graces of some mods but if those mods should go unmaintained, as many in this area do, then we're suddenly without a CMS, and I don't want to go in that direction. WordPress is now at its core a CMS, so that functionality isn't going to go anywhere. And that's where I need it. Also, as frosting on the cake, WordPress is now dominating worldwide as a CMS/blog. It's popularity is compelling as this increases the likeliness for stability, longer-living mods, etc. The reality is that SMF still only occupies a small segment of the forum world, and it's not widely recognized as a CMS at all. So for all of these reasons combined, I've decided to use this specific hybrid of best platforms.

@ahrasis - I'll look into PortaMX, thanks. I really need stability, and features help as well.

[Kindred]

SMF is not a CMS... but neither is WordPress.

Either one can be used to run a full featured site though.... 
but using the two of them is likely to cause issues.

So - determine what the focus of your site is

If it is a content site, then an actual CMS like Joomla or Drupal is the way to go. Both of those have blogs, forums and other add-ons.
If it is a blog site, focused on on person's thoughts and the readers only respond to that, then go for something like WordPress...
If it is a forum site, focused on community and conversation, then use SMF with a portal and other add-on mods.

[TonyG]

Respectfully, I disagree with a lot of this but I think we'll be more in-sync as you see where I'm focused.

SMF is not a CMS... but neither is WordPress.

A few years ago that was true but today a world of people would disagree that WordPress is not a CMS.

Either one can be used to run a full featured site though.... but using the two of them is likely to cause issues.

And how and why is that? I've been running these together for years without a single interaction.

So - determine what the focus of your site is

Yes, good to do this. However, my site is a portal where users are welcomed and provided with information, and then they are provided with access to multiple forums. So it's not a forum site nor a blog. It's focus is to provide content and to allow registered visitors to contribute. So I'm using best of breed software to accomplish the task, not defining the site by the kind of software I'm using.

If it is a content site, then an actual CMS like Joomla or Drupal is the way to go. Both of those have blogs, forums and other add-ons.

I've already said that I used to use Drupal and it does a lot of things badly. Really, both the blog functionality and forums are horrible via any of the mods. So I added WordPress for blogging and SMF for a forum and then I had to deal with separate logins for each part of the site as well as completely separate themes and L&F. Joomla is popular but doesn't have as much of a following as other solutions discussed here.

If it is a blog site, focused on on person's thoughts and the readers only respond to that, then go for something like WordPress...

Agreed.

If it is a forum site, focused on community and conversation, then use SMF with a portal and other add-on mods.

Agreed here as well.

But for a site that provides different services, I want to use the best software available. If/when I add wiki functionality, I'm not going to use a plugin for WordPress or SMF. I'm going to go with the industry standard WikiMedia. That way I won't get frustrated with one mod supported by perhaps one person who is trying diligently to build in functionality that already exists in a package supported by thousands of other people.

I just wish more effort were invested in trying to get the great FOSS in this world to integrate better. Right now they all have different login mechanisms, different themes, separate searching, and other features that are completely non-integrated. None of the noble individual attempts in this area have been successful. And yet we continue to have forum-oriented people trying to build blog and wiki functionality, blog people building in forum features, and CMS people trying to build in everything and getting very little of it right. There's a lot of effort being spent in (IMHO) the wrong areas, and with all of the failed attempt as examples you'd think some group would organize to help put an end to the futility.

Thanks for your patience.

[Kindred]

except I use SMF for all of that.

forum
blog
wiki
knowledgebase
downloads
gallery

all within SMF...

and WordPress is not a CMS. period. You can disagree all you want... but it's not. It is a blog software which has layout and addons.
I worked for a company that used WordPress to manage its site (mostly because some marketing person thought like you do)...   it was a nightmare and we couldn't get them off the platform fast enough...

If you call Wordpress a CMS, then you would have to call SMF+Portal a CMS as well, since they can do the same thing.

Wordpress does blogs. You can add plugins to have it do other stuff... but it does none of that as well as a dedicated CMS would. And it is not particularly well programmed.
Joomla, on the other hand, is fast, lightweight and well programmed. The add-ons for Joomla also extend it's capabilities, but it was DESIGNED form the ground, up... to manage a full site

as for what issues...   running a bridged wordpress/smf site is prone to MANY issues - not the least of which is variable collisions (although I hope most of those are dealt with at this point) but also runs into issues because wordpress ASSUMES that it is the only thing running on the site and "takes over", sometimes to the effect of actually preventing access to other softwares.

[stmaxx]

If I'm reading into this correctly, I can see off hand, that there may be big issue with the database, unless you are referring to just having separate forum categories, and not forum sites? If you have 2 sites, using the same database all will need to be in sync, the theme, categories in most everything else in most cases.

There may be a way to bridge from an WP site to the SMF, But keep in mind that all things do not get upgraded in sync or at the same time ( portals the WP and Mods ) It appear to me to be  A cluster fx effect.

I tested WP many times and had to take it, it uses many DB tables, just for testing themes, I make for it!

Now there are CMS packages that corporation type sites use, but they are bit harder to manage than both WP and SMF together, unless you have an IT guy. Some of the CMS systems allow use from from different areas (locations) to administrate the main site CMS, or can be broken down into groups. and even conferences.

Now if I'm way out of the park, sorry for that!

Don't get me wrong the SMF ( none Better) and the Portals are great and you can do allot but sounds to me you need, or are looking for some else!

regards,
Maxx

[TonyG]

At its base SMF is a forum. At it's base WordPress was a blog, but that's not the case anymore. There's no sense in debating this as it's quite easy for anyone to do a search on the current status and popularity of this platform, and thus verify the points. It's reasonable to assume that most folks here have had bad experiences in the past with one platform or another, and that good experiences in the past have led to current preferences. We need to make decisions today based on the software available today, and that's what I'm trying to do here. For my purposes I'll assert that I Am using WordPress along with SMF, and looking for the best practices for making them live in harmony with current tools and practices.

I do keep all software in different databases. So I might have a database wp_sitename and smf_sitename. I prefer not to host common software for more than one site in a single database, differentiated with table prefixes - like site1_users and site2_users. I'm sorry if I was ambiguous on these points. I do this so that I can easily move databases to different servers, and so that usage in one site doesn't impact others.

As to running software together, I was also ambiguous about this, and I apologize to @Kindred. I used to install as /home/site/blog and /home/site/forum. That's why I never had issues. But now I install WordPress (remember, as full site CMS, not as blog) under /home/site/ and the forum is in /home/site/forum. So far I have not seen conflicts but I understand that there's potential for .htaccess modifications from WP that might eventually affect SMF. Without a warning about a specific conflict with current releases, I can't assume there will be issues, but if someone here speaks up on this I'll test and re-work the structure if required.

I hope that gets us back to a common base here. I need to do my research on the mods suggested so far.

Thanks all!!

[Arantor]

current status and popularity of this platform

Doesn't make it a *good* platform under the hood.

I do keep all software in different databases.

This will give you headaches if you plan on doing any bridging of any kind.

Without a warning about a specific conflict with current releases, I can't assume there will be issues

There will almost certainly be issues depending on the exact confluence of plugins added to WP, especially the 'security' ones.

I would still dispute the need for WP at all to be honest. There is nothing you can do in WP that cannot be done in SMF - it all comes down to how idiot-proof you want it and how much legwork you need to do to achieve it. Since you're already talking about multi-site (though curiously avoiding the question of multiple subdomains or multiple domains), you immediately have a fair amount of SMF work to deal with and going from there to removing WP from the equation is not a huge amount of extra... especially since you get to avoid having two separate platforms plus all their plugins (which is a security risk, the more code, the more attack surface)

[TonyG]

Thanks for your notes. I guess I'm at the point where I should describe my specific application in order to give you folks an opportunity to make specific recommendations. In short it's a portal which allows visitors to run a private subsite for individual teams of game players. These can be online game guilds or clans, hockey or soccer teams, or even teams for board games.

The site is http://GameChat.Club. That entry portal is in WordPress. The buttons to Create and Enter subsites are being coded now so that page doesn't go anywhere at the moment except to the related social media sites. The "Go to your site" link will go to SMF instances.

An initial SMF installation serving as a prototype and live test environment is http://Extinct.GameChat.Club. (Note @Arantor that I am using subdomains.) Currently, each team site is getting a full file system and a separate database. As mentioned in this thread, this is certainly going to cause headaches as the number of sites grows. The subdomain name must be unique and will be issued on a first-come/first-serve basis (and I will soon offer "vanity" subdomains for a small fee), but the file system path to the SMF instance is designed to avoid conflicts with other teams by the same name.

The styling for that SMF instance is not complete. I've started using the default theme only because that's the one that's most compatible with current mods. I was faced with a choice of limiting mods to offer a choice of more themes, but I opted to limit the themes initially so that I can offer more features.

I won't provide access to the forum here - I don't think it's necessary for this discussion.

Initially I want to create a set of default boards for specific game types and not allow team leaders to create their own boards. I need to see the directions that people are going to go, as well as limit the options for confusion, permissions issues, etc. If this goes anywhere, I'll add more features to give team leaders more control over their subsite. Similarly, as time goes on I'll try to offer more options for team leaders to customize their own subsites, but for now I need to manage all styling for each one. Obviously that's not going to scale well, which is why I'm looking for options for growth.

It's quite likely that individuals will have memberships in multiple SMF instances. In each subsite individuals may identify themselves with different display names and other profile details. So for now, individuals will need a completely separate login for each subsite. Per this thread, I'd like to move toward a single signon where members are simply allowed/denied access to individual SMF instances - while still having forum-specific display names, etc. Yeah, I know ... that might be asking a lot ...

So now you can see where I'm going with this. Except for adding a jump point for team members to get from the portal to their SMF subsite, there's no integration, so no need for me to force SMF into the portal tier. While I do intend to have a SMF instance for the portal itself, I do not want to provide any kind of end-user login access to the portal site.

Comments on the overall model as well as implementation are most welcome. This is the first time I'm discussing this site in public so this is a big deal for me. 

BTW, if anyone wants to join this effort with some sweat-equity, please PM me. Right now this is all non-for-profit but if it goes anywhere then I don't intend for it to stay that way, and I'm going to need some SMF talent to keep it running.

Thanks as always.

[Arantor]

Currently, each team site is getting a full file system and a separate database. As mentioned in this thread, this is certainly going to cause headaches as the number of sites grows.

For what reason, exactly?

while still having forum-specific display names, etc. Yeah, I know ... that might be asking a lot ...

It's doable, just not with any of the solutions that have been discussed.

However I still do not understand the need for multiple physical installations, and the 'compatibility with mods' is going to go to hell very quickly whatever you do, in all honesty, because so much of this is going to require custom code to actually function even remotely well, whatever you do.

[Gwenwyfar]

But for a site that provides different services, I want to use the best software available.

Then you're choosing the wrong one, because in all honesty, WP is one of those things that just suck, but is popular anyways, for whatever reason. For one, why you'd want a software that only limits anything you do and has so many problems just waiting to happen?

Also:

There is nothing you can do in WP that cannot be done in SMF

[Kindred]

Seriously.

Drop wordpress.

Use a single smf installation with simpleportal and dougiefresh's Split forum mod
or
Use a single smf installation with PortaMX and the subforums extension.

either one of those will get you 98+% of the way to what you described...
and a) will be easier to maintain b) more secure c) faster and d) less prone to failures due to conflicting "stuff"

both of the mods listed above handle multiple subdomains (even multiple domains, as long as they are on the same server) and use the SMF permissions system to deal with access to individual boards on individual "sites" - and deal with custom themes per "site"

[TonyG]

About WP quality,  we can all cite examples of how statistics lie, but there are some basic numbers that indicate WP is in wide use, no matter how crappy some folks believe it is. Let me be clear - we're talking about mass worldwide adoption and daily use, full-stop. We can't claim these sites are largely dis-functional or deficient, or that their site admins are all deluded or incompetent.

• WordPress CMS usage is at 45% of marketshare compared to 20+% for Drupal, and all others follow. -- from data at http://trends.builtwith.com/cms
• "WordPress is used by 61.0% of all the websites whose content management system we know. This is 23.3% of all websites." -- W3techs.com

It's unreasonable to dismiss such numbers, and I'm sorry but SMF isn't even in these lists (yet).
In summary on this topic, I'm hoping we can move on from this non-technical and unrelated aspect of this thread as it's not related to the multi-site SMF. Thanks.


As to why I use two environments, WP (or otherwise) with SMF, I'm spreading the risk. If WP security glitches are found in the portal front-end, I haven't compromised thousands of people using the service. If SMF security glitches, we all suffer here, but at least my portal stays up and my business site isn't affected. I'm also just getting practice with the latest tools so that I can make my own evaluations - and I continue to find both WordPress and SMF equally competent and enjoyable. In the interest of due diligence, I will build a new front-end for another site purely with SMF, and will consider a retrofit of this site later.

But again, that's not the focus of this thread...


About multiple instances versus a multi-site solution, recall that I am asking how to implement the multi-site solution here so I'm open to suggestions and I will evaluate the mods referenced so far. My initial approach is to split up the SMF services to distribute load. I know having separate installs is a rotten solution, but I can move an installation to separate hardware with almost no pain at all.

I don't know how big a multi-site instance with a single MySQL database is going to grow, with tens or hundreds of sites (teams, guilds, etc for us) per installation. At some point I'll be faced with splitting some sites out of the single database and into their own - and to my knowledge, that's currently another undefined option. The only way I can think of to support that would be to copy a live site, then remove some sites from one side and remove the other sites from the other side. Again, I don't have a handle on the multisite mods yet but I'm getting the impression that they aren't scalable. Even if they are, I think I'll soon need multiple instances of the multi-site solution.

Finally, I know linking this all up is going to take custom code. I'll do that part. I just need to decide what I'm going to be linking up first.

I hope I've addressed all points. And I'm eager to benefit from real-world experience with multi-site solutions.
Thanks.

[Gwenwyfar]

Popularity never is and never will be a synonym for quality. Something could be popular or not popular to a variety of reasons, quality is just one of them. There are thousands of things like this all around the world, both good things that didn't become popular or crappy products that became popular.

If you base all your assumptions of quality based on popularity you're doing it wrong.

[Kindred]

First. Smf has the best security record of nearly any widely used php software.
Second... Just because Wordpress is widely used does not actually make it "good" as in well programmed....   It is actually one of the most inefficient pieces of software out there. It uses 2 to 3 times as many queries per page load as needed or used by any other software.
This makes it resource heavy...


I think that your plan is poor, myself...

However, that being said, if you are actually looking at hundreds of sites, then all I can say is that you are trying to do things that the software is not intended to do and perhaps smf is not actually the best option for you

[Arantor]

We can't claim these sites are largely dis-functional or deficient, or that their site admins are all deluded or incompetent

Most people use it because it is easy to use - and I won't fault *that* part of it. But most people using it are not technically minded. Nor are they aware of the fragility of its code - code that WP literally cannot move from at this point to something more stable because of how it will break the myriad plugins out there.

I don't really care about statistics, they don't actually mean anything in practice. For example: 100% of people will die. 100% of all people that die consume water. Is water a leading cause of death? Answer: no. But I just spun it to look somewhat like that way. Whatever statistics you can come up with, I can twist them to make them say anything I want and this should not be a good argument technique.

Just because something is popular does not make it good. You should go to some of these programming forums and ask what people think of PHP - as a computer language, it's utterly terrible (and I speak as a 12 year veteran with certification in the language). There is a reason why an article entitled "PHP: a fractal of bad design" exists and is routinely cited. There is a reason PHP is likened to a hammer with two claws and no hammerhead. Something being popular does not mean it is actually good at what it does. It merely means it is reasonably not terrible (but not even necessarily least worst) in its category and that nothing has come along that has challenged it in mass market adoption. For most, WordPress is 'good enough'. Doesn't make it great, just makes it 'good enough'. And for most people, good enough is good enough, and the perks outweigh the downsides to some degree.

If WP security glitches are found in the portal front-end, I haven't compromised thousands of people using the service

There is absolutely no guarantee of that whatsoever. I have had to fix SMF sites infected via WP front ends before now.

and I continue to find both WordPress and SMF equally competent

From a user facing perspective, sure. From a technical perspective, it's no contest. (Compare SMF's and WP's security track record if nothing else. Compare the number of patches for security since July 2011 for WP and for SMF and I think you might be alarmed at how many vulnerabilities continue to be found in WP)

My initial approach is to split up the SMF services to distribute load

One does not build an entire house to hang a coat peg up.

I'm hoping we can move on from this non-technical and unrelated aspect of this thread as it's not related to the multi-site SMF.

Of course it is technical and related unless you plan to have them completely separate in EVERY way. Because if it's not completely separated (as in, could run on a separate physical server with absolutely no code changes), it is relevant.

but I can move an installation to separate hardware with almost no pain at all.

Except for the whole 'single sign on' part that is the single largest technical challenge and doubly so when separating hardware?

I don't know how big a multi-site instance with a single MySQL database is going to grow, with tens or hundreds of sites (teams, guilds, etc for us) per installation

I point out, again, that the largest installation of SMF to date runs on a much smaller architecture than you are proposing, and they have 73 million posts, 1.6 million topics.

In any case, I don't build anything for that scale right from the off. Ever. Because I know from experience that no matter what I build, no matter how much I think I've tuned it, I will be retuning it and tweaking it as things change.

More importantly, what works from a dev perspective for smaller/growing sites will not work the same at huge scales. The rules change as you get bigger. Building now for that level of capacity when you're... several years away (being realistic) from actually using that capacity will be a major headache compared to getting something going now. You can always refine the design as circumstances change - rather than speccing out a huge system and never reaching capacity.

Scope creep is a very, very real consideration here and is another factor in the recommendation against WP incidentally. Keep the complexity as small as possible. The fewest moving parts and all that.

Again, I don't have a handle on the multisite mods yet but I'm getting the impression that they aren't scalable

Depends on your definition of scalable and to what scale you are actually planning to grow to - and I don't mean 'hopefully one day'.

Facebook did not design itself from the start to use HHVM or Cassandra or massive amounts of AJAX or indeed most of the structure/implementation it currently has. It built those as it scaled, not building them in the hopes of scaling.

And I'm eager to benefit from real-world experience with multi-site solutions

1. Fewest moving parts.
2. Design for realistic scale, not theoretical scale one day at some time in the future.
3. Don't try to hammer a round peg into a square hole. This is a real danger here.

[TonyG]

I like you @Arantor. I can't disagree with a single thing you're saying and I appreciate your demeanor.

I've been in I.T. for almost 40 years. Technology and paradigms change quickly. What I want to avoid is short-sightedness and getting caught with my unscalable pants down when I know better than to let that happen. I don't want to design for a (loosely defined) small site and then realize that I need a completely new topology to support an increasing user count that's moving faster than my ability to respond. I don't respect that sort of folly in others and I won't tolerate it in my own endeavors. So no, I'm not imagining "millions of users" up front, but somewhere between a startup and Facebook I think it's prudent to plan proactively for that first or second surge in growth so that I don't have a reactive effort built into the model.

I accept that the largest SMF site supports 73 million posts, 1.6 million topics, and that this site will never get that big. What I need to determine is roughly how many sites I can realistically support with the current multi-site mods. If we can project a number like 100 or 1000 sites per instance with something like a max of 300 members each (at an "average" load of threads per member), then I'll have some idea of when the rubber-band is going to break so that I can plan for a second instance and more integration code.

Thanks again.