Practical Licensing for distributed companies

[JonE]

Everyone,

The organisation I work for is a Sparx customer, and I'm currently trying to organise better management of those licenses. The main problem I have is that the nature of our work means most of the staff are offsite 90%+ of the time. This makes it extremely difficult for them to use floating licenses.

For example, if a staff member takes an EA license from the office key server on a 4 week lease, that license is now denied to everyone else. even if the person only uses it a few times in that period. If that person doesn't take the license they can't work in EA without returning to the office.

Conversely, if a user goes on-site for 6 to 9 months without a license. We don't want to force them to commute several hours just to use EA.

We've tried the "mixed mode" license arrangement with some licenses on short lease and others on long lease, but it has proven less than perfect. We've also tried aquiring licenses from the office network via VPN, but due to infrastructure limitations this can't work for everyone.

Appart from buying every potential user their own license, or purchasing enough floating licenses to run 6+ month leases, do I have any other options?

What I would really like is a web service I could host locally and secure against our domain, which would allow the users to retrive/release a key from a central server from any location with basic internet access over http.

[pwhiston]

Hi

One way that you could consider is delivering EA across Citrix. We have tried this and it works very well. ie you host it on a central server against your floating licenses and users log in remotely to use. Performance is fine and with proper configuration you can have access to all your lcal disk drives. You can also have centralised model as well using a database for your models.

[JonE]

I appreciate your suggestion, but it has a couple of fatal flaws:

1. The remote users then require reasonably performant constantly connected internet connection to work. A lot of our users are notebook based and mobile, so it won't fly for them.

2. If you're working on sensitive or CiC material, the client may not want their plans and designs shipped off-site. (even if it's a good idea for back-up purposes!)

3. It requires a reasonable citrix server, so investment in a citrix licenses (or Windows TS) and hardware to run it. The cost of that server plus the bandwidth expansion we'd need probably exceeds the cost difference of purchasing permanent licenses for everyone. The economic arguement would be different if we already had that infrastructure.

So, close but no cigar.

[«Midnight»]

Perhaps Rational then.

Simple, out of the box, and it's a great brand name.

Other than the 20-40 times price increase, expensive learning curve, and heavy (but justified, since it's a big name) infrastructure cost, it should serve you as well as EA.

Now think "fatal flaw" through again. If anything looks different, we can continue this thread.

[JonE]

Perhaps Rational then.

Simple, out of the box, and it's a great brand name.

Other than the 20-40 times price increase, expensive learning curve, and heavy (but justified, since it's a big name) infrastructure cost, it should serve you as well as EA.

Now think "fatal flaw" through again. If anything looks different, we can continue this thread.

I don't see the relevance of your post?

I want to use EA, I am using EA, I'm trying to get more EA - my only issue is with the practical distribution of licenses in an efficient manner. I want a way I can purchase a minimum number of licenses for maximum utilisation - that is the nature of economic reality. The "fatal flaw" is in the citrix as a deployment platform for my particular scenario.

Talking about how expensive an alternative solution is has no relevance to my post. If you have a helpful suggestion like pwhiston did, we can continue this thread.

[pwhiston]

Jon
You are right - if you don't have Citrix then its too heavy an investment. Regarding data I thought it was yours as it would reside on your employees laptops.
I didn't read your first post close enough but you say that you have had some success with getting license via a VPN connection which sounds the most promising way ahead. The issue seems that if you don't have some network connection to your license server you will have to rely on the check out and expiry facility - or buy standalone licenses.
The EA floating license manager is still a good way of maximising license usage.

[JonE]

pwhiston,

I admit I was just parsing potential issues in my head - I don't think data confidentiality would be a major issue, as you said.

The issue I have is the slack time in the license usage - if a user checks the license out for 2 weeks and goes off-network but only uses it onces, well, that isn't fantastic utilisation.

I have an infrastructure limitation on the number of VPN users I can support, which is why I'm reluctant to go down that path.

[«Midnight»]

Some musing here, but take a look at the last paragraph if you don't want to wade through the rest.

Let's go back to your original idea of a Web service.

To make floating licenses work, your clients (the end users) need access to the license file (KeyStore) on the server. Is there some way you can contrive to have a Web service provide this 'access?' I believe the client has to think it has run an executable located on your key server.

This might prove tricky, since you'd in essence be passing an executable through a Web service. However, perhaps it could work somehow. You are only passing outwards, which may mitigate firewall concerns. And you might only need to pass as far as the Web service host, since access to the program could likely be served from there.

A tricky part might be handling the current status of your licenses (available and reserved). This context likely needs to stay with KeyStore. Somehow you've got to capture it and provide it to the spot from which the Web service actually points your clients.

AFAIK, floating licenses can be divided into pools, which you've tried. The big problem here is that you can't divide too many times (a single division is often too many) before each of the pools seems to be the wrong size. Unless you have a really big user community you'll quickly be faced with an inventory of unused licenses in one pool and disconnected users queued up at the other. Adjusting the pools would take time, irritate everyone, and merely move the problem around so everyone gets to experience all the problems.

Perhaps you can create a Web service that directly serves access to KeyStore, and make it available to everyone who can authenticate at the Web service level, inside and out. Would this work for your security infrastructure? You might have to let internal users connect indirectly, by going right out to the internet and back in to your Web service (rather than via your intranet).

At the very least, the ability to connect via a Web service is worth raising a feature request with Sparx. We (the 'regulars' in this forum) have a set practice for doing these, which helps keep us all in the loop. You can do a search for recent (past few months) posts using the keywords "Request-a-Feature" or "Feature Request" (without quotes) and perhaps "Best Practice" as well. This will walk you through. I'd support this, and I'm sure others would too. The nice thing is that this would not require extensive change to the core EA product, so could be done at any time - it would be less likely to be postponed to the "next major version."

David

[bioform]

Seems like you just need to buy some fixed licenses... IF your really as committed to using EA as you say... then you are NOT talking about a lot of money...

Licenses are part of doing business... And you should check into getting a volume discount then...

ALSO, if your users are OFF site, and do NOT have highspeed access... then that sounds a little strange...

IF they are working off-site for what you seem to be saying is the VAST amount of their time... I would expect your company to have a policy that supports your staff... I mean if they are saving you money by NOT requiring office space (AND THAT IS ALOT MORE EXPENSIVE THEN a year license... say less that 30$ bucks a month!!!)

So I think you need to consider all of the cost saving factors involved in supporting 90% of your staff that tele-commutes and buy some freaking licenses...

The bang for your buck is enormous...

[JonE]

The big problem here is that you can't divide too many times (a single division is often too many) before each of the pools seems to be the wrong size. Unless you have a really big user community you'll quickly be faced with an inventory of unused licenses in one pool and disconnected users queued up at the other. Adjusting the pools would take time, irritate everyone, and merely move the problem around so everyone gets to experience all the problems.

You've hit the nail on the head with that one. I don't think it would be as important a problem for a larger organisation where the distribution between groups remained more constant, but with a small number of licences and users it's impossible to staticly define the distribution. Perhaps a keystore that allowed the requesting client to define the lease-length rather than determining it globally at the server would make more sense.

Perhaps you can create a Web service that directly serves access to KeyStore, and make it available to everyone who can authenticate at the Web service level, inside and out. Would this work for your security infrastructure? You might have to let internal users connect indirectly, by going right out to the internet and back in to your Web service (rather than via your intranet).

Yes, I think that would work. I don't think the direct or indirect nature of the connection really matters much. I don't know how I'd go about creating a web service to give direct access to the KeyStore - I was under the impression it's a file that has it's network path embedded in it, so there could be some issues with just exposing it.

At the very least, the ability to connect via a Web service is worth raising a feature request with Sparx.
David

Thanks for your advice on the process - I think I'm going to be meeting some of the Sparx guys in a couple of weeks at the Architecture Forum ( http://www.architectureforum.net.au/Pages/aaf.aspx ), so I'll see what they say and then follow it up here.

Jon

[JonE]

Seems like you just need to buy some fixed licenses... IF your really as committed to using EA as you say... then you are NOT talking about a lot of money...

Licenses are part of doing business... And you should check into getting a volume discount then...

Arguing that the point is moot because it's "not a lot of money" goes directly against basic business principals - why would I want to spend more to achieve the same end result if I could achieve it by better efficiently instead?

Even if I figured out that for my organisation licencing everyone was the best option, surely organisations with different usage patterns and scales are going to run into very similar issues to what I have, and want a better answer than "throw money at it"?

Re: Commitment to EA, I'm pragmatic. I always take the attitude "I will use this until I can't, or I find something better". It's a tool after all, not a religion.

ALSO, if your users are OFF site, and do NOT have highspeed access... then that sounds a little strange...

Not so, most large corporates strictly control their internet access and so a direct VPN out may not be possible. Hence the discussion of the web service idea.

IF they are working off-site for what you seem to be saying is the VAST amount of their time... I would expect your company to have a policy that supports your staff... I mean if they are saving you money by NOT requiring office space (AND THAT IS ALOT MORE EXPENSIVE THEN a year license... say less that 30$ bucks a month!!!)

There are two issues with this point. Firstly, you made the assumption that we lease office space for the entire staff regardless of usage - clearly not an efficient practice! Secondly, It's well and good to say "tool X only costs $30 / month" but you must then multiply that by the number of tools we use. Even if you ignore these two points the simple economically rational argument prevails - why should we spend more money than we have to?

So I think you need to consider all of the cost saving factors involved in supporting 90% of your staff that tele-commutes and buy some freaking licenses...

I said "offsite" not "tele-commute" (although I guess we have some of both). Again, throwing money at the problem may not be the best way to solve it.

The bang for your buck is enormous...

I never disputed that EA is a good ROI product!

Jon

[bioform]

I guess I missed the original point of your post...

If you were discussing suggestions to EA for improvements with it's license distribution, yes you have raised some good points...

If you were trying to find a solution to your license issues using the existing capability of EA, then I again believe the answer is to optimize your licenses strategy using existing functions:

1. Smaller groups of licenses (trying to optimize these groups with the smaller workgroups, rather then dumping all licenses into a big pool.)

2. Check-Out licenses to power users (heavy use users, rather than the occasional modeler.)

3. Make sure your users are NOT consumming licenses when they need to just VIEW the model (web publish feature of EA, or use EA Viewer application)

4. Purchase licenses at a different functional level (hey if your not implementing security or using a RDBMS to store the model, cut cost by NOT getting the corporate edition)

Business must plan for the costs associated with any tools, and optimize there short-term costs with WHAT the tool provides me today, AND plan on how those costs may change in the future (but realize those wanted changes may never occur...)

So IF your ROI is still on the positive side, then yes, "throwing more money at it.." is the answer for now and the near future..