Thanks for the light
Dave Maguire has a post up where he discusses licensing options for ArcGIS server based products. It is an example of a good blog post – it covers facts, it throws in some of the thinking about the licensing options, and it has the comments open. Thanks Dave…
[UPDATE – some people have pointed out that I probably read the blog post wrong and that you do have to pay for a seperate ADF machine and that machine can be licensed for the ADF alone at 50% of the two socket price. There does not seem to be a discussion of how many CPUs can be on the web server. So really seems to be a reduced price for deploying just the ADF on any machine you want. This price seems a bit steep to me unless this price is for the cheapest of the Server licenses. Anyway, it seems like they went with what would be my third option so now my complaint is just about pricing. ]
That being said I do want to discuss the decision made to license the ADF separately if it is placed on a separate machine from the SOM/SOC. I like that you can deploy the ADF to a 1 socket web serving machine, which feels almost right. Dave is right that server class machines are cheaper now with better horsepower. Because of this fact, organizations that used to put all their apps on one server can now afford to split services, such as web servers separate from DBMS and other Application Servers. This is now standard practice in even some of the smaller consulting, NGO, and government agencies, with the web server going outside the firewall. What the 1 socket policy is saying is that if you want the security of placing your web server in the DMZ go for it.
The other thing being said here is that if you need more than the horsepower of 1 CPU you must also be able to afford a separate $20-40K to license the ADF so now you have to pay to put it on a separate machine.
But the are a couple drawbacks I see here:
1. Most 1 socket machines are not built with redundant power supplies or hardware RAID capabilities.
2. The price for an end user between a 1 socket and 2 socket machine is minimal, so the ability to afford a 2 CPU machine does not mean they have the extra 20-40K to license the ADF.
3. If you want to build in redundancy by putting two 1 socket machines outside the DMZ then you are paying for the extra ADF license.
So while this solution is better than the previously rumored solution – separate license to put the ADF on any separate machine – I do not think it is the right solution.
Given all the great capabilities in the ADF that David mentions I can understand not wanting to give it away to people who don’t buy ArcIMS or ArcGIS Server. For people who don’t want to buy those products make them pay a reduced licensing fee for the ADF. I think you are missing a marketing opportunity by not doing this.
I also think you should allow people to buy ArcGIS server based products to have some limited deployments of the ADF per server product they buy, I don’t know say 4 ADFs per 1 Server. But by default all server products should include at least 1 separate ADF license – If I buy a server product I should be able to put the ADF on 1 separate machine with up to the same number of sockets as my SOM/SOC/Spatial Server machine.
A third option is to license the ADF at a reduced price to current server license holders. Making us pay the $20-40K for just the ADF not only seems ridiculous but also brings up interesting questions. If I buy enterprise server for my SOM/SOC machine do I have to buy enterprise server for the ADF to deploy mobile apps or can I just buy standard? And what about extensions? If I buy them for my SOM/SOC server do I need to buy them for my ADF server?
To sum up –
Good blog post because it was open to discussion and now we can talk about facts rather than rumors
While I think the licensing described in the post is better than what I heard before I think there are still some problems. The model for pricing the deployment of the ADF separately still needs some tuning to bring it in line with the way people work with server based products.