>
I'm not sure you understand my "basis". I'm saying that using proprietary features to get "scaling" limits your choices and may not be worth it.
>

I think this is where we differ in opinion. The whole proprietary v. non-proprietary issue. If you are using a particular back-end, why not use the features of that backend?


>
Cache is a good example. Internally it uses Cache Objects which is a whole new non-SQL language. It's meant to be quite cool, but I have no interest in learning it or retaining expertise in the company. Nevertheless the same "version" that runs for SQL server runs on Cache. Had I developed SP for SQL server, Cache would not be an option.
>

Again, I cannot comment on Cache since I have not used it, nor have I heard of it.


>
To you, this is clearly less valuable than the Scaling issue. My problem us that while my point is quite easy for each of us to value in terms of our own business, "better scaling" is quite hard to quantify and value by itself.
>

If I interpret your point correctly, you are asserting that I value scaling more than being non-proprietary. If I am wrong, my appologies in advance. Assuming I am right, to me, the proprietary v. non-propriatary issue is a moot point. Therefore, any point that relates or ties back to that is moot to me as well. To me, it is off topic and one that I will not comment on any further. I wont go as far to say that we will agree to disagree since the issue at best is indeterminate...


>I'd quite like to see some sample performance figures so we can see the "real life" differences between Remote Views and SP. Not trivial stuff that means nothing in real life, but useful examples. I'd be interested in (say) a 50 user fat client scenario (which encompasses most business) with maybe 10 3-row reads and 4 one-row writes per minute per user against a fairly standard client/invoice database, plus one user doing meaty un-indexed reports. With (say) SQL server on a dual PIII600, 256M machine ans a 100Mbs network, what is the difference experienced by each user for SP Vs RV? What if you increase it to 100 users, assuming the developers have been smart and have downloaded the View definitions to the Client, and you are using SQL Server 7 and have set up ODBC to use prepared statements properly? What about 200 users? If you say to me that the difference is a second per user per write with (say) a standard customer table, I'll give a sickly grin and go look at DataClas.<g> But I
>doubt the users would be able to see any difference. If I am right, who are we satisfying by increasing Scalability?
>

The issue is also about manageability. Imagine a remote view that is composed of multiple joins. You now need to update multiple tables. This is a bear - if not downright impossible. The issue is also about security. You may be forbidden to render client-side SQL. i.e., you may be forced to deal exclusively with SP's. From an integrity standpoint, SP's can be wrapped in individual transactions. Finally, SP's - assuming they are optimized, will always be faster than their Remote View Counterparts.