Is there a formula to help determine the access speed of a file that is hosted on filemaker server given x number of simultaneous connections?
There are many variables involved here, each one can be a bottleneck.
First question though: how is the file accessed: on a LAN or across the WAN?
WAN access adds a lot to the # of variables.
Also: what "speed" do you want to measure? The speed of opening the file? The speed of running certain routines?
Mac or Windows?
FMS comes with its own statistics log (off by default) that can give you a wealth of information to use for this kind of diagnostics. On Windows these "counters" are also available through PerfMon so you can correlate them with the other counters that are available to give you a broad overall picture of the machine's performance.
I would be interested in both LAN and WAN variables.
Speed - mean - enter data and it uploads to the file on the server. - So, running routines.
I am designing on Mac - but the users will access via Mac, Windows, iOS
There are 4 traditional bottlenecks that influence performance:
1) Memory - RAM is cheap so this one should not be a problem, either on client on server
2) Processing Power - this is a factor of a number of things
On the server
the number of concurrent users (including server-side schedules, PSoS sessions,...)
the design of the solution
the nature of the routines (heavy data crunching vs. just displaying static data,...)
For the clients: largely the nature of their work and the design of the solution, keep in mind that iOS devices are not as performant as desktops/laptops so you need to factor that into your design
3) disk i/o
Slow disk i/o is disastrous for performance, depending on the amount of data changes, frequency of the backups, server-side routines etc
4) network i/o
On a LAN this is rarely an issue, but again depends on the nature of the solution & the number of users
On a WAN though this becomes THE botleneck. WAN access speeds are a fraction of normal LAN speeds and the latency on the route between client and server will slow things down a lot.
If WAN access is required, the solution must be explicitly designed for that to minimize the UI bits and the amount of data that is sent between client and server (so no long lists, no layouts with multiple tabs and portals on each tab, no big graphic elements in the UI design,...)
So there is no formula. But you should evaulate your intended deployment and solution design with these potential bottlenecks in mind and make adjustments where necessary.
Retrieving data ...