Well, network diameter (ND) of 7 does restrict you to how strung out your L2 switches can be. Depending on how the VLANs are distributed, it is conceivable that you could interconnect 100 or 200 switches or more; but keeping to a ND=7 would force you to introduce L3 separation to keep things manageable. No VLAN could be allowed to traverse the entire system of switches, unless it had no STP redundant links associated with it.
As to the topic of adjusting STP timers, Max Age (MA) and Forward Delay (FD) are ultimately derived from the Hello Time (HT) and Network Diameter (ND). Minimum HT is 1 second, default is 2 seconds; minimum ND is 2, default is 7.
If you know your ND and your HT (which is configured on the root bridge or switch, per VLAN), you can calculate your optimum MA and ND:
MA = 4*HT + 2*ND - 2
FD = 0.5*(4*HT + 3*ND -0.5)
If you get a value with a fraction, just round it up to the next whole number. Also, since HT can't go lower than 1, the only way you can really minimize your STP timers is to make the ND as small as possible.
While you can add switches and create networks with ND greater than 7, that increases the time required for STP to (re)converge. And, even the MA and FD values have maximums on Cisco switches: I have seen MA max at 200, FD max at 200. So there are some practical limits. (If you want to see what they are, plug these maximums into the formulas above, along with your HT of choice, and solve for ND; the smaller of the two ND's is your limit.)
FWIW, you can read all about Spanning Tree for free if you go to http://www.ciscopress.com, look up Cisco LAN Switching by Kennedy Clark and Kevin Hamilton. Last time I checked, two sample chapters are the ones devoted to STP. Everything you ever wanted to know.
Except the formula for calculating STP port cost of a FEC or GEC connection, which can be found here:
http://forum.cisco.com/eforum/servlet/NetProf?page=netprof&CommCmd=MB%3Fcmd%3Dpass_through%26location%3Doutline%40%5E1%40%40.ee9c473/0#selected_message
Hope this helps.
