For the time being we will not consider to split the FCB information in more than one block.
The information in the FCB is updated regularly and in an Adabas Cluster Service environment this information has to be propagated to the other nuclei running in this cluster. Having a multi-block update in this case would mean another internal logic for propagation (locking and releasing) which could end up in a decreased performance.
We think that for the next years the lifting from 5 to 99 extents should be enough and that in practice very less customers will need 99 extents.
To answer the second question correctly we need to look at the FCB DCSECT of Version 8 which is not yet available.
An estimate from todays knowledge is following:
How many bytes are used when having 99 extents is dependent on the type of extents ?
If there are 99 AC extents (for AC extents we need for each extent hex 10 bytes, for a DS extent for example only hex 0C bytes) then the blocksize of a 3380 device (2004 bytes) would not be enough. The blocksize of a 3390 device (2544 bytes) would be enough to hold this information. For DS, NI and UI we would only have one extent in this example.
If there are 99 extents of each extent type (DS, AC, NI and UI) then an artificial device type has to be defined, because we need more than 4092 Bytes, which is the largest blocksize available today.
If there are for example 25 extents of each then we need around 2000 Byte and then Devicetype 3380 would be large enough.
With your information I have made some calculations.
If extent sizing is unchanged, the average extent size is (12,5+25)/2 = 18,75% of current size.
5 extents will allow growth to 199% of original size.
25 extents will allow growth to 6183% of original size.
99 extents will allow growth to 2061078785% of original size.