Drivers are more expensive than cells unless the cell apertures are very large. Even if you already have a cell it may well be cheaper to replace it for a more suitable one.
Driver voltage requirements scale with wavelength, but power scales with voltage squared. Cost scales somewhere between the two.
Avalanche drivers are fast, but the available repetition rates are lower. Also pulse fidelity is poorer.
Avalanche pulsers cannot deliver current for a long time. To achieve long gates one needs to charge a pockels cell and leave it charged, this results in different configurations for long gates as compared to short gates.
The fastest rise times are only achievable in 50Ω or similar geometries, this means that long gates with very fast rise times are difficult to achieve.
The dielectric constant of KD*P (a good pockels cell material) is ~80. This means that the speed of electrical waves is low and that it is hard to get the electric field into a large aperture crystal quickly even with a low drive impedance. For fast gating smaller apertures are needed.
KD*P exhibits piezo ringing. One should avoid high repetition rates (>10kHz) and be careful when using long gates with fast edges. It is the strength of the fourier components at the peizo resonance frequency that matters.
A longitudinal KD*P device will offer a higher extinction ratio than most transverse devices.
We have a lot of experience in driving a variety of pockels cells in a variety of manners; consider giving us the problem rather than asking for a specific driver.
Information to supply when considering a driver
Wavelength(s)
Aperture
Rise time, fall time, gate length: required, desired.