There are many drivers to move to an alternative carrier gas, but whether your reasons are to increase analytical throughput, reduce analysis time, or find an alternative carrier to helium for cost and availability reasons, there are some key facts that you need to consider for a successful transition.
For most of us considering a transition from helium for cost or availability reasons, we naturally turn to hydrogen as a carrier gas, but in many instances nitrogen has been very successfully used as a carrier gas for capillary gas chromatography (GC). Many separations are carried out with "excess resolution" in GC; therefore a small loss in efficiency encountered when operating nitrogen at higher linear velocities than its Van Deemter optimum often do not critically impact the resolution between analytes. This is especially true when operating in constant flow mode with temperature programming, where the practical impact of increased linear velocity on efficiency is reduced.
If hydrogen is chosen as the alternative carrier gas, we have to be mindful that the viscosity of hydrogen is a little less than half that of helium, and so we will need to approximately double the carrier-gas pressure to maintain the same linear velocity (and analyte retention times) for the translated method. Most modern instruments can be programed in terms of linear velocity rather than flow rate or carrier-gas pressure, provided we tell the instrument what the carrier gas is. Simply set your instrument for 30-cm/s hydrogen rather than 30-cm/s helium, for example, to approximately match retention times, or use method translation software.One should always be mindful, however, that when switching carrier gas, the selectivity of the separation may change, especially when separating analytes of divergent polarity using temperature-programmed separations. There are some very simple relationships that allow us to guard against selectivity changes, peak inversions, and so on — these are shown in equation 1:
New program rate = original program rate × (new average linear velocity/original average linear velocity)
New isothermal time = original isothermal time × (original average linear velocity/new average linear velocity) 
In equation 1, the linear velocity of each carrier is required to obtain a consistent flow or retention time for an early eluted analyte with the chromatogram, or the figure set for the linear velocity of hydrogen to achieve the same flow rate as was used for the helium based separation.
If converting the method to take advantage of the fact that hydrogen can maintain high peak efficiency at higher linear velocities, then one might want to use a shorter column with a narrower internal diameter. This should achieve approximately the same number of plates, but at a higher linear velocity (dictated by the column internal diameter) and over a shorter column length, and thus an equivalent separation will be achieved in a shorter time. If this is the case then the simple empirical relationships shown in equation 2 can be used to calculate the new temperature program to avoid problems with selectivity changes:
New program rate = original program rate × (length column 1/length column 2) × (gas velocity 2/gas velocity 1)
New isothermal time = original isothermal time × (length column 2/length column 1) × (gas velocity 1/gas velocity 2) 
The calculations in equation 2 assume the phase ratio (β = column radius (r)/2 × stationary-phase film thickness [d f]) remains constant between columns.
For those wishing to optimize the separation further, a useful relationship is shown in equation 3, which determines a good starting point for temperature programming rate optimization:
Optimal ramp rate = 10 °C/void time 
where void or hold-up time is the time for a nonretained peak to be eluted into the detector.
In practical terms, one may need to consider using a 2-mm i.d. inlet liner (as opposed to the 4-mm liner that you probably use now) to avoid problems with band broadening, which will reduce the attainable efficiency at higher linear carrier gas velocities. Furthermore, it's important to install the column into the detector carefully to avoid creating any dead volumes by having the column too "low" relative to the jet or detector entrance aperture.
In our hands, fast GC separations using 0.15- or 0.18-mm i.d. columns tend to have an optimum practical gas velocity of around 120–140 cm/s with hydrogen carrier gas, above which the column dimensions need to be changed to achieve reasonable separation.
In detector terms, if switching to hydrogen using flame ionization detection (FID), remember that you will need to provide an alternative make-up gas such as nitrogen or helium and that the sampling rate of the detector may need to increase if peak width is particularly narrow.
For MS detection, check with your manufacturer if your system is designed to work with hydrogen — most systems will not be able to use a nitrogen carrier. Some hardware modifications may be required for hydrogen operation (typically to the pumps and ion source); note that you will need to operate a thorough background depletion routine and that the appearance of spectra (typically chemically labile or highly reactive species at lower concentrations) may alter because of changes in chemical and ionization processes within the inlet.