Well, it used to be that 1/3 of the horizontal acceleration was used for the vertical direction, however, since Northridge produced very high vertical accelerations I'm not sure what the codes have finally developed for this direction/application.
As for records, I assume you wish strong-motion records or accelograms and I'm not sure what agencies out there record that kind of data for public use. Try the COSMOS website. I don't know the address so just search using the acronym above.
As Qshake said the vertical component of earthquake generally would be of order 1/3 of horizontal coponent. But it is not a standard value since the earthquakes I have studied, for example Kobe, Northridge, Chalfant, whitter field and many more, the vertical componant is in comparision with that of horizontal motion, few cases it exceeded the horizontal motion.
The vertical component is not at danger, b'coz the stuructures designed to the respective building codes have adeaquate factor of safety against vertical loads and the additional axial load arising from vertical motion will be easily taken care off by this factor of safety.
The vertical motion produced cosiderable effect on thin-walled structures susceptible to local buckling. In this case the local buckling coupled with the additional axial load due to vertical ground motion produces considerable damage during an event of severe earthquake.
For dynamic analysis of buildings in the U.S. (per 1997 UBC 1631.2.5), the vertical component of ground motion may be defined by scaling corresponding horizontal accelerations by a factor of two-thirds. For near-source sites, site specific data is required.
