What kind of boundary it is
USGS describes the San Andreas fault system as the main surface break between the Pacific plate and the North American plate, a transform boundary where the plates slide past each other. Circular 1198 gives the system a length of about 1,300 km (800 miles) and an average relative rate of 5 cm per year (2 inches per year) for that plate pair. The cool-facts page says it is not a single continuous fault. It is a zone of many segments, more than 1,300 km long, and in places as much as 16 km (10 miles) deep. The name comes from the San Andreas Valley, according to that page’s note on A.C. Lawson.
How long it has been a boundary
The USGS “Back to the future” page says the zone started moving about 28–30 million years ago and has slipped about 300–350 km (186–220 miles) in total. In northern California the zone includes the Hayward and Calaveras faults as well as the northern San Andreas. In southern California it includes the San Jacinto and other faults. Motion on one named strand is not motion on all of them.
Where the volcanoes are
Transform boundaries do not build the same volcanic arc as a subduction zone. The Cascade volcanoes, including Mount St. Helens and Mount Rainier, belong to Cascadia, where a plate is going down. The San Andreas connects, in the USGS circular’s wording, the East Pacific Rise east of Baja California to the Gorda Ridge at the south end of the Juan de Fuca plate. South of the fault the boundary is divergent. North of it, Juan de Fuca dives under North America. That is why this site has a separate Cascadia page.
What we do not have in the distance tool
The USGS list of the 20 largest earthquakes since 1900 does not include a San Andreas epicenter. The distance tool therefore has no San Andreas vertex taken from that table. It will still tell you whether a point sits in the broad Cascadia / western North America rectangle, which is not the fault. For the fault itself, use USGS fault publications rather than a homemade line.