The main difference between multiversion and lock models is

that in MVCC locks acquired for querying (reading) data don't

conflict with locks acquired for writing data and so reading

never blocks writing and writing never blocks reading.