In general terms, overclocking means to run a particular component at a clock rate over the rated clock speed (and is the opposite of underclocking, or running the component under the rated clock speed). By running a component at more than the rated clock speed, it allows you to achieve higher performance. Underclocking is used when an application is not performance critical and heat and power consumption are more of an issue.
On the original Pentium processors, one overclocked by increasing the clock multiplier via jumpers on the motherboard or increasing the system bus clock speed, also via jumpers. In this way it was common to overclock a Pentium 133 to be a Pentium 166 or other such speed rating, and also to overclock a Pentium 90 to a Pentium 100 (this was more common since it involved moving the system bus from 60 to 66 MHz and the PCI bus from 30 to 33 MHz). This also works on K5 and K6 processors.
On the Pentium Pro processor introduced the locked multiplier, however all of the processors ran at the same multiplier (50×3-150, 60×3-180, 66×3=200) so by increasing the bus speed one could overclock the processor. There were a few motherboards that allowed for multipliers to be reset, and those chips could be overclocked to 233 MHz.
On the Pentium II and Celeron and Celeron A, all multiplers were locked, so you had to increase bus speed. This became especially useful with the i440BX chipset, where a 66 MHz bus part could be made to run at up to 100 MHz, such as in the case of the Celeron 300A and a particular s-spec Pentium II 300. This also continued onto the Celeron II processor and Pentium III processors. Around this time Abit also introduced the softmenu, where voltage and bus speed could be adjusted in BIOS. Onto newer processors, the AMD Athlon origianlly came with a locked multiplier, but with a special module from AMD they could be unlocked. They were overclocked in similar ways to the older CPUs. This also continued into the Pentium 4, with only bus speeds, chip speeds and voltages changing.