A line coding technique that maps eight data bits into a 10-bit symbol, or character. The eight-bit data octet is divided into two groups. The three most significant bits, or leftmost bits, are encoded into a four-bit group (3B/4B).The five least significant bits, or rightmost bits, are encoded into a six-bit group (5B/6B).The two groups are then concatenated, or joined together, and placed on the line. As eight bits yields 256 possible bit combinations (2 8 = 256) and 10 bits yields 1,024 (2 10 = 1024) bit combinations, each eight-bit data octet can be phrased two different ways, with one being the bit-wise inverse of the other. For example, a data octet of 11001010 might be expressed the first time as 1000100111, and the second time as 0111011000.That encoding scheme yields direct current (DC) electrical balance on the line, as the number of 1s and 0s will be equal in the long term. This approach also ensures proper clocking as there is sufficient ones density. 8B/10B also provides an additional embedded error control mechanism similar to that discussed in 4B/5B, which is used in 100Base-TX. On the downside, 8B/10B adds 25 percent overhead (10/8 = 1.25) to the serial datastream. Note: The 10B format also provides for a number of control characters. 8B/10B is used in 10 Gigabit Ethernet (10GbE or 10GigE), Gigabit Ethernet (GbE or GigE), ESCON and Fibre Channel Storage Area Networks (SANs). See also 10 Gigabit Ethernet, 4B/5B, 64B/66B, balanced, DC, encode, ESCON, Fibre Channel, GigE, ones density, and line coding.