[[TCP (Transmission Control Protocol)]] [[TCP Options]] ## Flow Control - Balances the rate of data being created from the sender to the rate of data that can be used - Send Window: - A buffer of data to be sent is made and the data essentially queues up to be sent. The data that is sent must have an acknowledgement before it can be discarded from buffer. - Receive Window: - Works the same way as Send Window. Bytes queue up in line to be moved into the buffer, any that cannot be received will not be acknowledge and re-sent. Once data is in the buffer it waits to be processed and once complete, acknowledgement is sent to sender and data is pushed to next layer. - Flow Control is based on feedback i.e acknowledgements to ensure that the data has been received, processed and discarded correctly.

Error Control

• Ensures that data arrives to destination without any issues
• Error Control is based on the use of three tools
  • Checksums
  • Acknowledgements
  • Time-Outs

Rules for Generating ACK

• 1. When an end sends a data segment to the other end. It must have an ACK as the ACK is the next sequence number that the end is expecting.
• 2. The receiver needs to delay sending an ACK segment if there is only one outstanding in-order segment. This is to prevent extra ACK traffic
• 3. There should not be more than 2 in-order unacknowledged segments at any time. It prevents unnecessary re-transmission.
• 4. When a segment arrives with an out-of-order sequence number that is higher than expected, the receiver immediately send an ACK segment announcing the next expected segment.
• 5. When a missing segment arrives, the receiver sends an ACK segment to announce the next sequence number expected
• 6. If a duplicate segment arrives, the receiver immediately sends an ACK.
• Note: Lost acknowledgements may create deadlock if they are not handled properly.

Congestion Control

• TCP uses a congestion window and a congestion policy to avoid congestion and to detect and alleviate congestion after it has occurred.
• rwnd = receive window
• cwnd = congestion window
• Three Phases:
  • 1 Slow Start, Exponential Increase
    • 1. cwnd starts at the MSS (max segment size). Doubles until cwnd = ssthresh
      • MSS = (MTU - IP hdr len - TCP hdr len)
  • 2 Congestion Avoidance, Additive Increase
    • 1. When cwnd = slow start threshold, stop exponential increase
    • 2. cwnd increases by 1 for every “window” it sends
    • 3. Continues to increase until congestion is detected
  • 3 Congestion Detection, Multiplicative Decrease
    • Congestion detection occurs when a sender has to re-transmit a segment.
    • A re-transmission can occur due to:
      • Re transmission time-out (RTO) expires
      • Receive 3 ACKs for the same segment (fast re-transmit condition)
    • When congestion is detected, the threshold is cut in half
    • When congestion is due to RTO: set cwnd = 1, start at Phase 1 (worst case scenario)
    • When congestion is due to fast re-transmit: set cwnd = threshold, start Phase 2

TCP Timers

• To perform smooth operation, most TCP implementations use at least four timers
  • Retransmission Timer (RTO)
    • Waits for ACK, sends first segment not ACK’d if it expires
    • Based on round trip time (RTT)
  • Persistence Timer - Starts when a zero-window size packed is received; probe segment sent to get and ACK
  • Keepalive Timer - used when a connection is open, but no dat is flowing; probe segment sent to get an ACK
  • TIME-WAIT Timer - Has to do with closing connections

Selective Acknowledgement (SACK)

• Allows the sender to have a better idea of which segments are actually lost and which have arrives out of order.
• Two parts:
  • SACK permitted: As an option during three way handshake
  • SACK option: As an option during ACKs
• SACK options includes a list for blocks arriving out of order
• 32-bit blocks that define the beginning and the end of the blocks received. -> 8 bytes per block, plus the SACK option of 2 bytes
• Up to 4 blocks may be specified (40 byte options limit)