| /freebsd/share/man/man4/ |
| H A D | mod_cc.4 | 35 .Nd Modular congestion control
37 The modular congestion control framework allows the TCP implementation to
38 dynamically change the congestion control algorithm used by new and existing
68 Unloading a congestion control module will fail if it is used as a
71 used to switch a connection to an alternate congestion control.
72 Note that the new congestion control module may fail to initialize its
85 Read-only list of currently available congestion control algorithms by name.
87 Returns the current default congestion control algorithm when read, and changes
95 which alters the window decrease factor applied to the congestion window in
96 response to an ECN congestion signal.
[all …]
|
| H A D | cc_newreno.4 | 36 .Nd NewReno Congestion Control Algorithm
66 the congestion window in response to a congestion signal per: cwnd = (cwnd *
71 the congestion window in response to an ECN congestion signal when
82 Note that hystart++ requires the TCP stack be able to call to the congestion
89 congestion controller is rack.
99 the congestion window in response to a congestion signal per: cwnd = (cwnd *
104 the congestion window in response to an ECN congestion signal when
124 .%T "TCP Congestion Control"
142 congestion control algorithm first appeared in its modular form in
145 This was the default congestion control algorithm in FreeBSD before
[all …]
|
| H A D | cc_cubic.4 | 36 .Nd CUBIC Congestion Control Algorithm
38 The CUBIC congestion control algorithm was designed to provide increased
40 The CUBIC congestion control algorithm is the default for TCP.
44 The congestion window is increased as a function of the time elapsed since the
45 last congestion event.
47 with the inflection point set to be the congestion window value reached at the
48 last congestion event.
49 CUBIC also calculates an estimate of the congestion window that NewReno would
50 have achieved at a given time after a congestion event.
51 When updating the congestion window, the algorithm will choose the larger of the
[all …]
|
| H A D | cc_dctcp.4 | 34 .Nd DCTCP Congestion Control Algorithm
36 The DCTCP (data center TCP) congestion control algorithm aims to maximise
38 proportion of Explicit Congestion Notification (ECN) marks received from capable
39 hardware as a congestion signal.
41 DCTCP uses fraction of ECN marked packets to update congestion window.
44 marked, congestion window is halved.
73 The initial value to estimate the congestion on the link.
74 The valid range is from 0 to 1024, where 1024 reduces the congestion
77 could not yet adjust to the congestion level on that path.
90 A flag if the congestion window should be reduced by one half after slow start.
[all …]
|
| H A D | cc_htcp.4 | 36 .Nd H-TCP Congestion Control Algorithm
38 The H-TCP congestion control algorithm was designed to provide increased
43 The congestion window is increased as a function of the time elapsed since the
44 last congestion event.
46 congestion event, and then switches to a high-speed mode based on a quadratic
62 network queues non-empty during congestion recovery episodes.
65 Controls use of the RTT scaling algorithm, which is designed to make congestion
66 window increase during congestion avoidance mode invariant with respect to RTT.
83 .%T "H-TCP: TCP Congestion Control for High Bandwidth-Delay Product Paths"
90 .%T "H-TCP: A framework for congestion control in high-speed and long-distance networks"
[all …]
|
| H A D | cc_hd.4 | 34 .Nd HD Congestion Control Algorithm
36 The HD congestion control algorithm is an implementation of the Hamilton
37 Institute's delay-based congestion control which aims to keep network queuing
40 HD probabilistically reduces the congestion window (cwnd) based on its estimate
45 Loss-based congestion control algorithms such as NewReno probe for network
47 HD competes with loss-based congestion control algorithms by allowing its
59 Queueing congestion threshold (qth) in ticks.
86 .%T "A strategy for fair coexistence of loss and delay-based congestion control algorithms"
105 congestion control module first appeared in
118 congestion control module and this manual page were written by
|
| H A D | cc_chd.4 | 34 .Nd CHD Congestion Control Algorithm
38 It provides tolerance to non-congestion related packet loss and improvements to
43 threshold (queue_threshold) and decides to reduce the congestion window (cwnd)
49 The probability of cwnd reduction due to congestion is calculated once per round
67 Queueing congestion threshold (qth) in ticks.
76 If 1, cwnd is adjusted using the shadow window when a congestion
101 .%T "Improved coexistence and loss tolerance for delay based TCP congestion control"
113 congestion control module first appeared in
126 congestion control module and this manual page were written by
|
| H A D | cc_cdg.4 | 31 .Nd CDG Congestion Control Algorithm
33 CAIA-Delay Gradient (CDG) is a hybrid congestion control algorithm which reacts
42 backoff mechanism, and will also try to infer non congestion related
76 Number of consecutive delay-gradient based congestion episodes which will
80 Number of consecutive delay-gradient based congestion episodes to hold the
87 RTTs during congestion avoidance mode.
113 .%T "Revisiting TCP Congestion Control using Delay Gradients"
133 congestion control module first appeared in
146 congestion control module was written by
|
| H A D | cc_vegas.4 | 34 .Nd Vegas Congestion Control Algorithm
36 The Vegas congestion control algorithm uses what the authors term the actual and
37 expected transmission rates to determine whether there is congestion along the
58 If diff > beta, congestion is inferred and cwnd is decremented by one packet (or
107 .%T "TCP Vegas: end to end congestion avoidance on a global internet"
121 congestion control module first appeared in
134 congestion control module and this manual page were written by
|
| H A D | tcp_bbr.4 | 44 Bottleneck bandwidth and round-trip time (BBR) is a congestion control algorithm which
46 It is a round-up redesign of congestion control, which is not loss-based, delay-based,
132 .%T "BBR: Congestion-Based Congestion Control"
143 .%T "Towards a Deeper Understanding of TCP BBR Congestion Control"
151 congestion control module first appeared in
157 congestion control module was written by
|
| H A D | sctp.4 | 377 implements micro-burst control so that as the congestion window
433 .It Sy Congestion Control
436 Default congestion control module.
439 A value of 0 enables the default congestion control algorithm.
440 A value of 1 enables the High Speed congestion control algorithm.
441 A value of 2 enables the HTCP congestion control algorithm.
442 A value of 3 enables the data center congestion control (DCCC) algorithm.
444 Defines the initial congestion window size in MTUs.
446 Use congestion control instead of 'blind' logic to limit maximum burst when sending.
449 Enable Explicit Congestion Notification (ECN).
[all …]
|
| H A D | tcp.4 | 138 Set or query congestion control algorithm specific parameters.
143 Select or query the congestion control algorithm that TCP will use for the
418 The default congestion control algorithm for
422 Other congestion control algorithms can be made available using the
463 A number of variables for congestion control are under the
469 Variables for NewReno congestion control are under the
492 Enable support for TCP Explicit Congestion Notification (ECN).
692 Enable the ability to specify initial congestion window in number of segments.
695 using congestion window from the hostcache.
701 buffers or the link is experiencing congestion.
[all …]
|
| /freebsd/sys/dev/mlx5/mlx5_ib/ |
| H A D | mlx5_ib_cong.c | 56 #define CONG_LOCK(dev) sx_xlock(&(dev)->congestion.lock)
57 #define CONG_UNLOCK(dev) sx_xunlock(&(dev)->congestion.lock)
58 #define CONG_LOCKED(dev) sx_xlocked(&(dev)->congestion.lock)
282 dev->congestion.arg[x] = mlx5_get_cc_param_val(field, x); in mlx5_ib_get_all_cc_params()
329 value = dev->congestion.arg[arg2]; in mlx5_ib_cong_params_handler()
333 value == dev->congestion.arg[arg2]) in mlx5_ib_cong_params_handler()
337 dev->congestion.arg[arg2] = value; in mlx5_ib_cong_params_handler()
345 dev->congestion.arg[arg2]); in mlx5_ib_cong_params_handler()
364 dev->congestion.c = MLX5_GET(query_cong_status_out, out, field); \ in mlx5_ib_get_all_cc_status()
382 value = dev->congestion.arg[arg2]; in mlx5_ib_cong_status_handler()
[all …]
|
| /freebsd/share/man/man9/ |
| H A D | mod_cc.9 | 39 .Nd Modular Congestion Control
49 framework allows congestion control algorithms to be implemented as dynamically
145 It should be implemented to initialise congestion control algorithm variables
153 argument as an input to their congestion management algorithms.
161 function is called when a congestion event is detected by the TCP stack.
164 argument as an input to their congestion management algorithms.
165 The congestion event types currently reported by the stack are CC_ECN, CC_RTO,
167 CC_ECN is reported when the TCP stack receives an explicit congestion notification
177 function is called after the TCP connection has recovered from a congestion event.
202 congestion controller.
[all …]
|
| /freebsd/contrib/ofed/infiniband-diags/man/ |
| H A D | ibccconfig.8 | 5 IBCCCONFIG \- configure congestion control settings
39 supports the configuration of congestion control settings on switches
56 Specify a congestion control (CC) key. If none is specified, a key of 0 is used.
173 …itchCongestionSetting 2 0x1F 0x1FFFFFFFFF 0x0 0xF 8 0 0:0 1 # Configure Switch Congestion Settings
174 ibccconfig CACongestionSetting 1 0 0x3 150 1 0 0 # Configure CA Congestion…
175 ibccconfig CACongestionSetting 1 0 0x4 200 1 0 0 # Configure CA Congestion…
176 …olTable 1 63 0 0:0 0:1 ... # Configure first block of Congestion Control Table
177 …lTable 1 127 0 0:64 0:65 ... # Configure second block of Congestion Control Table
|
| H A D | ibccquery.8 | 5 IBCCQUERY \- query congestion control settings/info
39 to congestion control.
55 Specify a congestion control (CC) key. If none is specified, a key of 0 is used.
180 ibccquery CongestionInfo 3 # Congestion Info by lid
181 ibccquery SwitchPortCongestionSetting 3 # Query all Switch Port Congestion Settings
182 ibccquery SwitchPortCongestionSetting 3 1 # Query Switch Port Congestion Setting for port 1
|
| /freebsd/sys/netinet/cc/ |
| H A D | cc.c | 93 MALLOC_DEFINE(M_CC_MEM, "CC Mem", "Congestion Control State memory");
354 * The first loaded congestion control module will become in cc_register_algo()
389 * Perform any necessary tasks before we exit congestion recovery.
428 * congestion window is no longer current and we have to reduce it to in newreno_cc_after_idle()
452 * Get a new congestion window size on a multiplicative decrease event.
461 * Other TCP congestion controls use newreno_cong_signal(), but in newreno_cc_cwnd_on_multiplicative_decrease()
471 * Perform any necessary tasks before we enter congestion recovery.
482 ("%s: congestion signal type 0x%08x is private\n", __func__, type)); in newreno_cc_cong_signal()
527 ("congestion control state not in congestion avoidance\n")); in newreno_cc_cwnd_in_cong_avoid()
530 * Regular in-order ACK, open the congestion window. in newreno_cc_cwnd_in_cong_avoid()
[all …]
|
| H A D | cc_chd.c | 41 * An implementation of the CAIA-Hamilton delay based congestion control
43 * TCP congestion control" by D. A. Hayes and G. Armitage., in 35th Annual IEEE
84 * Private signal type for rate based congestion signal.
102 * Shadow window - keeps track of what the NewReno congestion window
214 } else { /* Congestion avoidance. */ in chd_window_increase()
236 * congestion. However, window increases are only performed when
256 * is delay based congestion. in chd_ack_received()
265 * Probabilistic delay based congestion in chd_ack_received()
300 * Send delay-based congestion signal to the congestion signal in chd_ack_received()
361 * Only react to loss as a congestion signal if qdly > in chd_cong_signal()
[all …]
|
| H A D | cc_cubic.h | 80 #define CUBICFLAG_CONG_EVENT 0x00000001 /* congestion experienced */
95 /* Size of cwnd just before cwnd was reduced in the last congestion event */
97 /* An estimate for the congestion window in the Reno-friendly region */
99 /* The cwnd at the beginning of the current congestion avoidance stage */
104 * was reduced in the last congestion event
111 /* Mean observed rtt between congestion epochs. */
113 /* ACKs since last congestion event. */
267 * congestion event. RTT should be the average RTT estimate for the path
268 * measured over the previous congestion epoch and wmax is the value of cwnd at
269 * the last congestion event. The "TCP friendly" concept in the CUBIC I-D is
[all …]
|
| H A D | cc_cdg.c | 37 * CAIA Delay-Gradient (CDG) congestion control algorithm
39 * An implemention of the delay-gradient congestion control algorithm proposed
42 * D. A. Hayes and G. Armitage, "Revisiting TCP Congestion Control using Delay
85 /* Private delay-gradient induced congestion control signal. */
121 /* rttcount for window increase when in congestion avoidance */
131 /* consecutive congestion episode counter */
430 /* Congestion avoidance. */ in cdg_window_increase()
470 * If already responding to congestion OR we have guessed no in cdg_cong_signal()
479 * Loss is likely to be congestion related. We have in cdg_cong_signal()
582 int congestion, new_measurement, slowstart; in cdg_ack_received() local
[all …]
|
| /freebsd/contrib/ofed/opensm/include/opensm/ |
| H A D | osm_subnet.h | 157 * Subnet Congestion Control Table entry. See A10.2.2.1.1 for format details.
181 * Subnet CA Congestion entry. See A10.4.3.8.4 for format details.
662 * Boolean that specifies whether OpenSM congestion control configuration
666 * CCkey to use when configuring congestion control.
672 * Congestion Control Switch Congestion Setting Control Map
676 * Congestion Control Switch Congestion Setting Victim Mask
680 * Congestion Control Switch Congestion Setting Credit Mask
684 * Congestion Control Switch Congestion Setting Threshold
688 * Congestion Control Switch Congestion Setting Packet Size
692 * Congestion Control Switch Congestion Setting Credit Starvation Threshold
[all …]
|
| /freebsd/contrib/ofed/infiniband-diags/src/ |
| H A D | ibccquery.c | 107 return "congestion info query failed"; in congestion_info()
122 return "congestion key info query failed"; in congestion_key_info()
139 return "congestion log query failed"; in congestion_log()
193 return "switch congestion setting query failed"; in switch_congestion_setting()
231 return "switch port congestion setting query failed"; in switch_port_congestion_setting()
248 return "switch port congestion setting query failed"; in switch_port_congestion_setting()
271 return "ca congestion setting query failed"; in ca_congestion_setting()
296 return "congestion control table query failed"; in congestion_control_table()
313 return "congestion control table query failed"; in congestion_control_table()
378 "CongestionInfo 3\t\t\t# Congestion Info by lid", in main()
[all …]
|
| H A D | ibccconfig.c | 173 return "congestion key info config failed"; in congestion_key_info()
335 return "switch congestion setting config failed"; in switch_congestion_setting()
393 return "switch port congestion setting query failed"; in switch_port_congestion_setting()
419 return "switch port congestion setting config failed"; in switch_port_congestion_setting()
493 return "ca congestion setting config failed"; in ca_congestion_setting()
542 return "congestion control table config failed"; in congestion_control_table()
583 …"SwitchCongestionSetting 2 0x1F 0x1FFFFFFFFF 0x0 0xF 8 0 0:0 1\t# Configure Switch Congestion Sett… in main()
584 "CACongestionSetting 1 0 0x3 150 1 0 0\t\t# Configure CA Congestion Settings to SL 0 and SL 1", in main()
585 "CACongestionSetting 1 0 0x4 200 1 0 0\t\t# Configure CA Congestion Settings to SL 2", in main()
586 "CongestionControlTable 1 63 0 0:0 0:1 ...\t# Configure first block of Congestion Control Table", in main()
[all …]
|
| /freebsd/usr.sbin/tcpdrop/ |
| H A D | tcpdrop.8 | 63 will attempt to drop all connections using the TCP congestion control algorithm
98 will attempt to drop all TCP connections using the congestion control algorithm
153 is causing congestion on a network link, one can drop the TCP session
173 To drop all TCP connections using the new-reno congestion control algorithm use:
189 To drop all TCP connections using the congestion control algorithm new-reno and
|
| /freebsd/usr.sbin/tcpsso/ |
| H A D | tcpsso.8 | 147 congestion control algorithm
194 will apply the socket option to all TCP endpoints using the congestion
242 The following command will set the congestion control module of all TCP
243 endpoints currently using cubic as its congestion control algorithm to the
244 congestion control algorithm newreno:
|