2 # IP Virtual Server configuration
5 tristate "IP virtual server support"
6 depends on NET && INET && NETFILTER
7 depends on (NF_CONNTRACK || NF_CONNTRACK=n)
9 IP Virtual Server support will let you build a high-performance
10 virtual server based on cluster of two or more real servers. This
11 option must be enabled for at least one of the clustered computers
12 that will take care of intercepting incoming connections to a
13 single IP address and scheduling them to real servers.
15 Three request dispatching techniques are implemented, they are
16 virtual server via NAT, virtual server via tunneling and virtual
17 server via direct routing. The several scheduling algorithms can
18 be used to choose which server the connection is directed to,
19 thus load balancing can be achieved among the servers. For more
20 information and its administration program, please visit the
21 following URL: <http://www.linuxvirtualserver.org/>.
23 If you want to compile it in kernel, say Y. To compile it as a
24 module, choose M here. If unsure, say N.
29 bool "IPv6 support for IPVS"
30 depends on IPV6 = y || IP_VS = IPV6
32 Add IPv6 support to IPVS. This is incomplete and might be dangerous.
34 See http://www.mindbasket.com/ipvs for more information.
39 bool "IP virtual server debugging"
41 Say Y here if you want to get additional messages useful in
42 debugging the IP virtual server code. You can change the debug
43 level in /proc/sys/net/ipv4/vs/debug_level
46 int "IPVS connection table size (the Nth power of 2)"
50 The IPVS connection hash table uses the chaining scheme to handle
51 hash collisions. Using a big IPVS connection hash table will greatly
52 reduce conflicts when there are hundreds of thousands of connections
55 Note the table size must be power of 2. The table size will be the
56 value of 2 to the your input number power. The number to choose is
57 from 8 to 20, the default number is 12, which means the table size
58 is 4096. Don't input the number too small, otherwise you will lose
59 performance on it. You can adapt the table size yourself, according
60 to your virtual server application. It is good to set the table size
61 not far less than the number of connections per second multiplying
62 average lasting time of connection in the table. For example, your
63 virtual server gets 200 connections per second, the connection lasts
64 for 200 seconds in average in the connection table, the table size
65 should be not far less than 200x200, it is good to set the table
68 Another note that each connection occupies 128 bytes effectively and
69 each hash entry uses 8 bytes, so you can estimate how much memory is
72 You can overwrite this number setting conn_tab_bits module parameter
73 or by appending ip_vs.conn_tab_bits=? to the kernel command line
74 if IP VS was compiled built-in.
76 comment "IPVS transport protocol load balancing support"
78 config IP_VS_PROTO_TCP
79 bool "TCP load balancing support"
81 This option enables support for load balancing TCP transport
82 protocol. Say Y if unsure.
84 config IP_VS_PROTO_UDP
85 bool "UDP load balancing support"
87 This option enables support for load balancing UDP transport
88 protocol. Say Y if unsure.
90 config IP_VS_PROTO_AH_ESP
91 def_bool IP_VS_PROTO_ESP || IP_VS_PROTO_AH
93 config IP_VS_PROTO_ESP
94 bool "ESP load balancing support"
96 This option enables support for load balancing ESP (Encapsulation
97 Security Payload) transport protocol. Say Y if unsure.
100 bool "AH load balancing support"
102 This option enables support for load balancing AH (Authentication
103 Header) transport protocol. Say Y if unsure.
105 config IP_VS_PROTO_SCTP
106 bool "SCTP load balancing support"
109 This option enables support for load balancing SCTP transport
110 protocol. Say Y if unsure.
112 comment "IPVS scheduler"
115 tristate "round-robin scheduling"
117 The robin-robin scheduling algorithm simply directs network
118 connections to different real servers in a round-robin manner.
120 If you want to compile it in kernel, say Y. To compile it as a
121 module, choose M here. If unsure, say N.
124 tristate "weighted round-robin scheduling"
126 The weighted robin-robin scheduling algorithm directs network
127 connections to different real servers based on server weights
128 in a round-robin manner. Servers with higher weights receive
129 new connections first than those with less weights, and servers
130 with higher weights get more connections than those with less
131 weights and servers with equal weights get equal connections.
133 If you want to compile it in kernel, say Y. To compile it as a
134 module, choose M here. If unsure, say N.
137 tristate "least-connection scheduling"
139 The least-connection scheduling algorithm directs network
140 connections to the server with the least number of active
143 If you want to compile it in kernel, say Y. To compile it as a
144 module, choose M here. If unsure, say N.
147 tristate "weighted least-connection scheduling"
149 The weighted least-connection scheduling algorithm directs network
150 connections to the server with the least active connections
151 normalized by the server weight.
153 If you want to compile it in kernel, say Y. To compile it as a
154 module, choose M here. If unsure, say N.
157 tristate "locality-based least-connection scheduling"
159 The locality-based least-connection scheduling algorithm is for
160 destination IP load balancing. It is usually used in cache cluster.
161 This algorithm usually directs packet destined for an IP address to
162 its server if the server is alive and under load. If the server is
163 overloaded (its active connection numbers is larger than its weight)
164 and there is a server in its half load, then allocate the weighted
165 least-connection server to this IP address.
167 If you want to compile it in kernel, say Y. To compile it as a
168 module, choose M here. If unsure, say N.
171 tristate "locality-based least-connection with replication scheduling"
173 The locality-based least-connection with replication scheduling
174 algorithm is also for destination IP load balancing. It is
175 usually used in cache cluster. It differs from the LBLC scheduling
176 as follows: the load balancer maintains mappings from a target
177 to a set of server nodes that can serve the target. Requests for
178 a target are assigned to the least-connection node in the target's
179 server set. If all the node in the server set are over loaded,
180 it picks up a least-connection node in the cluster and adds it
181 in the sever set for the target. If the server set has not been
182 modified for the specified time, the most loaded node is removed
183 from the server set, in order to avoid high degree of replication.
185 If you want to compile it in kernel, say Y. To compile it as a
186 module, choose M here. If unsure, say N.
189 tristate "destination hashing scheduling"
191 The destination hashing scheduling algorithm assigns network
192 connections to the servers through looking up a statically assigned
193 hash table by their destination IP addresses.
195 If you want to compile it in kernel, say Y. To compile it as a
196 module, choose M here. If unsure, say N.
199 tristate "source hashing scheduling"
201 The source hashing scheduling algorithm assigns network
202 connections to the servers through looking up a statically assigned
203 hash table by their source IP addresses.
205 If you want to compile it in kernel, say Y. To compile it as a
206 module, choose M here. If unsure, say N.
209 tristate "shortest expected delay scheduling"
211 The shortest expected delay scheduling algorithm assigns network
212 connections to the server with the shortest expected delay. The
213 expected delay that the job will experience is (Ci + 1) / Ui if
214 sent to the ith server, in which Ci is the number of connections
215 on the ith server and Ui is the fixed service rate (weight)
218 If you want to compile it in kernel, say Y. To compile it as a
219 module, choose M here. If unsure, say N.
222 tristate "never queue scheduling"
224 The never queue scheduling algorithm adopts a two-speed model.
225 When there is an idle server available, the job will be sent to
226 the idle server, instead of waiting for a fast one. When there
227 is no idle server available, the job will be sent to the server
228 that minimize its expected delay (The Shortest Expected Delay
229 scheduling algorithm).
231 If you want to compile it in kernel, say Y. To compile it as a
232 module, choose M here. If unsure, say N.
234 comment 'IPVS SH scheduler'
236 config IP_VS_SH_TAB_BITS
237 int "IPVS source hashing table size (the Nth power of 2)"
241 The source hashing scheduler maps source IPs to destinations
242 stored in a hash table. This table is tiled by each destination
243 until all slots in the table are filled. When using weights to
244 allow destinations to receive more connections, the table is
245 tiled an amount proportional to the weights specified. The table
246 needs to be large enough to effectively fit all the destinations
247 multiplied by their respective weights.
249 comment 'IPVS application helper'
252 tristate "FTP protocol helper"
253 depends on IP_VS_PROTO_TCP && NF_CONNTRACK && NF_NAT
256 FTP is a protocol that transfers IP address and/or port number in
257 the payload. In the virtual server via Network Address Translation,
258 the IP address and port number of real servers cannot be sent to
259 clients in ftp connections directly, so FTP protocol helper is
260 required for tracking the connection and mangling it back to that of
263 If you want to compile it in kernel, say Y. To compile it as a
264 module, choose M here. If unsure, say N.
267 bool "Netfilter connection tracking"
268 depends on NF_CONNTRACK
270 The Netfilter connection tracking support allows the IPVS
271 connection state to be exported to the Netfilter framework
272 for filtering purposes.
275 tristate "SIP persistence engine"
276 depends on IP_VS_PROTO_UDP
277 depends on NF_CONNTRACK_SIP
279 Allow persistence based on the SIP Call-ID