aboutsummaryrefslogtreecommitdiffstats
path: root/src/plugins/lb/lb.h
blob: c6773a4108fae8738774672cf130466c4fefe611 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
/*
 * Copyright (c) 2016 Cisco and/or its affiliates.
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at:
 *
 *     http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */

/**
 * lb-plugin implements a MagLev-like load balancer.
 * http://research.google.com/pubs/pub44824.html
 *
 * It hasn't been tested for interoperability with the original MagLev
 * but intends to provide similar functionality.
 * The load-balancer receives traffic destined to VIP (Virtual IP)
 * addresses from one or multiple(ECMP) routers.
 * The load-balancer tunnels the traffic toward many application servers
 * ensuring session stickyness (i.e. that a single sessions is tunneled
 * towards a single application server).
 *
 */

#ifndef LB_PLUGIN_LB_LB_H_
#define LB_PLUGIN_LB_LB_H_

#include <lb/util.h>
#include <vnet/util/refcount.h>

#include <vnet/vnet.h>
#include <vnet/ip/ip.h>
#include <vnet/dpo/dpo.h>
#include <vnet/fib/fib_table.h>
#include <vppinfra/hash.h>
#include <vppinfra/bihash_8_8.h>
#include <vppinfra/bihash_24_8.h>
#include <lb/lbhash.h>

#define LB_DEFAULT_PER_CPU_STICKY_BUCKETS 1 << 10
#define LB_DEFAULT_FLOW_TIMEOUT 40
#define LB_MAPPING_BUCKETS  1024
#define LB_MAPPING_MEMORY_SIZE  64<<20

typedef enum {
  LB_NEXT_DROP,
  LB_N_NEXT,
} lb_next_t;

typedef enum {
  LB_NAT4_IN2OUT_NEXT_DROP,
  LB_NAT4_IN2OUT_NEXT_LOOKUP,
  LB_NAT4_IN2OUT_N_NEXT,
} LB_nat4_in2out_next_t;

typedef enum {
  LB_NAT6_IN2OUT_NEXT_DROP,
  LB_NAT6_IN2OUT_NEXT_LOOKUP,
  LB_NAT6_IN2OUT_N_NEXT,
} LB_nat6_in2out_next_t;

#define foreach_lb_nat_in2out_error                       \
_(UNSUPPORTED_PROTOCOL, "Unsupported protocol")         \
_(IN2OUT_PACKETS, "Good in2out packets processed")      \
_(NO_TRANSLATION, "No translation")

typedef enum {
#define _(sym,str) LB_NAT_IN2OUT_ERROR_##sym,
  foreach_lb_nat_in2out_error
#undef _
  LB_NAT_IN2OUT_N_ERROR,
} lb_nat_in2out_error_t;

/**
 * lb for kube-proxy supports three types of service
 */
typedef enum {
  LB_SRV_TYPE_CLUSTERIP,
  LB_SRV_TYPE_NODEPORT,
  LB_SRV_N_TYPES,
} lb_svr_type_t;

typedef enum {
  LB4_NODEPORT_NEXT_IP4_NAT4,
  LB4_NODEPORT_NEXT_DROP,
  LB4_NODEPORT_N_NEXT,
} lb4_nodeport_next_t;

typedef enum {
  LB6_NODEPORT_NEXT_IP6_NAT6,
  LB6_NODEPORT_NEXT_DROP,
  LB6_NODEPORT_N_NEXT,
} lb6_nodeport_next_t;

/**
 * Each VIP is configured with a set of
 * application server.
 */
typedef struct {
  /**
   * Registration to FIB event.
   */
  fib_node_t fib_node;

  /**
   * Destination address used to tunnel traffic towards
   * that application server.
   * The address is also used as ID and pseudo-random
   * seed for the load-balancing process.
   */
  ip46_address_t address;

  /**
   * ASs are indexed by address and VIP Index.
   * Which means there will be duplicated if the same server
   * address is used for multiple VIPs.
   */
  u32 vip_index;

  /**
   * Some per-AS flags.
   * For now only LB_AS_FLAGS_USED is defined.
   */
  u8 flags;

#define LB_AS_FLAGS_USED 0x1

  /**
   * Rotating timestamp of when LB_AS_FLAGS_USED flag was last set.
   *
   * AS removal is based on garbage collection and reference counting.
   * When an AS is removed, there is a race between configuration core
   * and worker cores which may still add a reference while it should not
   * be used. This timestamp is used to not remove the AS while a race condition
   * may happen.
   */
  u32 last_used;

  /**
   * The FIB entry index for the next-hop
   */
  fib_node_index_t next_hop_fib_entry_index;

  /**
   * The child index on the FIB entry
   */
  u32 next_hop_child_index;

  /**
   * The next DPO in the graph to follow.
   */
  dpo_id_t dpo;

} lb_as_t;

format_function_t format_lb_as;

typedef struct {
  u32 as_index;
} lb_new_flow_entry_t;

#define lb_foreach_vip_counter \
 _(NEXT_PACKET, "packet from existing sessions", 0) \
 _(FIRST_PACKET, "first session packet", 1) \
 _(UNTRACKED_PACKET, "untracked packet", 2) \
 _(NO_SERVER, "no server configured", 3)

typedef enum {
#define _(a,b,c) LB_VIP_COUNTER_##a = c,
  lb_foreach_vip_counter
#undef _
  LB_N_VIP_COUNTERS
} lb_vip_counter_t;

typedef enum {
  LB_ENCAP_TYPE_GRE4,
  LB_ENCAP_TYPE_GRE6,
  LB_ENCAP_TYPE_L3DSR,
  LB_ENCAP_TYPE_NAT4,
  LB_ENCAP_TYPE_NAT6,
  LB_ENCAP_N_TYPES,
} lb_encap_type_t;

/**
 * The load balancer supports IPv4 and IPv6 traffic
 * and GRE4, GRE6, L3DSR and NAT4, NAT6 encap.
 */
typedef enum {
  LB_VIP_TYPE_IP6_GRE6,
  LB_VIP_TYPE_IP6_GRE4,
  LB_VIP_TYPE_IP4_GRE6,
  LB_VIP_TYPE_IP4_GRE4,
  LB_VIP_TYPE_IP4_L3DSR,
  LB_VIP_TYPE_IP4_NAT4,
  LB_VIP_TYPE_IP6_NAT6,
  LB_VIP_N_TYPES,
} lb_vip_type_t;

format_function_t format_lb_vip_type;
unformat_function_t unformat_lb_vip_type;


/* args for different vip encap types */
typedef struct {
  union
  {
    struct
    {
      /* Service type. clusterip or nodeport */
      u8 srv_type;

      /* Service port. network byte order */
      u16 port;

      /* Pod's port corresponding to specific service. network byte order */
      u16 target_port;

      /* Node's port, can access service via NodeIP:node_port. network byte order */
      u16 node_port;
    };
    /* DSCP bits for L3DSR */
    u8 dscp;
    u64 as_u64;
  };
} lb_vip_encap_args_t;

/**
 * Load balancing service is provided per VIP.
 * In this data model, a VIP can be a whole prefix.
 * But load balancing only
 * occurs on a per-source-address/port basis. Meaning that if a given source
 * reuses the same port for multiple destinations within the same VIP,
 * they will be considered as a single flow.
 */
typedef struct {

  //Runtime

  /**
   * Vector mapping (flow-hash & new_connect_table_mask) to AS index.
   * This is used for new flows.
   */
  lb_new_flow_entry_t *new_flow_table;

  /**
   * New flows table length - 1
   * (length MUST be a power of 2)
   */
  u32 new_flow_table_mask;

  /**
   * Last time garbage collection was run to free the ASs.
   */
  u32 last_garbage_collection;

  //Not runtime

  /**
   * A Virtual IP represents a given service delivered
   * by a set of application servers. It can be a single
   * address or a prefix.
   * IPv4 prefixes are encoded using IPv4-in-IPv6 embedded address
   * (i.e. ::/96 prefix).
   */
  ip46_address_t prefix;

  /**
   * The VIP prefix length.
   * In case of IPv4, plen = 96 + ip4_plen.
   */
  u8 plen;

  /**
   * The type of traffic for this.
   * LB_TYPE_UNDEFINED if unknown.
   */
  lb_vip_type_t type;

  /* args for different vip encap types */
  lb_vip_encap_args_t encap_args;

  /**
   * Flags related to this VIP.
   * LB_VIP_FLAGS_USED means the VIP is active.
   * When it is not set, the VIP in the process of being removed.
   * We cannot immediately remove a VIP because the VIP index still may be stored
   * in the adjacency index.
   */
  u8 flags;
#define LB_VIP_FLAGS_USED 0x1

  /**
   * Pool of AS indexes used for this VIP.
   * This also includes ASs that have been removed (but are still referenced).
   */
  u32 *as_indexes;
} lb_vip_t;

#define lb_vip_is_ip4(vip) ((vip)->type == LB_VIP_TYPE_IP4_GRE6 \
                            || (vip)->type == LB_VIP_TYPE_IP4_GRE4 \
                            || (vip)->type == LB_VIP_TYPE_IP4_L3DSR \
                            || (vip)->type == LB_VIP_TYPE_IP4_NAT4 )

#define lb_vip_is_gre4(vip) ((vip)->type == LB_VIP_TYPE_IP6_GRE4 \
                             || (vip)->type == LB_VIP_TYPE_IP4_GRE4)

#define lb_vip_is_gre6(vip) ((vip)->type == LB_VIP_TYPE_IP6_GRE6 \
                             || (vip)->type == LB_VIP_TYPE_IP4_GRE6)

#define lb_encap_is_ip4(vip) ((vip)->type == LB_VIP_TYPE_IP6_GRE4 \
                             || (vip)->type == LB_VIP_TYPE_IP4_GRE4 \
                             || (vip)->type == LB_VIP_TYPE_IP4_L3DSR \
                             || (vip)->type == LB_VIP_TYPE_IP4_NAT4 )

always_inline bool
lb_vip_is_l3dsr(const lb_vip_t *vip)
{
  return vip->type == LB_VIP_TYPE_IP4_L3DSR;
}
always_inline bool
lb_vip_is_nat4(const lb_vip_t *vip)
{
  return vip->type == LB_VIP_TYPE_IP4_NAT4;
}
always_inline bool
lb_vip_is_nat6(const lb_vip_t *vip)
{
  return vip->type == LB_VIP_TYPE_IP6_NAT6;
}

format_function_t format_lb_vip;
format_function_t format_lb_vip_detailed;

#define foreach_lb_nat_protocol \
  _(UDP, 0, udp, "udp")       \
  _(TCP, 1, tcp, "tcp")

typedef enum {
#define _(N, i, n, s) LB_NAT_PROTOCOL_##N = i,
  foreach_lb_nat_protocol
#undef _
} lb_nat_protocol_t;

always_inline u32
lb_ip_proto_to_nat_proto (u8 ip_proto)
{
  u32 nat_proto = ~0;

  nat_proto = (ip_proto == IP_PROTOCOL_UDP) ? LB_NAT_PROTOCOL_UDP : nat_proto;
  nat_proto = (ip_proto == IP_PROTOCOL_TCP) ? LB_NAT_PROTOCOL_TCP : nat_proto;

  return nat_proto;
}

/* Key for Pod's egress SNAT */
typedef struct {
  union
  {
    struct
    {
      ip4_address_t addr;
      u16 port;
      u16 protocol:3,
          fib_index:13;
    };
    u64 as_u64;
  };
} lb_snat4_key_t;

typedef struct
{
  union
  {
    struct
    {
      ip6_address_t addr;
      u16 port;
      u16 protocol;
      u32 fib_index;
    };
    u64 as_u64[3];
  };
} lb_snat6_key_t;

typedef struct {
  /**
   * for vip + port case, src_ip = vip;
   * for node ip + node_port, src_ip = node_ip
   */
  ip46_address_t src_ip;
  ip46_address_t as_ip;
  u8 src_ip_is_ipv6;
  u8 as_ip_is_ipv6;
  /**
   * Network byte order
   * for vip + port case, src_port = port;
   * for node ip + node_port, src_port = node_port
   */
  u16 src_port;
  u16 target_port; /* Network byte order */
  u32 vrf_id;
  u32 fib_index;
} lb_snat_mapping_t;

typedef struct {
  /**
   * Each CPU has its own sticky flow hash table.
   * One single table is used for all VIPs.
   */
  lb_hash_t *sticky_ht;
} lb_per_cpu_t;

typedef struct {
  /**
   * Pool of all Virtual IPs
   */
  lb_vip_t *vips;

  /**
   * Pool of ASs.
   * ASs are referenced by address and vip index.
   * The first element (index 0) is special and used only to fill
   * new_flow_tables when no AS has been configured.
   */
  lb_as_t *ass;

  /**
   * Each AS has an associated reference counter.
   * As ass[0] has a special meaning, its associated counter
   * starts at 0 and is decremented instead. i.e. do not use it.
   */
  vlib_refcount_t as_refcount;

  /* hash lookup vip_index by key: {u16: nodeport} */
  uword * vip_index_by_nodeport;

  /**
   * Some global data is per-cpu
   */
  lb_per_cpu_t *per_cpu;

  /**
   * Node next index for IP adjacencies, for each of the traffic types.
   */
  u32 ip_lookup_next_index[LB_VIP_N_TYPES];

  /**
   * Source address used in IPv6 encapsulated traffic
   */
  ip6_address_t ip6_src_address;

  /**
   * Source address used for IPv4 encapsulated traffic
   */
  ip4_address_t ip4_src_address;

  /**
   * Number of buckets in the per-cpu sticky hash table.
   */
  u32 per_cpu_sticky_buckets;

  /**
   * Flow timeout in seconds.
   */
  u32 flow_timeout;

  /**
   * Per VIP counter
   */
  vlib_simple_counter_main_t vip_counters[LB_N_VIP_COUNTERS];

  /**
   * DPO used to send packet from IP4/6 lookup to LB node.
   */
  dpo_type_t dpo_gre4_type;
  dpo_type_t dpo_gre6_type;
  dpo_type_t dpo_l3dsr_type;
  dpo_type_t dpo_nat4_type;
  dpo_type_t dpo_nat6_type;

  /**
   * Node type for registering to fib changes.
   */
  fib_node_type_t fib_node_type;

  /* Find a static mapping by AS IP : target_port */
  clib_bihash_8_8_t mapping_by_as4;
  clib_bihash_24_8_t mapping_by_as6;

  /* Static mapping pool */
  lb_snat_mapping_t * snat_mappings;

  /**
   * API dynamically registered base ID.
   */
  u16 msg_id_base;

  volatile u32 *writer_lock;

  /* convenience */
  vlib_main_t *vlib_main;
  vnet_main_t *vnet_main;
} lb_main_t;

/* args for different vip encap types */
typedef struct {
  ip46_address_t prefix;
  u8 plen;
  lb_vip_type_t type;
  u32 new_length;
  lb_vip_encap_args_t encap_args;
} lb_vip_add_args_t;

extern lb_main_t lb_main;
extern vlib_node_registration_t lb4_node;
extern vlib_node_registration_t lb6_node;
extern vlib_node_registration_t lb4_nodeport_node;
extern vlib_node_registration_t lb6_nodeport_node;
extern vlib_node_registration_t lb_nat4_in2out_node;
extern vlib_node_registration_t lb_nat6_in2out_node;

/**
 * Fix global load-balancer parameters.
 * @param ip4_address IPv4 source address used for encapsulated traffic
 * @param ip6_address IPv6 source address used for encapsulated traffic
 * @return 0 on success. VNET_LB_ERR_XXX on error
 */
int lb_conf(ip4_address_t *ip4_address, ip6_address_t *ip6_address,
            u32 sticky_buckets, u32 flow_timeout);

int lb_vip_add(lb_vip_add_args_t args, u32 *vip_index);

int lb_vip_del(u32 vip_index);

int lb_vip_find_index(ip46_address_t *prefix, u8 plen, u32 *vip_index);

#define lb_vip_get_by_index(index) (pool_is_free_index(lb_main.vips, index)?NULL:pool_elt_at_index(lb_main.vips, index))

int lb_vip_add_ass(u32 vip_index, ip46_address_t *addresses, u32 n);
int lb_vip_del_ass(u32 vip_index, ip46_address_t *addresses, u32 n);

u32 lb_hash_time_now(vlib_main_t * vm);

void lb_garbage_collection();

int lb_nat4_interface_add_del (u32 sw_if_index, int is_del);
int lb_nat6_interface_add_del (u32 sw_if_index, int is_del);

format_function_t format_lb_main;

#endif /* LB_PLUGIN_LB_LB_H_ */