]> git.proxmox.com Git - ceph.git/blame - ceph/src/dpdk/doc/guides/prog_guide/packet_framework.rst
projects / ceph.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
bump version to 12.2.12-pve1
[ceph.git] / ceph / src / dpdk / doc / guides / prog_guide / packet_framework.rst
CommitLineData
7c673cae
FG		 1.. BSD LICENSE
2 Copyright(c) 2010-2014 Intel Corporation. All rights reserved.
3 All rights reserved.
4
5 Redistribution and use in source and binary forms, with or without
6 modification, are permitted provided that the following conditions
7 are met:
8
9 * Redistributions of source code must retain the above copyright
10 notice, this list of conditions and the following disclaimer.
11 * Redistributions in binary form must reproduce the above copyright
12 notice, this list of conditions and the following disclaimer in
13 the documentation and/or other materials provided with the
14 distribution.
15 * Neither the name of Intel Corporation nor the names of its
16 contributors may be used to endorse or promote products derived
17 from this software without specific prior written permission.
18
19 THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
20 "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
21 LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
22 A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
23 OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
24 SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
25 LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
26 DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
27 THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
28 (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
29 OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
30
31Packet Framework
32================
33
34Design Objectives
35-----------------
36
37The main design objectives for the DPDK Packet Framework are:
38
39* Provide standard methodology to build complex packet processing pipelines.
40 Provide reusable and extensible templates for the commonly used pipeline functional blocks;
41
42* Provide capability to switch between pure software and hardware-accelerated implementations for the same pipeline functional block;
43
44* Provide the best trade-off between flexibility and performance.
45 Hardcoded pipelines usually provide the best performance, but are not flexible,
46 while developing flexible frameworks is never a problem, but performance is usually low;
47
48* Provide a framework that is logically similar to Open Flow.
49
50Overview
51--------
52
53Packet processing applications are frequently structured as pipelines of multiple stages,
54with the logic of each stage glued around a lookup table.
55For each incoming packet, the table defines the set of actions to be applied to the packet,
56as well as the next stage to send the packet to.
57
58The DPDK Packet Framework minimizes the development effort required to build packet processing pipelines
59by defining a standard methodology for pipeline development,
60as well as providing libraries of reusable templates for the commonly used pipeline blocks.
61
62The pipeline is constructed by connecting the set of input ports with the set of output ports
63through the set of tables in a tree-like topology.
64As result of lookup operation for the current packet in the current table,
65one of the table entries (on lookup hit) or the default table entry (on lookup miss)
66provides the set of actions to be applied on the current packet,
67as well as the next hop for the packet, which can be either another table, an output port or packet drop.
68
69An example of packet processing pipeline is presented in :numref:`figure_figure32`:
70
71.. _figure_figure32:
72
73.. figure:: img/figure32.*
74
75 Example of Packet Processing Pipeline where Input Ports 0 and 1
76 are Connected with Output Ports 0, 1 and 2 through Tables 0 and 1
77
78
79Port Library Design
80-------------------
81
82Port Types
83~~~~~~~~~~
84
85:numref:`table_qos_19` is a non-exhaustive list of ports that can be implemented with the Packet Framework.
86
87.. _table_qos_19:
88
89.. table:: Port Types
90
91 +---+------------------+---------------------------------------------------------------------------------------+
92 | # | Port type | Description |
93 | | | |
94 +===+==================+=======================================================================================+
95 | 1 | SW ring | SW circular buffer used for message passing between the application threads. Uses |
96 | | | the DPDK rte_ring primitive. Expected to be the most commonly used type of |
97 | | | port. |
98 | | | |
99 +---+------------------+---------------------------------------------------------------------------------------+
100 | 2 | HW ring | Queue of buffer descriptors used to interact with NIC, switch or accelerator ports. |
101 | | | For NIC ports, it uses the DPDK rte_eth_rx_queue or rte_eth_tx_queue |
102 | | | primitives. |
103 | | | |
104 +---+------------------+---------------------------------------------------------------------------------------+
105 | 3 | IP reassembly | Input packets are either IP fragments or complete IP datagrams. Output packets are |
106 | | | complete IP datagrams. |
107 | | | |
108 +---+------------------+---------------------------------------------------------------------------------------+
109 | 4 | IP fragmentation | Input packets are jumbo (IP datagrams with length bigger than MTU) or non-jumbo |
110 | | | packets. Output packets are non-jumbo packets. |
111 | | | |
112 +---+------------------+---------------------------------------------------------------------------------------+
113 | 5 | Traffic manager | Traffic manager attached to a specific NIC output port, performing congestion |
114 | | | management and hierarchical scheduling according to pre-defined SLAs. |
115 | | | |
116 +---+------------------+---------------------------------------------------------------------------------------+
117 | 6 | KNI | Send/receive packets to/from Linux kernel space. |
118 | | | |
119 +---+------------------+---------------------------------------------------------------------------------------+
120 | 7 | Source | Input port used as packet generator. Similar to Linux kernel /dev/zero character |
121 | | | device. |
122 | | | |
123 +---+------------------+---------------------------------------------------------------------------------------+
124 | 8 | Sink | Output port used to drop all input packets. Similar to Linux kernel /dev/null |
125 | | | character device. |
126 | | | |
127 +---+------------------+---------------------------------------------------------------------------------------+
128
129Port Interface
130~~~~~~~~~~~~~~
131
132Each port is unidirectional, i.e. either input port or output port.
133Each input/output port is required to implement an abstract interface that
134defines the initialization and run-time operation of the port.
135The port abstract interface is described in.
136
137.. _table_qos_20:
138
139.. table:: 20 Port Abstract Interface
140
141 +---+----------------+-----------------------------------------------------------------------------------------+
142 | # | Port Operation | Description |
143 | | | |
144 +===+================+=========================================================================================+
145 | 1 | Create | Create the low-level port object (e.g. queue). Can internally allocate memory. |
146 | | | |
147 +---+----------------+-----------------------------------------------------------------------------------------+
148 | 2 | Free | Free the resources (e.g. memory) used by the low-level port object. |
149 | | | |
150 +---+----------------+-----------------------------------------------------------------------------------------+
151 | 3 | RX | Read a burst of input packets. Non-blocking operation. Only defined for input ports. |
152 | | | |
153 +---+----------------+-----------------------------------------------------------------------------------------+
154 | 4 | TX | Write a burst of input packets. Non-blocking operation. Only defined for output ports. |
155 | | | |
156 +---+----------------+-----------------------------------------------------------------------------------------+
157 | 5 | Flush | Flush the output buffer. Only defined for output ports. |
158 | | | |
159 +---+----------------+-----------------------------------------------------------------------------------------+
160
161Table Library Design
162--------------------
163
164Table Types
165~~~~~~~~~~~
166
167:numref:`table_qos_21` is a non-exhaustive list of types of tables that can be implemented with the Packet Framework.
168
169.. _table_qos_21:
170
171.. table:: Table Types
172
173 +---+----------------------------+-----------------------------------------------------------------------------+
174 | # | Table Type | Description |
175 | | | |
176 +===+============================+=============================================================================+
177 | 1 | Hash table | Lookup key is n-tuple based. |
178 | | | |
179 | | | Typically, the lookup key is hashed to produce a signature that is used to |
180 | | | identify a bucket of entries where the lookup key is searched next. |
181 | | | |
182 | | | The signature associated with the lookup key of each input packet is either |
183 | | | read from the packet descriptor (pre-computed signature) or computed at |
184 | | | table lookup time. |
185 | | | |
186 | | | The table lookup, add entry and delete entry operations, as well as any |
187 | | | other pipeline block that pre-computes the signature all have to use the |
188 | | | same hashing algorithm to generate the signature. |
189 | | | |
190 | | | Typically used to implement flow classification tables, ARP caches, routing |
191 | | | table for tunnelling protocols, etc. |
192 | | | |
193 +---+----------------------------+-----------------------------------------------------------------------------+
194 | 2 | Longest Prefix Match (LPM) | Lookup key is the IP address. |
195 | | | |
196 | | | Each table entries has an associated IP prefix (IP and depth). |
197 | | | |
198 | | | The table lookup operation selects the IP prefix that is matched by the |
199 | | | lookup key; in case of multiple matches, the entry with the longest prefix |
200 | | | depth wins. |
201 | | | |
202 | | | Typically used to implement IP routing tables. |
203 | | | |
204 +---+----------------------------+-----------------------------------------------------------------------------+
205 | 3 | Access Control List (ACLs) | Lookup key is 7-tuple of two VLAN/MPLS labels, IP destination address, |
206 | | | IP source addresses, L4 protocol, L4 destination port, L4 source port. |
207 | | | |
208 | | | Each table entry has an associated ACL and priority. The ACL contains bit |
209 | | | masks for the VLAN/MPLS labels, IP prefix for IP destination address, IP |
210 | | | prefix for IP source addresses, L4 protocol and bitmask, L4 destination |
211 | | | port and bit mask, L4 source port and bit mask. |
212 | | | |
213 | | | The table lookup operation selects the ACL that is matched by the lookup |
214 | | | key; in case of multiple matches, the entry with the highest priority wins. |
215 | | | |
216 | | | Typically used to implement rule databases for firewalls, etc. |
217 | | | |
218 +---+----------------------------+-----------------------------------------------------------------------------+
219 | 4 | Pattern matching search | Lookup key is the packet payload. |
220 | | | |
221 | | | Table is a database of patterns, with each pattern having a priority |
222 | | | assigned. |
223 | | | |
224 | | | The table lookup operation selects the patterns that is matched by the |
225 | | | input packet; in case of multiple matches, the matching pattern with the |
226 | | | highest priority wins. |
227 | | | |
228 +---+----------------------------+-----------------------------------------------------------------------------+
229 | 5 | Array | Lookup key is the table entry index itself. |
230 | | | |
231 +---+----------------------------+-----------------------------------------------------------------------------+
232
233Table Interface
234~~~~~~~~~~~~~~~
235
236Each table is required to implement an abstract interface that defines the initialization
237and run-time operation of the table.
238The table abstract interface is described in :numref:`table_qos_29_1`.
239
240.. _table_qos_29_1:
241
242.. table:: Table Abstract Interface
243
244 +---+-----------------+----------------------------------------------------------------------------------------+
245 | # | Table operation | Description |
246 | | | |
247 +===+=================+========================================================================================+
248 | 1 | Create | Create the low-level data structures of the lookup table. Can internally allocate |
249 | | | memory. |
250 | | | |
251 +---+-----------------+----------------------------------------------------------------------------------------+
252 | 2 | Free | Free up all the resources used by the lookup table. |
253 | | | |
254 +---+-----------------+----------------------------------------------------------------------------------------+
255 | 3 | Add entry | Add new entry to the lookup table. |
256 | | | |
257 +---+-----------------+----------------------------------------------------------------------------------------+
258 | 4 | Delete entry | Delete specific entry from the lookup table. |
259 | | | |
260 +---+-----------------+----------------------------------------------------------------------------------------+
261 | 5 | Lookup | Look up a burst of input packets and return a bit mask specifying the result of the |
262 | | | lookup operation for each packet: a set bit signifies lookup hit for the corresponding |
263 | | | packet, while a cleared bit a lookup miss. |
264 | | | |
265 | | | For each lookup hit packet, the lookup operation also returns a pointer to the table |
266 | | | entry that was hit, which contains the actions to be applied on the packet and any |
267 | | | associated metadata. |
268 | | | |
269 | | | For each lookup miss packet, the actions to be applied on the packet and any |
270 | | | associated metadata are specified by the default table entry preconfigured for lookup |
271 | | | miss. |
272 | | | |
273 +---+-----------------+----------------------------------------------------------------------------------------+
274
275
276Hash Table Design
277~~~~~~~~~~~~~~~~~
278
279Hash Table Overview
280^^^^^^^^^^^^^^^^^^^
281
282Hash tables are important because the key lookup operation is optimized for speed:
283instead of having to linearly search the lookup key through all the keys in the table,
284the search is limited to only the keys stored in a single table bucket.
285
286**Associative Arrays**
287
288An associative array is a function that can be specified as a set of (key, value) pairs,
289with each key from the possible set of input keys present at most once.
290For a given associative array, the possible operations are:
291
292#. *add (key, value)*: When no value is currently associated with *key*, then the (key, *value* ) association is created.
293 When *key* is already associated value *value0*, then the association (*key*, *value0*) is removed
294 and association *(key, value)* is created;
295
296#. *delete key*: When no value is currently associated with *key*, this operation has no effect.
297 When *key* is already associated *value*, then association *(key, value)* is removed;
298
299#. *lookup key*: When no value is currently associated with *key*, then this operation returns void value (lookup miss).
300 When *key* is associated with *value*, then this operation returns *value*.
301 The *(key, value)* association is not changed.
302
303The matching criterion used to compare the input key against the keys in the associative array is *exact match*,
304as the key size (number of bytes) and the key value (array of bytes) have to match exactly for the two keys under comparison.
305
306**Hash Function**
307
308A hash function deterministically maps data of variable length (key) to data of fixed size (hash value or key signature).
309Typically, the size of the key is bigger than the size of the key signature.
310The hash function basically compresses a long key into a short signature.
311Several keys can share the same signature (collisions).
312
313High quality hash functions have uniform distribution.
314For large number of keys, when dividing the space of signature values into a fixed number of equal intervals (buckets),
315it is desirable to have the key signatures evenly distributed across these intervals (uniform distribution),
316as opposed to most of the signatures going into only a few of the intervals
317and the rest of the intervals being largely unused (non-uniform distribution).
318
319**Hash Table**
320
321A hash table is an associative array that uses a hash function for its operation.
322The reason for using a hash function is to optimize the performance of the lookup operation
323by minimizing the number of table keys that have to be compared against the input key.
324
325Instead of storing the (key, value) pairs in a single list, the hash table maintains multiple lists (buckets).
326For any given key, there is a single bucket where that key might exist, and this bucket is uniquely identified based on the key signature.
327Once the key signature is computed and the hash table bucket identified,
328the key is either located in this bucket or it is not present in the hash table at all,
329so the key search can be narrowed down from the full set of keys currently in the table
330to just the set of keys currently in the identified table bucket.
331
332The performance of the hash table lookup operation is greatly improved,
333provided that the table keys are evenly distributed among the hash table buckets,
334which can be achieved by using a hash function with uniform distribution.
335The rule to map a key to its bucket can simply be to use the key signature (modulo the number of table buckets) as the table bucket ID:
336
337 *bucket_id = f_hash(key) % n_buckets;*
338
339By selecting the number of buckets to be a power of two, the modulo operator can be replaced by a bitwise AND logical operation:
340
341 *bucket_id = f_hash(key) & (n_buckets - 1);*
342
343considering *n_bits* as the number of bits set in *bucket_mask = n_buckets - 1*,
344this means that all the keys that end up in the same hash table bucket have the lower *n_bits* of their signature identical.
345In order to reduce the number of keys in the same bucket (collisions), the number of hash table buckets needs to be increased.
346
347In packet processing context, the sequence of operations involved in hash table operations is described in :numref:`figure_figure33`:
348
349.. _figure_figure33:
350
351.. figure:: img/figure33.*
352
353 Sequence of Steps for Hash Table Operations in a Packet Processing Context
354
355
356
357Hash Table Use Cases
358^^^^^^^^^^^^^^^^^^^^
359
360**Flow Classification**
361
362*Description:* The flow classification is executed at least once for each input packet.
363This operation maps each incoming packet against one of the known traffic flows in the flow database that typically contains millions of flows.
364
365*Hash table name:* Flow classification table
366
367*Number of keys:* Millions
368
369*Key format:* n-tuple of packet fields that uniquely identify a traffic flow/connection.
370Example: DiffServ 5-tuple of (Source IP address, Destination IP address, L4 protocol, L4 protocol source port, L4 protocol destination port).
371For IPv4 protocol and L4 protocols like TCP, UDP or SCTP, the size of the DiffServ 5-tuple is 13 bytes, while for IPv6 it is 37 bytes.
372
373*Key value (key data):* actions and action meta-data describing what processing to be applied for the packets of the current flow.
374The size of the data associated with each traffic flow can vary from 8 bytes to kilobytes.
375
376**Address Resolution Protocol (ARP)**
377
378*Description:* Once a route has been identified for an IP packet (so the output interface and the IP address of the next hop station are known),
379the MAC address of the next hop station is needed in order to send this packet onto the next leg of the journey
380towards its destination (as identified by its destination IP address).
381The MAC address of the next hop station becomes the destination MAC address of the outgoing Ethernet frame.
382
383*Hash table name:* ARP table
384
385*Number of keys:* Thousands
386
387*Key format:* The pair of (Output interface, Next Hop IP address), which is typically 5 bytes for IPv4 and 17 bytes for IPv6.
388
389*Key value (key data):* MAC address of the next hop station (6 bytes).
390
391Hash Table Types
392^^^^^^^^^^^^^^^^
393
394:numref:`table_qos_22` lists the hash table configuration parameters shared by all different hash table types.
395
396.. _table_qos_22:
397
398.. table:: Configuration Parameters Common for All Hash Table Types
399
400 +---+---------------------------+------------------------------------------------------------------------------+
401 | # | Parameter | Details |
402 | | | |
403 +===+===========================+==============================================================================+
404 | 1 | Key size | Measured as number of bytes. All keys have the same size. |
405 | | | |
406 +---+---------------------------+------------------------------------------------------------------------------+
407 | 2 | Key value (key data) size | Measured as number of bytes. |
408 | | | |
409 +---+---------------------------+------------------------------------------------------------------------------+
410 | 3 | Number of buckets | Needs to be a power of two. |
411 | | | |
412 +---+---------------------------+------------------------------------------------------------------------------+
413 | 4 | Maximum number of keys | Needs to be a power of two. |
414 | | | |
415 +---+---------------------------+------------------------------------------------------------------------------+
416 | 5 | Hash function | Examples: jhash, CRC hash, etc. |
417 | | | |
418 +---+---------------------------+------------------------------------------------------------------------------+
419 | 6 | Hash function seed | Parameter to be passed to the hash function. |
420 | | | |
421 +---+---------------------------+------------------------------------------------------------------------------+
422 | 7 | Key offset | Offset of the lookup key byte array within the packet meta-data stored in |
423 | | | the packet buffer. |
424 | | | |
425 +---+---------------------------+------------------------------------------------------------------------------+
426
427Bucket Full Problem
428"""""""""""""""""""
429
430On initialization, each hash table bucket is allocated space for exactly 4 keys.
431As keys are added to the table, it can happen that a given bucket already has 4 keys when a new key has to be added to this bucket.
432The possible options are:
433
434#. **Least Recently Used (LRU) Hash Table.**
435 One of the existing keys in the bucket is deleted and the new key is added in its place.
436 The number of keys in each bucket never grows bigger than 4. The logic to pick the key to be dropped from the bucket is LRU.
437 The hash table lookup operation maintains the order in which the keys in the same bucket are hit, so every time a key is hit,
438 it becomes the new Most Recently Used (MRU) key, i.e. the last candidate for drop.
439 When a key is added to the bucket, it also becomes the new MRU key.
440 When a key needs to be picked and dropped, the first candidate for drop, i.e. the current LRU key, is always picked.
441 The LRU logic requires maintaining specific data structures per each bucket.
442
443#. **Extendable Bucket Hash Table.**
444 The bucket is extended with space for 4 more keys.
445 This is done by allocating additional memory at table initialization time,
446 which is used to create a pool of free keys (the size of this pool is configurable and always a multiple of 4).
447 On key add operation, the allocation of a group of 4 keys only happens successfully within the limit of free keys,
448 otherwise the key add operation fails.
449 On key delete operation, a group of 4 keys is freed back to the pool of free keys
450 when the key to be deleted is the only key that was used within its group of 4 keys at that time.
451 On key lookup operation, if the current bucket is in extended state and a match is not found in the first group of 4 keys,
452 the search continues beyond the first group of 4 keys, potentially until all keys in this bucket are examined.
453 The extendable bucket logic requires maintaining specific data structures per table and per each bucket.
454
455.. _table_qos_23:
456
457.. table:: Configuration Parameters Specific to Extendable Bucket Hash Table
458
459 +---+---------------------------+--------------------------------------------------+
460 | # | Parameter | Details |
461 | | | |
462 +===+===========================+==================================================+
463 | 1 | Number of additional keys | Needs to be a power of two, at least equal to 4. |
464 | | | |
465 +---+---------------------------+--------------------------------------------------+
466
467
468Signature Computation
469"""""""""""""""""""""
470
471The possible options for key signature computation are:
472
473#. **Pre-computed key signature.**
474 The key lookup operation is split between two CPU cores.
475 The first CPU core (typically the CPU core that performs packet RX) extracts the key from the input packet,
476 computes the key signature and saves both the key and the key signature in the packet buffer as packet meta-data.
477 The second CPU core reads both the key and the key signature from the packet meta-data
478 and performs the bucket search step of the key lookup operation.
479
480#. **Key signature computed on lookup ("do-sig" version).**
481 The same CPU core reads the key from the packet meta-data, uses it to compute the key signature
482 and also performs the bucket search step of the key lookup operation.
483
484.. _table_qos_24:
485
486.. table:: Configuration Parameters Specific to Pre-computed Key Signature Hash Table
487
488 +---+------------------+-----------------------------------------------------------------------+
489 | # | Parameter | Details |
490 | | | |
491 +===+==================+=======================================================================+
492 | 1 | Signature offset | Offset of the pre-computed key signature within the packet meta-data. |
493 | | | |
494 +---+------------------+-----------------------------------------------------------------------+
495
496Key Size Optimized Hash Tables
497""""""""""""""""""""""""""""""
498
499For specific key sizes, the data structures and algorithm of key lookup operation can be specially handcrafted for further performance improvements,
500so following options are possible:
501
502#. **Implementation supporting configurable key size.**
503
504#. **Implementation supporting a single key size.**
505 Typical key sizes are 8 bytes and 16 bytes.
506
507Bucket Search Logic for Configurable Key Size Hash Tables
508^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
509
510The performance of the bucket search logic is one of the main factors influencing the performance of the key lookup operation.
511The data structures and algorithm are designed to make the best use of Intel CPU architecture resources like:
512cache memory space, cache memory bandwidth, external memory bandwidth, multiple execution units working in parallel,
513out of order instruction execution, special CPU instructions, etc.
514
515The bucket search logic handles multiple input packets in parallel.
516It is built as a pipeline of several stages (3 or 4), with each pipeline stage handling two different packets from the burst of input packets.
517On each pipeline iteration, the packets are pushed to the next pipeline stage: for the 4-stage pipeline,
518two packets (that just completed stage 3) exit the pipeline,
519two packets (that just completed stage 2) are now executing stage 3, two packets (that just completed stage 1) are now executing stage 2,
520two packets (that just completed stage 0) are now executing stage 1 and two packets (next two packets to read from the burst of input packets)
521are entering the pipeline to execute stage 0.
522The pipeline iterations continue until all packets from the burst of input packets execute the last stage of the pipeline.
523
524The bucket search logic is broken into pipeline stages at the boundary of the next memory access.
525Each pipeline stage uses data structures that are stored (with high probability) into the L1 or L2 cache memory of the current CPU core and
526breaks just before the next memory access required by the algorithm.
527The current pipeline stage finalizes by prefetching the data structures required by the next pipeline stage,
528so given enough time for the prefetch to complete,
529when the next pipeline stage eventually gets executed for the same packets,
530it will read the data structures it needs from L1 or L2 cache memory and thus avoid the significant penalty incurred by L2 or L3 cache memory miss.
531
532By prefetching the data structures required by the next pipeline stage in advance (before they are used)
533and switching to executing another pipeline stage for different packets,
534the number of L2 or L3 cache memory misses is greatly reduced, hence one of the main reasons for improved performance.
535This is because the cost of L2/L3 cache memory miss on memory read accesses is high, as usually due to data dependency between instructions,
536the CPU execution units have to stall until the read operation is completed from L3 cache memory or external DRAM memory.
537By using prefetch instructions, the latency of memory read accesses is hidden,
538provided that it is preformed early enough before the respective data structure is actually used.
539
540By splitting the processing into several stages that are executed on different packets (the packets from the input burst are interlaced),
541enough work is created to allow the prefetch instructions to complete successfully (before the prefetched data structures are actually accessed) and
542also the data dependency between instructions is loosened.
543For example, for the 4-stage pipeline, stage 0 is executed on packets 0 and 1 and then,
544before same packets 0 and 1 are used (i.e. before stage 1 is executed on packets 0 and 1),
545different packets are used: packets 2 and 3 (executing stage 1), packets 4 and 5 (executing stage 2) and packets 6 and 7 (executing stage 3).
546By executing useful work while the data structures are brought into the L1 or L2 cache memory, the latency of the read memory accesses is hidden.
547By increasing the gap between two consecutive accesses to the same data structure, the data dependency between instructions is loosened;
548this allows making the best use of the super-scalar and out-of-order execution CPU architecture,
549as the number of CPU core execution units that are active (rather than idle or stalled due to data dependency constraints between instructions) is maximized.
550
551The bucket search logic is also implemented without using any branch instructions.
552This avoids the important cost associated with flushing the CPU core execution pipeline on every instance of branch misprediction.
553
554Configurable Key Size Hash Table
555""""""""""""""""""""""""""""""""
556
557:numref:`figure_figure34`, :numref:`table_qos_25` and :numref:`table_qos_26` detail the main data structures used to implement configurable key size hash tables (either LRU or extendable bucket,
558either with pre-computed signature or "do-sig").
559
560.. _figure_figure34:
561
562.. figure:: img/figure34.*
563
564 Data Structures for Configurable Key Size Hash Tables
565
566
567.. _table_qos_25:
568
569.. table:: Main Large Data Structures (Arrays) used for Configurable Key Size Hash Tables
570
571 +---+-------------------------+------------------------------+---------------------------+-------------------------------+
572 | # | Array name | Number of entries | Entry size (bytes) | Description |
573 | | | | | |
574 +===+=========================+==============================+===========================+===============================+
575 | 1 | Bucket array | n_buckets (configurable) | 32 | Buckets of the hash table. |
576 | | | | | |
577 +---+-------------------------+------------------------------+---------------------------+-------------------------------+
578 | 2 | Bucket extensions array | n_buckets_ext (configurable) | 32 | This array is only created |
579 | | | | | for extendable bucket tables. |
580 | | | | | |
581 +---+-------------------------+------------------------------+---------------------------+-------------------------------+
582 | 3 | Key array | n_keys | key_size (configurable) | Keys added to the hash table. |
583 | | | | | |
584 +---+-------------------------+------------------------------+---------------------------+-------------------------------+
585 | 4 | Data array | n_keys | entry_size (configurable) | Key values (key data) |
586 | | | | | associated with the hash |
587 | | | | | table keys. |
588 | | | | | |
589 +---+-------------------------+------------------------------+---------------------------+-------------------------------+
590
591.. _table_qos_26:
592
593.. table:: Field Description for Bucket Array Entry (Configurable Key Size Hash Tables)
594
595 +---+------------------+--------------------+------------------------------------------------------------------+
596 | # | Field name | Field size (bytes) | Description |
597 | | | | |
598 +===+==================+====================+==================================================================+
599 | 1 | Next Ptr/LRU | 8 | For LRU tables, this fields represents the LRU list for the |
600 | | | | current bucket stored as array of 4 entries of 2 bytes each. |
601 | | | | Entry 0 stores the index (0 .. 3) of the MRU key, while entry 3 |
602 | | | | stores the index of the LRU key. |
603 | | | | |
604 | | | | For extendable bucket tables, this field represents the next |
605 | | | | pointer (i.e. the pointer to the next group of 4 keys linked to |
606 | | | | the current bucket). The next pointer is not NULL if the bucket |
607 | | | | is currently extended or NULL otherwise. |
608 | | | | To help the branchless implementation, bit 0 (least significant |
609 | | | | bit) of this field is set to 1 if the next pointer is not NULL |
610 | | | | and to 0 otherwise. |
611 | | | | |
612 +---+------------------+--------------------+------------------------------------------------------------------+
613 | 2 | Sig[0 .. 3] | 4 x 2 | If key X (X = 0 .. 3) is valid, then sig X bits 15 .. 1 store |
614 | | | | the most significant 15 bits of key X signature and sig X bit 0 |
615 | | | | is set to 1. |
616 | | | | |
617 | | | | If key X is not valid, then sig X is set to zero. |
618 | | | | |
619 +---+------------------+--------------------+------------------------------------------------------------------+
620 | 3 | Key Pos [0 .. 3] | 4 x 4 | If key X is valid (X = 0 .. 3), then Key Pos X represents the |
621 | | | | index into the key array where key X is stored, as well as the |
622 | | | | index into the data array where the value associated with key X |
623 | | | | is stored. |
624 | | | | |
625 | | | | If key X is not valid, then the value of Key Pos X is undefined. |
626 | | | | |
627 +---+------------------+--------------------+------------------------------------------------------------------+
628
629
630:numref:`figure_figure35` and :numref:`table_qos_27` detail the bucket search pipeline stages (either LRU or extendable bucket,
631either with pre-computed signature or "do-sig").
632For each pipeline stage, the described operations are applied to each of the two packets handled by that stage.
633
634.. _figure_figure35:
635
636.. figure:: img/figure35.*
637
638 Bucket Search Pipeline for Key Lookup Operation (Configurable Key Size Hash
639 Tables)
640
641
642.. _table_qos_27:
643
644.. table:: Description of the Bucket Search Pipeline Stages (Configurable Key Size Hash Tables)
645
646 +---+---------------------------+------------------------------------------------------------------------------+
647 | # | Stage name | Description |
648 | | | |
649 +===+===========================+==============================================================================+
650 | 0 | Prefetch packet meta-data | Select next two packets from the burst of input packets. |
651 | | | |
652 | | | Prefetch packet meta-data containing the key and key signature. |
653 | | | |
654 +---+---------------------------+------------------------------------------------------------------------------+
655 | 1 | Prefetch table bucket | Read the key signature from the packet meta-data (for extendable bucket hash |
656 | | | tables) or read the key from the packet meta-data and compute key signature |
657 | | | (for LRU tables). |
658 | | | |
659 | | | Identify the bucket ID using the key signature. |
660 | | | |
661 | | | Set bit 0 of the signature to 1 (to match only signatures of valid keys from |
662 | | | the table). |
663 | | | |
664 | | | Prefetch the bucket. |
665 | | | |
666 +---+---------------------------+------------------------------------------------------------------------------+
667 | 2 | Prefetch table key | Read the key signatures from the bucket. |
668 | | | |
669 | | | Compare the signature of the input key against the 4 key signatures from the |
670 | | | packet. As result, the following is obtained: |
671 | | | |
672 | | | *match* |
673 | | | = equal to TRUE if there was at least one signature match and to FALSE in |
674 | | | the case of no signature match; |
675 | | | |
676 | | | *match_many* |
677 | | | = equal to TRUE is there were more than one signature matches (can be up to |
678 | | | 4 signature matches in the worst case scenario) and to FALSE otherwise; |
679 | | | |
680 | | | *match_pos* |
681 | | | = the index of the first key that produced signature match (only valid if |
682 | | | match is true). |
683 | | | |
684 | | | For extendable bucket hash tables only, set |
685 | | | *match_many* |
686 | | | to TRUE if next pointer is valid. |
687 | | | |
688 | | | Prefetch the bucket key indicated by |
689 | | | *match_pos* |
690 | | | (even if |
691 | | | *match_pos* |
692 | | | does not point to valid key valid). |
693 | | | |
694 +---+---------------------------+------------------------------------------------------------------------------+
695 | 3 | Prefetch table data | Read the bucket key indicated by |
696 | | | *match_pos*. |
697 | | | |
698 | | | Compare the bucket key against the input key. As result, the following is |
699 | | | obtained: |
700 | | | *match_key* |
701 | | | = equal to TRUE if the two keys match and to FALSE otherwise. |
702 | | | |
703 | | | Report input key as lookup hit only when both |
704 | | | *match* |
705 | | | and |
706 | | | *match_key* |
707 | | | are equal to TRUE and as lookup miss otherwise. |
708 | | | |
709 | | | For LRU tables only, use branchless logic to update the bucket LRU list |
710 | | | (the current key becomes the new MRU) only on lookup hit. |
711 | | | |
712 | | | Prefetch the key value (key data) associated with the current key (to avoid |
713 | | | branches, this is done on both lookup hit and miss). |
714 | | | |
715 +---+---------------------------+------------------------------------------------------------------------------+
716
717
718Additional notes:
719
720#. The pipelined version of the bucket search algorithm is executed only if there are at least 7 packets in the burst of input packets.
721 If there are less than 7 packets in the burst of input packets,
722 a non-optimized implementation of the bucket search algorithm is executed.
723
724#. Once the pipelined version of the bucket search algorithm has been executed for all the packets in the burst of input packets,
725 the non-optimized implementation of the bucket search algorithm is also executed for any packets that did not produce a lookup hit,
726 but have the *match_many* flag set.
727 As result of executing the non-optimized version, some of these packets may produce a lookup hit or lookup miss.
728 This does not impact the performance of the key lookup operation,
729 as the probability of matching more than one signature in the same group of 4 keys or of having the bucket in extended state
730 (for extendable bucket hash tables only) is relatively small.
731
732**Key Signature Comparison Logic**
733
734The key signature comparison logic is described in :numref:`table_qos_28`.
735
736.. _table_qos_28:
737
738.. table:: Lookup Tables for Match, Match_Many and Match_Pos
739
740 +----+------+---------------+--------------------+--------------------+
741 | # | mask | match (1 bit) | match_many (1 bit) | match_pos (2 bits) |
742 | | | | | |
743 +----+------+---------------+--------------------+--------------------+
744 | 0 | 0000 | 0 | 0 | 00 |
745 | | | | | |
746 +----+------+---------------+--------------------+--------------------+
747 | 1 | 0001 | 1 | 0 | 00 |
748 | | | | | |
749 +----+------+---------------+--------------------+--------------------+
750 | 2 | 0010 | 1 | 0 | 01 |
751 | | | | | |
752 +----+------+---------------+--------------------+--------------------+
753 | 3 | 0011 | 1 | 1 | 00 |
754 | | | | | |
755 +----+------+---------------+--------------------+--------------------+
756 | 4 | 0100 | 1 | 0 | 10 |
757 | | | | | |
758 +----+------+---------------+--------------------+--------------------+
759 | 5 | 0101 | 1 | 1 | 00 |
760 | | | | | |
761 +----+------+---------------+--------------------+--------------------+
762 | 6 | 0110 | 1 | 1 | 01 |
763 | | | | | |
764 +----+------+---------------+--------------------+--------------------+
765 | 7 | 0111 | 1 | 1 | 00 |
766 | | | | | |
767 +----+------+---------------+--------------------+--------------------+
768 | 8 | 1000 | 1 | 0 | 11 |
769 | | | | | |
770 +----+------+---------------+--------------------+--------------------+
771 | 9 | 1001 | 1 | 1 | 00 |
772 | | | | | |
773 +----+------+---------------+--------------------+--------------------+
774 | 10 | 1010 | 1 | 1 | 01 |
775 | | | | | |
776 +----+------+---------------+--------------------+--------------------+
777 | 11 | 1011 | 1 | 1 | 00 |
778 | | | | | |
779 +----+------+---------------+--------------------+--------------------+
780 | 12 | 1100 | 1 | 1 | 10 |
781 | | | | | |
782 +----+------+---------------+--------------------+--------------------+
783 | 13 | 1101 | 1 | 1 | 00 |
784 | | | | | |
785 +----+------+---------------+--------------------+--------------------+
786 | 14 | 1110 | 1 | 1 | 01 |
787 | | | | | |
788 +----+------+---------------+--------------------+--------------------+
789 | 15 | 1111 | 1 | 1 | 00 |
790 | | | | | |
791 +----+------+---------------+--------------------+--------------------+
792
793The input *mask* hash bit X (X = 0 .. 3) set to 1 if input signature is equal to bucket signature X and set to 0 otherwise.
794The outputs *match*, *match_many* and *match_pos* are 1 bit, 1 bit and 2 bits in size respectively and their meaning has been explained above.
795
796As displayed in :numref:`table_qos_29`, the lookup tables for *match* and *match_many* can be collapsed into a single 32-bit value and the lookup table for
797*match_pos* can be collapsed into a 64-bit value.
798Given the input *mask*, the values for *match*, *match_many* and *match_pos* can be obtained by indexing their respective bit array to extract 1 bit,
7991 bit and 2 bits respectively with branchless logic.
800
801.. _table_qos_29:
802
803.. table:: Collapsed Lookup Tables for Match, Match_Many and Match_Pos
804
805 +------------+------------------------------------------+-------------------+
806 | | Bit array | Hexadecimal value |
807 | | | |
808 +------------+------------------------------------------+-------------------+
809 | match | 1111_1111_1111_1110 | 0xFFFELLU |
810 | | | |
811 +------------+------------------------------------------+-------------------+
812 | match_many | 1111_1110_1110_1000 | 0xFEE8LLU |
813 | | | |
814 +------------+------------------------------------------+-------------------+
815 | match_pos | 0001_0010_0001_0011__0001_0010_0001_0000 | 0x12131210LLU |
816 | | | |
817 +------------+------------------------------------------+-------------------+
818
819
820The pseudo-code for match, match_many and match_pos is::
821
822 match = (0xFFFELLU >> mask) & 1;
823
824 match_many = (0xFEE8LLU >> mask) & 1;
825
826 match_pos = (0x12131210LLU >> (mask << 1)) & 3;
827
828Single Key Size Hash Tables
829"""""""""""""""""""""""""""
830
831:numref:`figure_figure37`, :numref:`figure_figure38`, :numref:`table_qos_30` and :numref:`table_qos_31` detail the main data structures used to implement 8-byte and 16-byte key hash tables
832(either LRU or extendable bucket, either with pre-computed signature or "do-sig").
833
834.. _figure_figure37:
835
836.. figure:: img/figure37.*
837
838 Data Structures for 8-byte Key Hash Tables
839
840
841.. _figure_figure38:
842
843.. figure:: img/figure38.*
844
845 Data Structures for 16-byte Key Hash Tables
846
847
848.. _table_qos_30:
849
850.. table:: Main Large Data Structures (Arrays) used for 8-byte and 16-byte Key Size Hash Tables
851
852 +---+-------------------------+------------------------------+----------------------+------------------------------------+
853 | # | Array name | Number of entries | Entry size (bytes) | Description |
854 | | | | | |
855 +===+=========================+==============================+======================+====================================+
856 | 1 | Bucket array | n_buckets (configurable) | *8-byte key size:* | Buckets of the hash table. |
857 | | | | | |
858 | | | | 64 + 4 x entry_size | |
859 | | | | | |
860 | | | | | |
861 | | | | *16-byte key size:* | |
862 | | | | | |
863 | | | | 128 + 4 x entry_size | |
864 | | | | | |
865 +---+-------------------------+------------------------------+----------------------+------------------------------------+
866 | 2 | Bucket extensions array | n_buckets_ext (configurable) | *8-byte key size:* | This array is only created for |
867 | | | | | extendable bucket tables. |
868 | | | | | |
869 | | | | 64 + 4 x entry_size | |
870 | | | | | |
871 | | | | | |
872 | | | | *16-byte key size:* | |
873 | | | | | |
874 | | | | 128 + 4 x entry_size | |
875 | | | | | |
876 +---+-------------------------+------------------------------+----------------------+------------------------------------+
877
878.. _table_qos_31:
879
880.. table:: Field Description for Bucket Array Entry (8-byte and 16-byte Key Hash Tables)
881
882 +---+---------------+--------------------+-------------------------------------------------------------------------------+
883 | # | Field name | Field size (bytes) | Description |
884 | | | | |
885 +===+===============+====================+===============================================================================+
886 | 1 | Valid | 8 | Bit X (X = 0 .. 3) is set to 1 if key X is valid or to 0 otherwise. |
887 | | | | |
888 | | | | Bit 4 is only used for extendable bucket tables to help with the |
889 | | | | implementation of the branchless logic. In this case, bit 4 is set to 1 if |
890 | | | | next pointer is valid (not NULL) or to 0 otherwise. |
891 | | | | |
892 +---+---------------+--------------------+-------------------------------------------------------------------------------+
893 | 2 | Next Ptr/LRU | 8 | For LRU tables, this fields represents the LRU list for the current bucket |
894 | | | | stored as array of 4 entries of 2 bytes each. Entry 0 stores the index |
895 | | | | (0 .. 3) of the MRU key, while entry 3 stores the index of the LRU key. |
896 | | | | |
897 | | | | For extendable bucket tables, this field represents the next pointer (i.e. |
898 | | | | the pointer to the next group of 4 keys linked to the current bucket). The |
899 | | | | next pointer is not NULL if the bucket is currently extended or NULL |
900 | | | | otherwise. |
901 | | | | |
902 +---+---------------+--------------------+-------------------------------------------------------------------------------+
903 | 3 | Key [0 .. 3] | 4 x key_size | Full keys. |
904 | | | | |
905 +---+---------------+--------------------+-------------------------------------------------------------------------------+
906 | 4 | Data [0 .. 3] | 4 x entry_size | Full key values (key data) associated with keys 0 .. 3. |
907 | | | | |
908 +---+---------------+--------------------+-------------------------------------------------------------------------------+
909
910and detail the bucket search pipeline used to implement 8-byte and 16-byte key hash tables (either LRU or extendable bucket,
911either with pre-computed signature or "do-sig").
912For each pipeline stage, the described operations are applied to each of the two packets handled by that stage.
913
914.. _figure_figure39:
915
916.. figure:: img/figure39.*
917
918 Bucket Search Pipeline for Key Lookup Operation (Single Key Size Hash
919 Tables)
920
921
922.. _table_qos_32:
923
924.. table:: Description of the Bucket Search Pipeline Stages (8-byte and 16-byte Key Hash Tables)
925
926 +---+---------------------------+-----------------------------------------------------------------------------+
927 | # | Stage name | Description |
928 | | | |
929 +===+===========================+=============================================================================+
930 | 0 | Prefetch packet meta-data | #. Select next two packets from the burst of input packets. |
931 | | | |
932 | | | #. Prefetch packet meta-data containing the key and key signature. |
933 | | | |
934 +---+---------------------------+-----------------------------------------------------------------------------+
935 | 1 | Prefetch table bucket | #. Read the key signature from the packet meta-data (for extendable bucket |
936 | | | hash tables) or read the key from the packet meta-data and compute key |
937 | | | signature (for LRU tables). |
938 | | | |
939 | | | #. Identify the bucket ID using the key signature. |
940 | | | |
941 | | | #. Prefetch the bucket. |
942 | | | |
943 +---+---------------------------+-----------------------------------------------------------------------------+
944 | 2 | Prefetch table data | #. Read the bucket. |
945 | | | |
946 | | | #. Compare all 4 bucket keys against the input key. |
947 | | | |
948 | | | #. Report input key as lookup hit only when a match is identified (more |
949 | | | than one key match is not possible) |
950 | | | |
951 | | | #. For LRU tables only, use branchless logic to update the bucket LRU list |
952 | | | (the current key becomes the new MRU) only on lookup hit. |
953 | | | |
954 | | | #. Prefetch the key value (key data) associated with the matched key (to |
955 | | | avoid branches, this is done on both lookup hit and miss). |
956 | | | |
957 +---+---------------------------+-----------------------------------------------------------------------------+
958
959Additional notes:
960
961#. The pipelined version of the bucket search algorithm is executed only if there are at least 5 packets in the burst of input packets.
962 If there are less than 5 packets in the burst of input packets, a non-optimized implementation of the bucket search algorithm is executed.
963
964#. For extendable bucket hash tables only,
965 once the pipelined version of the bucket search algorithm has been executed for all the packets in the burst of input packets,
966 the non-optimized implementation of the bucket search algorithm is also executed for any packets that did not produce a lookup hit,
967 but have the bucket in extended state.
968 As result of executing the non-optimized version, some of these packets may produce a lookup hit or lookup miss.
969 This does not impact the performance of the key lookup operation,
970 as the probability of having the bucket in extended state is relatively small.
971
972Pipeline Library Design
973-----------------------
974
975A pipeline is defined by:
976
977#. The set of input ports;
978
979#. The set of output ports;
980
981#. The set of tables;
982
983#. The set of actions.
984
985The input ports are connected with the output ports through tree-like topologies of interconnected tables.
986The table entries contain the actions defining the operations to be executed on the input packets and the packet flow within the pipeline.
987
988Connectivity of Ports and Tables
989~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
990
991To avoid any dependencies on the order in which pipeline elements are created,
992the connectivity of pipeline elements is defined after all the pipeline input ports,
993output ports and tables have been created.
994
995General connectivity rules:
996
997#. Each input port is connected to a single table. No input port should be left unconnected;
998
999#. The table connectivity to other tables or to output ports is regulated by the next hop actions of each table entry and the default table entry.
1000 The table connectivity is fluid, as the table entries and the default table entry can be updated during run-time.
1001
1002 * A table can have multiple entries (including the default entry) connected to the same output port.
1003 A table can have different entries connected to different output ports.
1004 Different tables can have entries (including default table entry) connected to the same output port.
1005
1006 * A table can have multiple entries (including the default entry) connected to another table,
1007 in which case all these entries have to point to the same table.
1008 This constraint is enforced by the API and prevents tree-like topologies from being created (allowing table chaining only),
1009 with the purpose of simplifying the implementation of the pipeline run-time execution engine.
1010
1011Port Actions
1012~~~~~~~~~~~~
1013
1014Port Action Handler
1015^^^^^^^^^^^^^^^^^^^
1016
1017An action handler can be assigned to each input/output port to define actions to be executed on each input packet that is received by the port.
1018Defining the action handler for a specific input/output port is optional (i.e. the action handler can be disabled).
1019
1020For input ports, the action handler is executed after RX function. For output ports, the action handler is executed before the TX function.
1021
1022The action handler can decide to drop packets.
1023
1024Table Actions
1025~~~~~~~~~~~~~
1026
1027Table Action Handler
1028^^^^^^^^^^^^^^^^^^^^
1029
1030An action handler to be executed on each input packet can be assigned to each table.
1031Defining the action handler for a specific table is optional (i.e. the action handler can be disabled).
1032
1033The action handler is executed after the table lookup operation is performed and the table entry associated with each input packet is identified.
1034The action handler can only handle the user-defined actions, while the reserved actions (e.g. the next hop actions) are handled by the Packet Framework.
1035The action handler can decide to drop the input packet.
1036
1037Reserved Actions
1038^^^^^^^^^^^^^^^^
1039
1040The reserved actions are handled directly by the Packet Framework without the user being able to change their meaning
1041through the table action handler configuration.
1042A special category of the reserved actions is represented by the next hop actions, which regulate the packet flow between input ports,
1043tables and output ports through the pipeline.
1044:numref:`table_qos_33` lists the next hop actions.
1045
1046.. _table_qos_33:
1047
1048.. table:: Next Hop Actions (Reserved)
1049
1050 +---+---------------------+-----------------------------------------------------------------------------------+
1051 | # | Next hop action | Description |
1052 | | | |
1053 +===+=====================+===================================================================================+
1054 | 1 | Drop | Drop the current packet. |
1055 | | | |
1056 +---+---------------------+-----------------------------------------------------------------------------------+
1057 | 2 | Send to output port | Send the current packet to specified output port. The output port ID is metadata |
1058 | | | stored in the same table entry. |
1059 | | | |
1060 +---+---------------------+-----------------------------------------------------------------------------------+
1061 | 3 | Send to table | Send the current packet to specified table. The table ID is metadata stored in |
1062 | | | the same table entry. |
1063 | | | |
1064 +---+---------------------+-----------------------------------------------------------------------------------+
1065
1066User Actions
1067^^^^^^^^^^^^
1068
1069For each table, the meaning of user actions is defined through the configuration of the table action handler.
1070Different tables can be configured with different action handlers, therefore the meaning of the user actions
1071and their associated meta-data is private to each table.
1072Within the same table, all the table entries (including the table default entry) share the same definition
1073for the user actions and their associated meta-data,
1074with each table entry having its own set of enabled user actions and its own copy of the action meta-data.
1075:numref:`table_qos_34` contains a non-exhaustive list of user action examples.
1076
1077.. _table_qos_34:
1078
1079.. table:: User Action Examples
1080
1081 +---+-----------------------------------+---------------------------------------------------------------------+
1082 | # | User action | Description |
1083 | | | |
1084 +===+===================================+=====================================================================+
1085 | 1 | Metering | Per flow traffic metering using the srTCM and trTCM algorithms. |
1086 | | | |
1087 +---+-----------------------------------+---------------------------------------------------------------------+
1088 | 2 | Statistics | Update the statistics counters maintained per flow. |
1089 | | | |
1090 +---+-----------------------------------+---------------------------------------------------------------------+
1091 | 3 | App ID | Per flow state machine fed by variable length sequence of packets |
1092 | | | at the flow initialization with the purpose of identifying the |
1093 | | | traffic type and application. |
1094 | | | |
1095 +---+-----------------------------------+---------------------------------------------------------------------+
1096 | 4 | Push/pop labels | Push/pop VLAN/MPLS labels to/from the current packet. |
1097 | | | |
1098 +---+-----------------------------------+---------------------------------------------------------------------+
1099 | 5 | Network Address Translation (NAT) | Translate between the internal (LAN) and external (WAN) IP |
1100 | | | destination/source address and/or L4 protocol destination/source |
1101 | | | port. |
1102 | | | |
1103 +---+-----------------------------------+---------------------------------------------------------------------+
1104 | 6 | TTL update | Decrement IP TTL and, in case of IPv4 packets, update the IP |
1105 | | | checksum. |
1106 | | | |
1107 +---+-----------------------------------+---------------------------------------------------------------------+
1108
1109Multicore Scaling
1110-----------------
1111
1112A complex application is typically split across multiple cores, with cores communicating through SW queues.
1113There is usually a performance limit on the number of table lookups
1114and actions that can be fitted on the same CPU core due to HW constraints like:
1115available CPU cycles, cache memory size, cache transfer BW, memory transfer BW, etc.
1116
1117As the application is split across multiple CPU cores, the Packet Framework facilitates the creation of several pipelines,
1118the assignment of each such pipeline to a different CPU core
1119and the interconnection of all CPU core-level pipelines into a single application-level complex pipeline.
1120For example, if CPU core A is assigned to run pipeline P1 and CPU core B pipeline P2,
1121then the interconnection of P1 with P2 could be achieved by having the same set of SW queues act like output ports
1122for P1 and input ports for P2.
1123
1124This approach enables the application development using the pipeline, run-to-completion (clustered) or hybrid (mixed) models.
1125
1126It is allowed for the same core to run several pipelines, but it is not allowed for several cores to run the same pipeline.
1127
1128Shared Data Structures
1129~~~~~~~~~~~~~~~~~~~~~~
1130
1131The threads performing table lookup are actually table writers rather than just readers.
1132Even if the specific table lookup algorithm is thread-safe for multiple readers
1133(e. g. read-only access of the search algorithm data structures is enough to conduct the lookup operation),
1134once the table entry for the current packet is identified, the thread is typically expected to update the action meta-data stored in the table entry
1135(e.g. increment the counter tracking the number of packets that hit this table entry), and thus modify the table entry.
1136During the time this thread is accessing this table entry (either writing or reading; duration is application specific),
1137for data consistency reasons, no other threads (threads performing table lookup or entry add/delete operations) are allowed to modify this table entry.
1138
1139Mechanisms to share the same table between multiple threads:
1140
1141#. **Multiple writer threads.**
1142 Threads need to use synchronization primitives like semaphores (distinct semaphore per table entry) or atomic instructions.
1143 The cost of semaphores is usually high, even when the semaphore is free.
1144 The cost of atomic instructions is normally higher than the cost of regular instructions.
1145
1146#. **Multiple writer threads, with single thread performing table lookup operations and multiple threads performing table entry add/delete operations.**
1147 The threads performing table entry add/delete operations send table update requests to the reader (typically through message passing queues),
1148 which does the actual table updates and then sends the response back to the request initiator.
1149
1150#. **Single writer thread performing table entry add/delete operations and multiple reader threads that perform table lookup operations with read-only access to the table entries.**
1151 The reader threads use the main table copy while the writer is updating the mirror copy.
1152 Once the writer update is done, the writer can signal to the readers and busy wait until all readers swaps between the mirror copy (which now becomes the main copy) and
1153 the mirror copy (which now becomes the main copy).
1154
1155Interfacing with Accelerators
1156-----------------------------
1157
1158The presence of accelerators is usually detected during the initialization phase by inspecting the HW devices that are part of the system (e.g. by PCI bus enumeration).
1159Typical devices with acceleration capabilities are:
1160
1161* Inline accelerators: NICs, switches, FPGAs, etc;
1162
1163* Look-aside accelerators: chipsets, FPGAs, etc.
1164
1165Usually, to support a specific functional block, specific implementation of Packet Framework tables and/or ports and/or actions has to be provided for each accelerator,
1166with all the implementations sharing the same API: pure SW implementation (no acceleration), implementation using accelerator A, implementation using accelerator B, etc.
1167The selection between these implementations could be done at build time or at run-time (recommended), based on which accelerators are present in the system,
1168with no application changes required.
Ceph