1 /*! \page platform Describing the virtual platform
5 As @ref starting_components "explained in the introduction," any
6 SimGrid study must entail the description of the platform on which you
7 want to simulate your application. You have to describe **each element
8 of your platform**, such as computing hosts, clusters, each disks,
9 links, etc. You must also define the **routing on your platform**, ie
10 which path is taken between two hosts. Finally, you may also describe
11 an **experimental scenario**, with qualitative changes (e.g.,
12 bandwidth changes representing an external load) and qualitative
13 changes (representing how some elements fail and restart over time).
15 You should really separate your application from the platform
16 description, as it will ease your experimental campain afterward.
17 Mixing them is seen as a really bad experimental practice. The easiest
18 to enforce this split is to put the platform description in a XML
19 file. Many example platforms are provided in the archive, and this
20 page gives all needed details to write such files, as well as some
21 hints and tricks about describing your platform.
23 On the other side, XML is sometimes not expressive enough for some
24 platforms, in particular large platforms exhibiting repetitive
25 patterns that are not simply expressed in XML. In practice, many
26 users end up generating their XML platform files from some sort of
27 scripts. It is probably preferable to rewrite your XML @ref
28 platform_lua "platform using the lua scripting language" instead.
29 In the future, it should be possible to describe the platform directly
30 in C++, but this is not possible yet.
32 As usual, SimGrid is a versatile framework, and you should find the
33 way of describing your platform that best fits your experimental
36 \section pf_overview Describing the platform with XML
38 Your platform description should follow the specification presented in
39 the [simgrid.dtd](http://simgrid.gforge.inria.fr/simgrid/simgrid.dtd)
40 DTD file. The same DTD is used for both the platform and deployment
43 From time to time, this DTD evolves to introduce possibly
44 backward-incompatible changes. That is why each platform desciption is
45 enclosed within a @c platform tag, that have a @c version attribute.
46 The current version is <b>4.1</b>. The @c simgrid_update_xml program can
47 upgrade most of the past platform files to the recent formalism.
49 \section pf_first_example First Platform Example
51 Here is a very simple platform file, containing 3 resources (two hosts
52 and one link), and explicitly giving the route between the hosts.
56 <!DOCTYPE platform SYSTEM "http://simgrid.gforge.inria.fr/simgrid/simgrid.dtd">
57 <platform version="4.1">
58 <zone id="first zone" routing="Full">
59 <!-- the resources -->
60 <host id="host1" speed="1Mf"/>
61 <host id="host2" speed="2Mf"/>
62 <link id="link1" bandwidth="125MBps" latency="100us"/>
63 <!-- the routing: specify how the hosts are interconnected -->
64 <route src="host1" dst="host2">
65 <link_ctn id="link1"/>
71 As we said, the englobing @ref pf_overview "<platform>" tag is
72 used to specify the dtd version used for this file.
74 Then, every resource (specified with @ref pf_tag_host, @ref
75 pf_tag_link or others) must be located within a given **networking
76 zone**. Each zone is in charge of the routing between its
77 resources. It means that when an host wants to communicate with
78 another host of the same zone, it is the zone's duty to find the list
79 of links that are involved in the communication. Here, since the @ref
80 pf_tag_zone tag has **Full** as a **routing attribute**, all routes
81 must be explicitely given using the @ref pf_tag_route and @ref
82 pf_tag_linkctn tags (this @ref pf_rm "routing model" is both simple
83 and inefficient :) It is OK to not specify the route between two
84 hosts, as long as the processes located on them never try to
87 A zone can contain several zones itself, leading to a hierarchical
88 decomposition of the platform. This can be more efficient (as the
89 inter-zone routing gets factorized with @ref pf_tag_zoneroute), and
90 allows to have more than one routing model in your platform. For
91 example, you could have a coordinate-based routing for the WAN parts
92 of your platforms, a full routing within each datacenter, and a highly
93 optimized routing within each cluster of the datacenter. In this
94 case, determining the route between two given hosts gets @ref
95 routing_basics "somewhat more complex" but SimGrid still computes
96 these routes for you in a time- and space-efficient manner.
97 Here is an illustration of these concepts:
99 
101 Circles represent processing units and squares represent network
102 routers. Bold lines represent communication links. The zone "AS2"
103 models the core of a national network interconnecting a small flat
104 cluster (AS4) and a larger hierarchical cluster (AS5), a subset of a
105 LAN (AS6), and a set of peers scattered around the world (AS7).
107 \section pf_res Resource description
109 \subsection pf_res_computing Computing Resources
111 \subsubsection pf_tag_host <host>
113 An host is the computing resource on which an actor can execute.
115 Attribute | Values | Description
116 ----------------- | -------------------------------------- | -----------
117 id | String (mandatory) | The identifier of the host. facilitates referring to this AS.
118 speed | double (mandatory) | Computational power of every core of this host in FLOPS (must be positive)
119 core | int (defaults to 1) | Number of cores (see @ref howto_multicore)
120 state | optionally "OFF" | If set to OFF, the host is initially turned off.
121 availability_file | File name (optional) | (Relative or absolute) filename to use as input; must contain availability traces for this host. The syntax of this file is defined below.
122 state_file | File name (optional) | File to use as a state profile (see @ref howto_churn)
123 coordinates | String (mandatory when using Vivaldi routing) | The coordinates of this host (see @ref pf_P2P_tags).
124 pstate | Double (Defaults to 0) | FIXME: Not yet documented.
126 #### Included tags ####
128 - @ref pf_tag_mount Specifies the storages mounted on that host
129 - @ref pf_tag_prop Specifies a user-defined property of that host, that you can retrieve with MSG_host_get_property_value() or simgrid::s4u::Host::property().
134 <host id="host1" speed="1000000000"/>
135 <host id="host2" speed="1000000000">
136 <prop id="color" value="blue"/>
137 <prop id="rendershape" value="square"/>
141 \anchor pf_host_dynamism
142 ### Expressing dynamism ###
144 SimGrid provides mechanisms to change a hosts' availability over
145 time, using the ``availability_file`` attribute to the ``\<host\>`` tag
146 and a separate text file whose syntax is exemplified below.
148 #### Adding a trace file ####
151 <platform version="4">
152 <host id="bob" speed="500Gf" availability_file="bob.trace" />
156 #### Example of "bob.trace" file ####
165 Let us begin to explain this example by looking at line 2. (Line 1 will become clear soon).
166 The first column describes points in time, in this case, time 0. The second column
167 describes the relative amount of power this host is able to deliver (relative
168 to the maximum performance specified in the ``\<host\>`` tag). (Clearly, the
169 second column needs to contain values that are not smaller than 0 and not larger than 1).
170 In this example, our host will deliver 500 Mflop/s at time 0, as 500 Mflop/s is the
171 maximum performance of this host. At time 11.0, it will
172 deliver half of its maximum performance, i.e., 250 Mflop/s until time 20.0 when it will
173 will start delivering 80\% of its power. In this example, this amounts to 400 Mflop/s.
175 Since the periodicity in line 1 was set to be 1.0, i.e., 1 timestep, this host will
176 continue to provide 500 Mflop/s from time 21. From time 32 it will provide 250 MFlop/s and so on.
178 \subsubsection pf_tag_cluster <cluster>
180 ``<cluster />`` represents a machine-cluster. It is most commonly used
181 when one wants to define many hosts and a network quickly. Technically,
182 ``cluster`` is a meta-tag: <b>from the inner SimGrid point of
183 view, a cluster is an AS where some optimized routing is defined</b>.
184 The default inner organization of the cluster is as follow:
190 ____________|__________|_____________ backbone
192 l0| l1| l2| l97| l96 | | l99
198 Here, a set of <b>host</b>s is defined. Each of them has a <b>link</b>
199 to a central backbone (backbone is a link itself, as a link can
200 be used to represent a switch, see the switch / link section
201 below for more details about it). A <b>router</b> allows to connect a
202 <b>cluster</b> to the outside world. Internally,
203 SimGrid treats a cluster as an AS containing all hosts: the router is the default
204 gateway for the cluster.
206 There is an alternative organization, which is as follows:
220 The principle is the same, except that there is no backbone. This representation
221 can be obtained easily: just do not set the bb_* attributes.
224 Attribute name | Mandatory | Values | Description
225 --------------- | --------- | ------ | -----------
226 id | yes | string | The identifier of the cluster. Facilitates referring to this cluster.
227 prefix | yes | string | Each node of the cluster has to have a name. This name will be prefixed with this prefix.
228 suffix | yes | string | Each node of the cluster will be suffixed with this suffix
229 radical | yes | string | Regexp used to generate cluster nodes name. Syntax: "10-20" will give you 11 machines numbered from 10 to 20, "10-20;2" will give you 12 machines, one with the number 2, others numbered as before. The produced number is concatenated between prefix and suffix to form machine names.
230 speed | yes | int | Same as the ``speed`` attribute of the ``\<host\>`` tag.
231 core | no | int (default: 1) | Same as the ``core`` attribute of the ``\<host\>`` tag.
232 bw | yes | int | Bandwidth for the links between nodes and backbone (if any). See the \ref pf_tag_link "link section" for syntax/details.
233 lat | yes | int | Latency for the links between nodes and backbone (if any). See <b>link</b> section for syntax/details.
234 sharing_policy | no | string | Sharing policy for the links between nodes and backbone (if any). See <b>link</b> section for syntax/details.
235 bb_bw | no | int | Bandwidth for backbone (if any). See <b>link</b> section for syntax/details. If bb_bw and bb_lat (see below) attributes are omitted, no backbone is created (alternative cluster architecture <b>described before</b>).
236 bb_lat | no | int | Latency for backbone (if any). See <b>link</b> section for syntax/details. If bb_lat and bb_bw (see above) attributes are omitted, no backbone is created (alternative cluster architecture <b>described before</b>).
237 bb_sharing_policy | no | string | Sharing policy for the backbone (if any). See <b>link</b> section for syntax/details.
238 limiter_link | no | int | Bandwidth for limiter link (if any). This adds a specific link for each node, to set the maximum bandwidth reached when communicating in both directions at the same time. In theory this value should be 2*bw for fullduplex links, but in reality this might be less. This value will depend heavily on the communication model, and on the cluster's hardware, so no default value can be set, this has to be measured. More details can be obtained in <a href="https://hal.inria.fr/hal-00919507/"> "Toward Better Simulation of MPI Applications on Ethernet/TCP Networks"</a>
239 loopback_bw | no | int | Bandwidth for loopback (if any). See <b>link</b> section for syntax/details. If loopback_bw and loopback_lat (see below) attributes are omitted, no loopback link is created and all intra-node communication will use the main network link of the node. Loopback link is a \ref pf_sharing_policy_fatpipe "\b FATPIPE".
240 loopback_lat | no | int | Latency for loopback (if any). See <b>link</b> section for syntax/details. See loopback_bw for more info.
241 topology | no | FLAT\|TORUS\|FAT_TREE\|DRAGONFLY (default: FLAT) | Network topology to use. SimGrid currently supports FLAT (with or without backbone, as described before), <a href="http://en.wikipedia.org/wiki/Torus_interconnect">TORUS </a>, FAT_TREE, and DRAGONFLY attributes for this tag.
242 topo_parameters | no | string | Specific parameters to pass for the topology defined in the topology tag. For torus networks, comma-separated list of the number of nodes in each dimension of the torus. Please refer to the specific documentation for \ref simgrid::kernel::routing::FatTreeZone "FatTree NetZone", \ref simgrid::kernel::routing::DragonflyZone "Dragonfly NetZone".
245 the router name is defined as the resulting String in the following
249 router_name = prefix + clusterId + "_router" + suffix;
253 #### Cluster example ####
255 Consider the following two (and independent) uses of the ``cluster`` tag:
258 <cluster id="my_cluster_1" prefix="" suffix="" radical="0-262144"
259 speed="1e9" bw="125e6" lat="5E-5"/>
261 <cluster id="my_cluster_2" prefix="c-" suffix=".me" radical="0-99"
262 speed="1e9" bw="125e6" lat="5E-5"
263 bb_bw="2.25e9" bb_lat="5E-4"/>
266 The second example creates one router and 100 machines with the following names:
268 c-my_cluster_2_router.me
276 \subsubsection pf_cabinet <cabinet>
279 This tag is only available when the routing mode of the AS
280 is set to ``Cluster``.
282 The ``<cabinet />`` tag is, like the \ref pf_tag_cluster "<cluster>" tag,
283 a meta-tag. This means that it is simply a shortcut for creating a set of (homogenous) hosts and links quickly;
284 unsurprisingly, this tag was introduced to setup cabinets in data centers quickly. Unlike
285 <cluster>, however, the <cabinet> assumes that you create the backbone
286 and routers yourself; see our examples below.
290 Attribute name | Mandatory | Values | Description
291 --------------- | --------- | ------ | -----------
292 id | yes | string | The identifier of the cabinet. Facilitates referring to this cluster.
293 prefix | yes | string | Each node of the cabinet has to have a name. This name will be prefixed with this prefix.
294 suffix | yes | string | Each node of the cabinet will be suffixed with this suffix
295 radical | yes | string | Regexp used to generate cabinet nodes name. Syntax: "10-20" will give you 11 machines numbered from 10 to 20, "10-20;2" will give you 12 machines, one with the number 2, others numbered as before. The produced number is concatenated between prefix and suffix to form machine names.
296 speed | yes | int | Same as the ``speed`` attribute of the \ref pf_tag_host "<host>" tag.
297 bw | yes | int | Bandwidth for the links between nodes and backbone (if any). See the \ref pf_tag_link "link section" for syntax/details.
298 lat | yes | int | Latency for the links between nodes and backbone (if any). See the \ref pf_tag_link "link section" for syntax/details.
301 Please note that as of now, it is impossible to change attributes such as,
302 amount of cores (always set to 1), the initial state of hosts/links
303 (always set to ON), the sharing policy of the links (always set to \ref pf_sharing_policy_fullduplex "FULLDUPLEX").
307 The following example was taken from ``examples/platforms/meta_cluster.xml`` and
308 shows how to use the cabinet tag.
311 <AS id="my_cluster1" routing="Cluster">
312 <cabinet id="cabinet1" prefix="host-" suffix=".cluster1"
313 speed="1Gf" bw="125MBps" lat="100us" radical="1-10"/>
314 <cabinet id="cabinet2" prefix="host-" suffix=".cluster1"
315 speed="1Gf" bw="125MBps" lat="100us" radical="11-20"/>
316 <cabinet id="cabinet3" prefix="host-" suffix=".cluster1"
317 speed="1Gf" bw="125MBps" lat="100us" radical="21-30"/>
319 <backbone id="backbone1" bandwidth="2.25GBps" latency="500us"/>
324 Please note that you must specify the \ref pf_backbone "<backbone>"
325 tag by yourself; this is not done automatically and there are no checks
326 that ensure this backbone was defined.
328 The hosts generated in the above example are named host-1.cluster, host-2.cluster1
332 \subsubsection pf_peer \<peer\> (Vivaldi netzones only)
334 This tag represents a peer, as in Peer-to-Peer (P2P) networks. This
335 can only be used in Vivaldi NetZones. It creates the following
336 resources to the NetZone:
339 \li Two links: One for download and one for upload. This is
340 convenient to use and simulate stuff under the last mile model (e.g., ADSL peers).
341 \li It connects the two links to the host
345 Attribute name | Mandatory | Values | Description
346 --------------- | --------- | ------ | -----------
347 id | yes | string | The identifier of the peer. Facilitates referring to this peer.
348 speed | yes | int | See the description of the ``host`` tag for this attribute
349 bw_in | yes | int | Bandwidth of the private downstream link
350 bw_out | yes | int | Bandwidth of the private upstream link
351 coordinates | no | string | Coordinates of the gateway for this peer. Example value: 12.8 14.4 6.4
352 sharing_policy | no | SHARED\|FULLDUPLEX (default: FULLDUPLEX) | Sharing policy for links. See <b>link</b> description for details.
353 availability_file| no | string | Availability file for the peer. Same as host availability file. See <b>host</b> description for details.
354 state_file | no | string | State file for the peer. Same as host state file. See <b>host</b> description for details.
357 The communication latency between an host A=(xA,yA,zA) and an host
358 B=(xB,yB,zB) is computed as follows:
360 latency = sqrt( (xA-xB)² + (yA-yB)² ) + zA + zB
362 See the documentation of simgrid::kernel::routing::VivaldiZone for
363 details on how the latency is computed from the coordinate, and on the
364 the up and down bandwidth are used.
366 \subsection pf_ne Network equipments
368 There are two tags at all times available to represent network entities and
369 several other tags that are available only in certain contexts.
370 1. ``<link>``: Represents a entity that has a limited bandwidth, a
371 latency, and that can be shared according to TCP way to share this
374 The concept of links in SimGrid may not be intuitive, as links are not
375 limited to connecting (exactly) two entities; in fact, you can have more than
376 two equipments connected to it. (In graph theoretical terms: A link in
377 SimGrid is not an edge, but a hyperedge)
379 2. ``<router/>``: Represents an entity that a message can be routed
380 to, but that is unable to execute any code. In SimGrid, routers have also
381 no impact on the performance: Routers do not limit any bandwidth nor
382 do they increase latency. As a matter of fact, routers are (almost) ignored
383 by the simulator when the simulation has begun.
385 3. ``<backbone/>``: This tag is only available when the containing AS is
386 used as a cluster (i.e., mode="Cluster")
389 If you want to represent an entity like a switch, you must use ``<link>`` (see section). Routers are used
390 to run some routing algorithm and determine routes (see Section \ref pf_routing for details).
392 \subsubsection pf_router <router/>
394 As said before, <b>router</b> is used only to give some information
395 for routing algorithms. So, it does not have any attributes except :
399 Attribute name | Mandatory | Values | Description
400 --------------- | --------- | ------ | -----------
401 id | yes | string | The identifier of the router to be used when referring to it.
402 coordinates | no | string | Must be provided when choosing the Vivaldi, coordinate-based routing model for the AS the router belongs to. More details can be found in the Section \ref pf_P2P_tags.
407 <router id="gw_dc1_horizdist"/>
410 \subsubsection pf_tag_link <link>
412 Network links can represent one-hop network connections. They are
413 characterized by their id and their bandwidth; links can (but may not) be subject
418 Attribute name | Mandatory | Values | Description
419 --------------- | --------- | ------ | -----------
420 id | yes | string | The identifier of the link to be used when referring to it.
421 bandwidth | yes | int | Maximum bandwidth for this link, given in bytes/s
422 latency | no | double (default: 0.0) | Latency for this link.
423 sharing_policy | no | \ref sharing_policy_shared "SHARED"\|\ref pf_sharing_policy_fatpipe "FATPIPE"\|\ref pf_sharing_policy_fullduplex "FULLDUPLEX" (default: SHARED) | Sharing policy for the link.
424 state | no | ON\|OFF (default: ON) | Allows you to to turn this link on or off (working / not working)
425 bandwidth_file | no | string | Allows you to use a file as input for bandwidth.
426 latency_file | no | string | Allows you to use a file as input for latency.
427 state_file | no | string | Allows you to use a file as input for states.
430 #### Possible shortcuts for ``latency`` ####
432 When using the latency attribute, you can specify the latency by using the scientific
433 notation or by using common abbreviations. For instance, the following three tags
437 <link id="LINK1" bandwidth="125000000" latency="5E-6"/>
438 <link id="LINK1" bandwidth="125000000" latency="5us"/>
439 <link id="LINK1" bandwidth="125000000" latency="0.000005"/>
442 Here, the second tag uses "us", meaning "microseconds". Other shortcuts are:
444 Name | Abbreviation | Time (in seconds)
445 ---- | ------------ | -----------------
446 Week | w | 7 * 24 * 60 * 60
447 Day | d | 24 * 60 * 60
451 Millisecond | ms | 0.001 = 10^(-3)
452 Microsecond | us | 0.000001 = 10^(-6)
453 Nanosecond | ns | 0.000000001 = 10^(-9)
454 Picosecond | ps | 0.000000000001 = 10^(-12)
456 #### Sharing policy ####
458 \anchor sharing_policy_shared
459 By default a network link is \b SHARED, i.e., if two or more data flows go
460 through a link, the bandwidth is shared fairly among all data flows. This
461 is similar to the sharing policy TCP uses.
463 \anchor pf_sharing_policy_fatpipe
464 On the other hand, if a link is defined as a \b FATPIPE,
465 each flow going through this link will be provided with the complete bandwidth,
466 i.e., no sharing occurs and the bandwidth is only limiting each flow individually.
467 Please note that this is really on a per-flow basis, not only on a per-host basis!
468 The complete bandwidth provided by this link in this mode
469 is ``number_of_flows*bandwidth``, with at most ``bandwidth`` being available per flow.
471 Using the FATPIPE mode allows to model backbones that won't affect performance
474 \anchor pf_sharing_policy_fullduplex
475 The last mode available is \b FULLDUPLEX. This means that SimGrid will
476 automatically generate two links (one carrying the suffix _UP and the other the
477 suffix _DOWN) for each ``<link>`` tag. This models situations when the direction
478 of traffic is important.
481 Transfers from one side to the other will interact similarly as
482 TCP when ACK returning packets circulate on the other direction. More
483 discussion about it is available in the description of link_ctn description.
485 In other words: The SHARED policy defines a physical limit for the bandwidth.
486 The FATPIPE mode defines a limit for each application,
487 with no upper total limit.
490 Tip: By using the FATPIPE mode, you can model big backbones that
491 won't affect performance (except latency).
496 <link id="SWITCH" bandwidth="125000000" latency="5E-5" sharing_policy="FATPIPE" />
499 #### Expressing dynamism and failures ####
501 Similar to hosts, it is possible to declare links whose state, bandwidth
502 or latency changes over time (see Section \ref pf_host_dynamism for details).
504 In the case of network links, the ``bandwidth`` and ``latency`` attributes are
505 replaced by the ``bandwidth_file`` and ``latency_file`` attributes.
506 The following XML snippet demonstrates how to use this feature in the platform
507 file. The structure of the files "link1.bw" and "link1.lat" is shown below.
510 <link id="LINK1" state_file="link1.fail" bandwidth="80000000" latency=".0001" bandwidth_file="link1.bw" latency_file="link1.lat" />
514 Even if the syntax is the same, the semantic of bandwidth and latency
515 trace files differs from that of host availability files. For bandwidth and
516 latency, the corresponding files do not
517 express availability as a fraction of the available capacity but directly in
518 bytes per seconds for the bandwidth and in seconds for the latency. This is
519 because most tools allowing to capture traces on real platforms (such as NWS)
520 express their results this way.
522 ##### Example of "link1.bw" file #####
530 In this example, the bandwidth changes repeatedly, with all changes
531 being repeated every 12 seconds.
533 At the beginning of the the simulation, the link's bandwidth is 80,000,000
534 B/s (i.e., 80 Mb/s); this value was defined in the XML snippet above.
535 After four seconds, it drops to 40 Mb/s (line 2), and climbs
536 back to 60 Mb/s after another 4 seconds (line 3). The value does not change any
537 more until the end of the period, that is, after 12 seconds have been simulated).
538 At this point, periodicity kicks in and this behavior is repeated: Seconds
539 12-16 will experience 80 Mb/s, 16-20 40 Mb/s etc.).
541 ##### Example of "link1.lat" file #####
550 In this example, the latency varies with a period of 5 seconds.
551 In the xml snippet above, the latency is initialized to be 0.0001s (100µs). This
552 value will be kept during the first second, since the latency_file contains
553 changes to this value at second one, two and three.
554 At second one, the value will be 0.001, i.e., 1ms. One second later it will
555 be adjusted to 0.01 (or 10ms) and one second later it will be set again to 1ms. The
556 value will not change until second 5, when the periodicity defined in line 1
557 kicks in. It then loops back, starting at 100µs (the initial value) for one second.
559 #### The ``<prop/>`` tag ####
561 Similar to the ``<host>`` tag, a link may also contain the ``<prop/>`` tag; see the host
562 documentation (Section \ref pf_host) for an example.
565 \subsubsection pf_backbone <backbone/>
568 This tag is <b>only available</b> when the containing AS uses the "Cluster" routing mode!
570 Using this tag, you can designate an already existing link to be a backbone.
572 Attribute name | Mandatory | Values | Description
573 --------------- | --------- | ------ | -----------
574 id | yes | string | Name of the link that is supposed to act as a backbone.
576 \subsection pf_storage Storage
579 This is a prototype version that should evolve quickly, hence this
580 is just some doc valuable only at the time of writing.
581 This section describes the storage management under SimGrid ; nowadays
582 it's only usable with MSG. It relies basically on linux-like concepts.
583 You also may want to have a look to its corresponding section in
584 @ref msg_file ; access functions are organized as a POSIX-like
587 \subsubsection pf_sto_conc Storage - Main Concepts
589 The storage facilities implemented in SimGrid help to model (and account for)
590 storage devices, such as tapes, hard-drives, CD or DVD devices etc.
591 A typical situation is depicted in the figure below:
593 \image html ./webcruft/storage_sample_scenario.png
594 \image latex ./webcruft/storage_sample_scenario.png "storage_sample_scenario" width=\textwidth
596 In this figure, two hosts called Bob and Alice are interconnected via a network
597 and each host is physically attached to a disk; it is not only possible for each host to
598 mount the disk they are attached to directly, but they can also mount disks
599 that are in a remote location. In this example, Bob mounts Alice's disk remotely
600 and accesses the storage via the network.
602 SimGrid provides 3 different entities that can be used to model setups
603 that include storage facilities:
605 Entity name | Description
606 --------------- | -----------
607 \ref pf_storage_entity_storage_type "storage_type" | Defines a template for a particular kind of storage (such as a hard-drive) and specifies important features of the storage, such as capacity, performance (read/write), contents, ... Different models of hard-drives use different storage_types (because the difference between an SSD and an HDD does matter), as they differ in some specifications (e.g., different sizes or read/write performance).
608 \ref pf_storage_entity_storage "storage" | Defines an actual instance of a storage type (disk, RAM, ...); uses a ``storage_type`` template (see line above) so that you don't need to re-specify the same details over and over again.
609 \ref pf_tag_mount "mount" | Must be wrapped by a \ref pf_tag_host tag; declares which storage(s) this host has mounted and where (i.e., the mountpoint).
612 \anchor pf_storage_content_file
613 ### %Storage Content File ###
615 In order to assess exactly how much time is spent reading from the storage,
616 SimGrid needs to know what is stored on the storage device (identified by distinct (file-)name, like in a file system)
617 and what size this content has.
620 The content file is never changed by the simulation; it is parsed once
621 per simulation and kept in memory afterwards. When the content of the
622 storage changes, only the internal SimGrid data structures change.
624 \anchor pf_storage_content_file_structure
625 #### Structure of a %Storage Content File ####
627 Here is an excerpt from two storage content file; if you want to see the whole file, check
628 the file ``examples/platforms/content/storage_content.txt`` that comes with the
631 SimGrid essentially supports two different formats: UNIX-style filepaths should
632 follow the well known format:
635 /lib/libsimgrid.so.3.6.2 12710497
639 /bin/simgrid_update_xml 5018
640 /bin/graphicator 66986
641 /bin/simgrid-colorizer 2993
646 Windows filepaths, unsurprisingly, use the windows style:
649 \Windows\avastSS.scr 41664
650 \Windows\bfsvc.exe 75264
651 \Windows\bootstat.dat 67584
652 \Windows\CoreSingleLanguage.xml 31497
654 \Windows\dchcfg64.exe 335464
655 \Windows\dcmdev64.exe 93288
659 The different file formats come at a cost; in version 3.12 (and most likely
660 in later versions, too), copying files from windows-style storages to unix-style
661 storages (and vice versa) is not supported.
663 \anchor pf_storage_content_file_create
664 #### Generate a %Storage Content File ####
666 If you want to generate a storage content file based on your own filesystem (or at least a filesystem you have access to),
667 try running this command (works only on unix systems):
670 find . -type f -exec ls -1s --block=1 {} \; 2>/dev/null | awk '{ print $2 " " $1}' > ./content.txt
673 \subsubsection pf_storage_entities The Storage Entities
675 These are the entities that you can use in your platform files to include
676 storage in your model. See also the list of our \ref pf_storage_example_files "example files";
677 these might also help you to get started.
679 \anchor pf_storage_entity_storage_type
680 #### \<storage_type\> ####
682 Attribute name | Mandatory | Values | Description
683 --------------- | --------- | ------ | -----------
684 id | yes | string | Identifier of this storage_type; used when referring to it
685 model | yes | string | For reasons of future backwards compatibility only; specifies the name of the model for the storage that should be used
686 size | yes | string | Specifies the amount of available storage space; you can specify storage like "500GiB" or "500GB" if you want. (TODO add a link to all the available abbreviations)
687 content | yes | string | Path to a \ref pf_storage_content_file "Storage Content File" on your system. This file must exist.
689 This tag must contain some predefined model properties, specified via the <model_prop> tag. Here is a list,
690 see below for an example:
692 Property id | Mandatory | Values | Description
693 --------------- | --------- | ------ | -----------
694 Bwrite | yes | string | Bandwidth for write access; in B/s (but you can also specify e.g. "30MBps")
695 Bread | yes | string | Bandwidth for read access; in B/s (but you can also specify e.g. "30MBps")
696 Bconnexion | yes | string | Throughput (of the storage connector) in B/s.
699 A storage_type can also contain the <b><prop></b> tag. The <prop> tag allows you
700 to associate additional information to this <storage_type> and follows the
701 attribute/value schema; see the example below. You may want to use it to give information to
702 the tool you use for rendering your simulation, for example.
704 Here is a complete example for the ``storage_type`` tag:
706 <storage_type id="single_HDD" model="linear_no_lat" size="4000">
707 <model_prop id="Bwrite" value="30MBps" />
708 <model_prop id="Bread" value="100MBps" />
709 <model_prop id="Bconnection" value="150MBps" />
710 <prop id="Brand" value="Western Digital" />
714 @subsubsection pf_tag_storage <storage>
716 Attributes | Mandatory | Values | Description
717 -------------- | --------- | ------ | -----------
718 id | yes | string | Identifier of this ``storage``; used when referring to it
719 typeId | yes | string | Here you need to refer to an already existing \ref pf_storage_entity_storage_type "\<storage_type\>"; the storage entity defined by this tag will then inherit the properties defined there.
720 attach | yes | string | Name of a host (see Section \ref pf_host) to which this storage is <i>physically</i> attached to (e.g., a hard drive in a computer)
721 content | no | string | When specified, overwrites the content attribute of \ref pf_storage_entity_storage_type "\<storage_type\>"
723 Here are two examples:
726 <storage id="Disk1" typeId="single_HDD" attach="bob" />
728 <storage id="Disk2" typeId="single_SSD"
729 content="content/win_storage_content.txt" />
732 The first example is straightforward: A disk is defined and called "Disk1"; it is
733 of type "single_HDD" (shown as an example of \ref pf_storage_entity_storage_type "\<storage_type\>" above) and attached
734 to a host called "bob" (the definition of this host is omitted here).
736 The second storage is called "Disk2", is still of the same type as Disk1 but
737 now specifies a new content file (so the contents will be different from Disk1)
738 and the filesystem uses the windows style; finally, it is attached to a second host,
739 called alice (which is again not defined here).
741 \subsubsection pf_tag_mount <mount>
743 | Attribute | Mandatory | Values | Description |
744 | ----------- | ----------- | -------- | ------------- |
745 | id | yes | string | Refers to a \ref pf_storage_entity_storage "<storage>" entity that will be mounted on that computer |
746 | name | yes | string | Path/location to/of the logical reference (mount point) of this disk
748 This tag must be enclosed by a \ref pf_tag_host tag. It then specifies where the mountpoint of a given storage device (defined by the ``id`` attribute)
749 is; this location is specified by the ``name`` attribute.
751 Here is a simple example, taken from the file ``examples/platform/storage.xml``:
754 <storage_type id="single_SSD" model="linear_no_lat" size="500GiB">
755 <model_prop id="Bwrite" value="60MBps" />
756 <model_prop id="Bread" value="200MBps" />
757 <model_prop id="Bconnection" value="220MBps" />
760 <storage id="Disk2" typeId="single_SSD"
761 content="content/win_storage_content.txt"
763 <storage id="Disk4" typeId="single_SSD"
764 content="content/small_content.txt"
767 <host id="alice" speed="1Gf">
768 <mount storageId="Disk2" name="c:"/>
771 <host id="denise" speed="1Gf">
772 <mount storageId="Disk2" name="c:"/>
773 <mount storageId="Disk4" name="/home"/>
777 This example is quite interesting, as the same device, called "Disk2", is mounted by
778 two hosts at the same time! Note, however, that the host called ``alice`` is actually
779 attached to this storage, as can be seen in the \ref pf_storage_entity_storage "<storage>"
780 tag. This means that ``denise`` must access this storage through the network, but SimGrid automatically takes
781 care of that for you.
783 Furthermore, this example shows that ``denise`` has mounted two storages with different
784 filesystem types (unix and windows). In general, a host can mount as many storage devices as
788 Again, the difference between ``attach`` and ``mount`` is simply that
789 an attached storage is always physically inside (or connected to) that machine;
790 for instance, a USB stick is attached to one and only one machine (where it's plugged-in)
791 but it can only be mounted on others, as mounted storage can also be a remote location.
793 ###### Example files #####
795 \verbinclude example_filelist_xmltag_mount
797 \subsubsection pf_storage_example_files Example files
799 Several examples were already discussed above; if you're interested in full examples,
800 check the the following platforms:
802 1. ``examples/platforms/storage.xml``
803 2. ``examples/platforms/remote_io.xml``
805 If you're looking for some examplary C code, you may find the source code
806 available in the directory ``examples/msg/io/`` useful.
808 \subsubsection pf_storage_examples_modelling Modelling different situations
810 The storage functionality of SimGrid is type-agnostic, that is, the implementation
811 does not presume any type of storage, such as HDDs/SSDs, RAM,
812 CD/DVD devices, USB sticks etc.
814 This allows the user to apply the simulator for a wide variety of scenarios; one
815 common scenario would be the access of remote RAM.
817 #### Modelling the access of remote RAM ####
819 How can this be achieved in SimGrid? Let's assume we have a setup where three hosts
820 (HostA, HostB, HostC) need to access remote RAM:
830 An easy way to model this scenario is to setup and define the RAM via the
831 \ref pf_storage_entity_storage "storage" and \ref pf_storage_entity_storage_type "storage type"
832 entities and attach it to a remote dummy host; then, every host can have their own links
833 to this host (modelling for instance certain scenarios, such as PCIe ...)
838 RAM - Dummy -- Host B
843 Now, if read from this storage, the host that mounts this storage
844 communicates to the dummy host which reads from RAM and
845 sends the information back.
848 \section pf_routing Routing
850 To achieve high performance, the routing tables used within SimGrid are
851 static. This means that routing between two nodes is calculated once
852 and will not change during execution. The SimGrid team chose to use this
853 approach as it is rare to have a real deficiency of a resource;
854 most of the time, a communication fails because the links experience too much
855 congestion and hence, your connection stops before the timeout or
856 because the computer designated to be the destination of that message
859 We also chose to use shortest paths algorithms in order to emulate
860 routing. Doing so is consistent with the reality: [RIP](https://en.wikipedia.org/wiki/Routing_Information_Protocol),
861 [OSPF](https://en.wikipedia.org/wiki/Open_Shortest_Path_First), [BGP](https://en.wikipedia.org/wiki/Border_Gateway_Protocol)
862 are all calculating shortest paths. They do require some time to converge, but
863 eventually, when the routing tables have stabilized, your packets will follow
866 \subsection pf_tag_zone <zone>
868 Before SimGrid v3.16, networking zones used to be called Autonomous
869 Systems, but this was misleading as zones may include other zones in a
870 hierarchical manner. If you find any remaining reference to ASes,
871 please report this as a bug.
873 Attribute | Value | Description
874 ----------- | ------------------------------------------------- | ----------------------------------------------
875 id | String (mandatory) | The identifier of this zone (must be unique)
876 routing | One of the existing routing algorithm (mandatory) | See Section \ref pf_rm for details.
880 <AS id="AS0" routing="Full">
881 <host id="host1" speed="1000000000"/>
882 <host id="host2" speed="1000000000"/>
883 <link id="link1" bandwidth="125000000" latency="0.000100"/>
884 <route src="host1" dst="host2"><link_ctn id="link1"/></route>
888 In this example, AS0 contains two hosts (host1 and host2). The route
889 between the hosts goes through link1.
891 \subsection pf_rm Routing models
893 For each AS, you must define explicitly which routing model will
894 be used. There are 3 different categories for routing models:
896 1. \ref pf_routing_model_shortest_path "Shortest-path" based models: SimGrid calculates shortest
897 paths and manages them. Behaves more or less like most real life
899 2. \ref pf_routing_model_manual "Manually-entered" route models: you have to define all routes
900 manually in the platform description file; this can become
901 tedious very quickly, as it is very verbose.
902 Consistent with some manually managed real life routing.
903 3. \ref pf_routing_model_simple "Simple/fast models": those models offer fast, low memory routing
904 algorithms. You should consider to use this type of model if
905 you can make some assumptions about your AS.
906 Routing in this case is more or less ignored.
908 \subsubsection pf_raf The router affair
910 Using routers becomes mandatory when using shortest-path based
911 models or when using the bindings to the ns-3 packet-level
912 simulator instead of the native analytical network model implemented
915 For graph-based shortest path algorithms, routers are mandatory, because these
916 algorithms require a graph as input and so we need to have source and
917 destination for each edge.
919 Routers are naturally an important concept ns-3 since the
920 way routers run the packet routing algorithms is actually simulated.
921 SimGrid's analytical models however simply aggregate the routing time
922 with the transfer time.
924 So why did we incorporate routers in SimGrid? Rebuilding a graph representation
925 only from the route information turns out to be a very difficult task, because
926 of the missing information about how routes intersect. That is why we
927 introduced routers, which are simply used to express these intersection points.
928 It is important to understand that routers are only used to provide topological
931 To express this topological information, a <b>route</b> has to be
932 defined in order to declare which link is connected to a router.
935 \subsubsection pf_routing_model_shortest_path Shortest-path based models
937 The following table shows all the models that compute routes using
938 shortest-paths algorithms are currently available in SimGrid. More detail on how
939 to choose the best routing model is given in the Section called \"\ref pf_routing_howto_choose_wisely\".
941 | Name | Description |
942 | --------------------------------------------------- | -------------------------------------------------------------------------- |
943 | \ref pf_routing_model_floyd "Floyd" | Floyd routing data. Pre-calculates all routes once |
944 | \ref pf_routing_model_dijkstra "Dijkstra" | Dijkstra routing data. Calculates routes only when needed |
945 | \ref pf_routing_model_dijkstracache "DijkstraCache" | Dijkstra routing data. Handles some cache for already calculated routes. |
947 All those shortest-path models are instanciated in the same way and are
948 completely interchangeable. Here are some examples:
950 \anchor pf_routing_model_floyd
955 <AS id="AS0" routing="Floyd">
957 <cluster id="my_cluster_1" prefix="c-" suffix=""
958 radical="0-1" speed="1000000000" bw="125000000" lat="5E-5"
959 router_id="router1"/>
961 <AS id="AS1" routing="None">
962 <host id="host1" speed="1000000000"/>
965 <link id="link1" bandwidth="100000" latency="0.01"/>
967 <ASroute src="my_cluster_1" dst="AS1"
970 <link_ctn id="link1"/>
976 ASroute given at the end gives a topological information: link1 is
977 between router1 and host1.
979 #### Example platform files ####
981 This is an automatically generated list of example files that use the Floyd
982 routing model (the path is given relative to SimGrid's source directory)
984 \verbinclude example_filelist_routing_floyd
986 \anchor pf_routing_model_dijkstra
989 #### Example platform files ####
991 This is an automatically generated list of example files that use the Dijkstra
992 routing model (the path is given relative to SimGrid's source directory)
994 \verbinclude example_filelist_routing_dijkstra
998 <AS id="AS_2" routing="Dijkstra">
999 <host id="AS_2_host1" speed="1000000000"/>
1000 <host id="AS_2_host2" speed="1000000000"/>
1001 <host id="AS_2_host3" speed="1000000000"/>
1002 <link id="AS_2_link1" bandwidth="1250000000" latency="5E-4"/>
1003 <link id="AS_2_link2" bandwidth="1250000000" latency="5E-4"/>
1004 <link id="AS_2_link3" bandwidth="1250000000" latency="5E-4"/>
1005 <link id="AS_2_link4" bandwidth="1250000000" latency="5E-4"/>
1006 <router id="central_router"/>
1007 <router id="AS_2_gateway"/>
1008 <!-- routes providing topological information -->
1009 <route src="central_router" dst="AS_2_host1"><link_ctn id="AS_2_link1"/></route>
1010 <route src="central_router" dst="AS_2_host2"><link_ctn id="AS_2_link2"/></route>
1011 <route src="central_router" dst="AS_2_host3"><link_ctn id="AS_2_link3"/></route>
1012 <route src="central_router" dst="AS_2_gateway"><link_ctn id="AS_2_link4"/></route>
1016 \anchor pf_routing_model_dijkstracache
1017 ### DijkstraCache ###
1019 DijkstraCache example:
1021 <AS id="AS_2" routing="DijkstraCache">
1022 <host id="AS_2_host1" speed="1000000000"/>
1024 (platform unchanged compared to upper example)
1027 #### Example platform files ####
1029 This is an automatically generated list of example files that use the DijkstraCache
1030 routing model (the path is given relative to SimGrid's source directory):
1032 Editor's note: At the time of writing, no platform file used this routing model - so
1033 if there are no example files listed here, this is likely to be correct.
1035 \verbinclude example_filelist_routing_dijkstra_cache
1037 \subsubsection pf_routing_model_manual Manually-entered route models
1039 | Name | Description |
1040 | ---------------------------------- | ------------------------------------------------------------------------------ |
1041 | \ref pf_routing_model_full "Full" | You have to enter all necessary routers manually; that is, every single route. This may consume a lot of memory when the XML is parsed and might be tedious to write; i.e., this is only recommended (if at all) for small platforms. |
1043 \anchor pf_routing_model_full
1048 <AS id="AS0" routing="Full">
1049 <host id="host1" speed="1000000000"/>
1050 <host id="host2" speed="1000000000"/>
1051 <link id="link1" bandwidth="125000000" latency="0.000100"/>
1052 <route src="host1" dst="host2"><link_ctn id="link1"/></route>
1056 #### Example platform files ####
1058 This is an automatically generated list of example files that use the Full
1059 routing model (the path is given relative to SimGrid's source directory):
1061 \verbinclude example_filelist_routing_full
1063 \subsubsection pf_routing_model_simple Simple/fast models
1065 | Name | Description |
1066 | ---------------------------------------- | ------------------------------------------------------------------------------ |
1067 | \ref pf_routing_model_cluster "Cluster" | This is specific to the \ref pf_tag_cluster "<cluster/>" tag and should not be used by the user, as several assumptions are made. |
1068 | \ref pf_routing_model_none "None" | No routing at all. Unless you know what you're doing, avoid using this mode in combination with a non-constant network model. |
1069 | \ref pf_routing_model_vivaldi "Vivaldi" | Perfect when you want to use coordinates. Also see the corresponding \ref pf_P2P_tags "P2P section" below. |
1071 \anchor pf_routing_model_cluster
1075 In this mode, the \ref pf_cabinet "<cabinet/>" tag is available.
1077 #### Example platform files ####
1079 This is an automatically generated list of example files that use the Cluster
1080 routing model (the path is given relative to SimGrid's source directory):
1082 \verbinclude example_filelist_routing_cluster
1084 \anchor pf_routing_model_none
1087 This model does exactly what it's name advertises: Nothing. There is no routing
1088 available within this model and if you try to communicate within the AS that
1089 uses this model, SimGrid will fail unless you have explicitly activated the
1090 \ref options_model_select_network_constant "Constant Network Model" (this model charges
1091 the same for every single communication). It should
1092 be noted, however, that you can still attach an \ref pf_tag_asroute "ASroute",
1093 as is demonstrated in the example below:
1095 \verbinclude platforms/cluster_and_one_host.xml
1097 #### Example platform files ####
1099 This is an automatically generated list of example files that use the None
1100 routing model (the path is given relative to SimGrid's source directory):
1102 \verbinclude example_filelist_routing_none
1105 \anchor pf_routing_model_vivaldi
1108 For more information on how to use the [Vivaldi Coordinates](https://en.wikipedia.org/wiki/Vivaldi_coordinates),
1109 see also Section \ref pf_P2P_tags "P2P tags".
1111 Note that it is possible to combine the Vivaldi routing model with other routing models;
1112 an example can be found in the file \c examples/platforms/cloud.xml. This
1113 examples models a NetZone using Vivaldi that contains other NetZones that use different
1116 #### Example platform files ####
1118 This is an automatically generated list of example files that use the None
1119 routing model (the path is given relative to SimGrid's source directory):
1121 \verbinclude example_filelist_routing_vivaldi
1124 \subsection ps_dec Defining routes
1126 There are currently four different ways to define routes:
1128 | Name | Description |
1129 | ------------------------------------------------- | ----------------------------------------------------------------------------------- |
1130 | \ref pf_tag_route "route" | Used to define route between host/router |
1131 | \ref pf_tag_zoneroute "zoneRoute" | Used to define route between different zones |
1132 | \ref pf_tag_bypassroute "bypassRoute" | Used to supersede normal routes as calculated by the network model between host/router; e.g., can be used to use a route that is not the shortest path for any of the shortest-path routing models. |
1133 | \ref pf_tag_bypassasroute "bypassZoneRoute" | Used in the same way as bypassRoute, but for zones |
1135 Basically all those tags will contain an (ordered) list of references
1136 to link that compose the route you want to define.
1138 Consider the example below:
1141 <route src="Alice" dst="Bob">
1142 <link_ctn id="link1"/>
1143 <link_ctn id="link2"/>
1144 <link_ctn id="link3"/>
1148 The route here from host Alice to Bob will be first link1, then link2,
1149 and finally link3. What about the reverse route? \ref pf_tag_route "Route" and
1150 \ref pf_tag_zoneroute "ASroute" have an optional attribute \c symmetrical, that can
1151 be either \c YES or \c NO. \c YES means that the reverse route is the same
1152 route in the inverse order, and is set to \c YES by default. Note that
1153 this is not the case for bypass*Route, as it is more probable that you
1154 want to bypass only one default route.
1156 For an \ref pf_tag_zoneroute "ASroute", things are just slightly more complicated, as you have
1157 to give the id of the gateway which is inside the AS you want to access ...
1158 So it looks like this:
1161 <ASroute src="AS1" dst="AS2"
1162 gw_src="router1" gw_dst="router2">
1163 <link_ctn id="link1"/>
1167 gw == gateway, so when any message are trying to go from AS1 to AS2,
1168 it means that it must pass through router1 to get out of the AS, then
1169 pass through link1, and get into AS2 by being received by router2.
1170 router1 must belong to AS1 and router2 must belong to AS2.
1172 \subsubsection pf_tag_linkctn <link_ctn>
1174 This entity has only one purpose: Refer to an already existing
1175 \ref pf_tag_link "<link/>" when defining a route, i.e., it
1176 can only occur as a child of \ref pf_tag_route "<route/>"
1178 | Attribute name | Mandatory | Values | Description |
1179 | --------------- | --------- | ------ | ----------- |
1180 | id | yes | String | The identifier of the link that should be added to the route. |
1181 | direction | maybe | UP\|DOWN | If the link referenced by \c id has been declared as \ref pf_sharing_policy_fullduplex "FULLDUPLEX", this indicates which direction the route traverses through this link: UP or DOWN. If you don't use FULLDUPLEX, do not use this attribute or SimGrid will not find the right link.
1183 #### Example Files ####
1185 This is an automatically generated list of example files that use the \c <link_ctn/>
1186 entity (the path is given relative to SimGrid's source directory):
1188 \verbinclude example_filelist_xmltag_linkctn
1190 \subsubsection pf_tag_zoneroute <zoneRoute>
1192 The purpose of this entity is to define a route between two ASes.
1193 This is mainly useful when you're in the \ref pf_routing_model_full "Full routing model".
1195 #### Attributes ####
1197 | Attribute name | Mandatory | Values | Description |
1198 | --------------- | --------- | ------ | ----------- |
1199 | src | yes | String | The identifier of the source AS |
1200 | dst | yes | String | See the \c src attribute |
1201 | gw_src | yes | String | The gateway that will be used within the src AS; this can be any \ref pf_tag_host "Host" or \ref pf_router "Router" defined within the src AS. |
1202 | gw_dst | yes | String | Same as \c gw_src, but with the dst AS instead. |
1203 | symmetrical | no | YES\|NO (Default: YES) | If this route is symmetric, the opposite route (from dst to src) will also be declared implicitly. |
1208 <AS id="AS0" routing="Full">
1209 <cluster id="my_cluster_1" prefix="c-" suffix=".me"
1210 radical="0-149" speed="1000000000" bw="125000000" lat="5E-5"
1211 bb_bw="2250000000" bb_lat="5E-4"/>
1213 <cluster id="my_cluster_2" prefix="c-" suffix=".me"
1214 radical="150-299" speed="1000000000" bw="125000000" lat="5E-5"
1215 bb_bw="2250000000" bb_lat="5E-4"/>
1217 <link id="backbone" bandwidth="1250000000" latency="5E-4"/>
1219 <ASroute src="my_cluster_1" dst="my_cluster_2"
1220 gw_src="c-my_cluster_1_router.me"
1221 gw_dst="c-my_cluster_2_router.me">
1222 <link_ctn id="backbone"/>
1224 <ASroute src="my_cluster_2" dst="my_cluster_1"
1225 gw_src="c-my_cluster_2_router.me"
1226 gw_dst="c-my_cluster_1_router.me">
1227 <link_ctn id="backbone"/>
1232 \subsubsection pf_tag_route <route>
1234 The principle is the same as for
1235 \ref pf_tag_zoneroute "ASroute": The route contains a list of links that
1236 provide a path from \c src to \c dst. Here, \c src and \c dst can both be either a
1237 \ref pf_tag_host "host" or \ref pf_router "router". This is mostly useful for the
1238 \ref pf_routing_model_full "Full routing model" as well as for the
1239 \ref pf_routing_model_shortest_path "shortest-paths" based models (as they require
1240 topological information).
1243 | Attribute name | Mandatory | Values | Description |
1244 | --------------- | --------- | ---------------------- | ----------- |
1245 | src | yes | String | The value given to the source's "id" attribute |
1246 | dst | yes | String | The value given to the destination's "id" attribute. |
1247 | symmetrical | no | YES\| NO (Default: YES) | If this route is symmetric, the opposite route (from dst to src) will also be declared implicitly. |
1252 A route in the \ref pf_routing_model_full "Full routing model" could look like this:
1254 <route src="Tremblay" dst="Bourassa">
1255 <link_ctn id="4"/><link_ctn id="3"/><link_ctn id="2"/><link_ctn id="0"/><link_ctn id="1"/><link_ctn id="6"/><link_ctn id="7"/>
1259 A route in the \ref pf_routing_model_shortest_path "Shortest-Path routing model" could look like this:
1261 <route src="Tremblay" dst="Bourassa">
1266 You must only have one link in your routes when you're using them to provide
1267 topological information, as the routes here are simply the edges of the
1268 (network-)graph and the employed algorithms need to know which edge connects
1269 which pair of entities.
1271 \subsubsection pf_tag_bypassasroute bypassASroute
1273 As said before, once you choose
1274 a model, it (most likely; the constant network model, for example, doesn't) calculates routes for you. But maybe you want to
1275 define some of your routes, which will be specific. You may also want
1276 to bypass some routes defined in lower level AS at an upper stage:
1277 <b>bypassASroute</b> is the tag you're looking for. It allows to
1278 bypass routes defined between already defined between AS (if you want
1279 to bypass route for a specific host, you should just use byPassRoute).
1280 The principle is the same as ASroute : <b>bypassASroute</b> contains
1281 list of links that are in the path between src and dst.
1283 #### Attributes ####
1285 | Attribute name | Mandatory | Values | Description |
1286 | --------------- | --------- | ---------------------- | ----------- |
1287 | src | yes | String | The value given to the source AS's "id" attribute |
1288 | dst | yes | String | The value given to the destination AS's "id" attribute. |
1289 | gw_src | yes | String | The value given to the source gateway's "id" attribute; this can be any host or router within the src AS |
1290 | gw_dst | yes | String | The value given to the destination gateway's "id" attribute; this can be any host or router within the dst AS|
1291 | symmetrical | no | YES\| NO (Default: YES) | If this route is symmetric, the opposite route (from dst to src) will also be declared implicitly. |
1296 <bypassASRoute src="my_cluster_1" dst="my_cluster_2"
1297 gw_src="my_cluster_1_router"
1298 gw_dst="my_cluster_2_router">
1299 <link_ctn id="link_tmp"/>
1303 This example shows that link \c link_tmp (definition not displayed here) directly
1304 connects the router \c my_cluster_1_router in the source cluster to the router
1305 \c my_cluster_2_router in the destination router. Additionally, as the \c symmetrical
1306 attribute was not given, this route is presumed to be symmetrical.
1308 \subsubsection pf_tag_bypassroute bypassRoute
1310 As said before, once you choose
1311 a model, it (most likely; the constant network model, for example, doesn't) calculates routes for you. But maybe you want to
1312 define some of your routes, which will be specific. You may also want
1313 to bypass some routes defined in lower level AS at an upper stage :
1314 <b>bypassRoute</b> is the tag you're looking for. It allows to bypass
1315 routes defined between <b>host/router</b>. The principle is the same
1316 as route : <b>bypassRoute</b> contains list of links references of
1317 links that are in the path between src and dst.
1319 #### Attributes ####
1321 | Attribute name | Mandatory | Values | Description |
1322 | --------------- | --------- | ---------------------- | ----------- |
1323 | src | yes | String | The value given to the source AS's "id" attribute |
1324 | dst | yes | String | The value given to the destination AS's "id" attribute. |
1325 | symmetrical | no | YES \| NO (Default: YES) | If this route is symmetric, the opposite route (from dst to src) will also be declared implicitly. |
1330 <bypassRoute src="host_1" dst="host_2">
1331 <link_ctn id="link_tmp"/>
1335 This example shows that link \c link_tmp (definition not displayed here) directly
1336 connects host \c host_1 to host \c host_2. Additionally, as the \c symmetrical
1337 attribute was not given, this route is presumed to be symmetrical.
1339 \subsection pb_baroex Basic Routing Example
1341 Let's say you have an AS named AS_Big that contains two other AS, AS_1
1342 and AS_2. If you want to make a host (h1) from AS_1 with another one
1343 (h2) from AS_2 then you'll have to proceed as follows:
1344 \li First, you have to ensure that a route is defined from h1 to the
1345 AS_1's exit gateway and from h2 to AS_2's exit gateway.
1346 \li Then, you'll have to define a route between AS_1 to AS_2. As those
1347 AS are both resources belonging to AS_Big, then it has to be done
1348 at AS_big level. To define such a route, you have to give the
1349 source AS (AS_1), the destination AS (AS_2), and their respective
1350 gateway (as the route is effectively defined between those two
1351 entry/exit points). Elements of this route can only be elements
1352 belonging to AS_Big, so links and routers in this route should be
1353 defined inside AS_Big. If you choose some shortest-path model,
1354 this route will be computed automatically.
1356 As said before, there are mainly 2 tags for routing :
1357 \li <b>ASroute</b>: to define routes between two <b>AS</b>
1358 \li <b>route</b>: to define routes between two <b>host/router</b>
1360 As we are dealing with routes between AS, it means that those we'll
1361 have some definition at AS_Big level. Let consider AS_1 contains 1
1362 host, 1 link and one router and AS_2 3 hosts, 4 links and one router.
1363 There will be a central router, and a cross-like topology. At the end
1364 of the crosses arms, you'll find the 3 hosts and the router that will
1365 act as a gateway. We have to define routes inside those two AS. Let
1366 say that AS_1 contains full routes, and AS_2 contains some Floyd
1367 routing (as we don't want to bother with defining all routes). As
1368 we're using some shortest path algorithms to route into AS_2, we'll
1369 then have to define some <b>route</b> to gives some topological
1370 information to SimGrid. Here is a file doing it all :
1373 <AS id="AS_Big" routing="Dijkstra">
1374 <AS id="AS_1" routing="Full">
1375 <host id="AS_1_host1" speed="1000000000"/>
1376 <link id="AS_1_link" bandwidth="1250000000" latency="5E-4"/>
1377 <router id="AS_1_gateway"/>
1378 <route src="AS_1_host1" dst="AS_1_gateway">
1379 <link_ctn id="AS_1_link"/>
1382 <AS id="AS_2" routing="Floyd">
1383 <host id="AS_2_host1" speed="1000000000"/>
1384 <host id="AS_2_host2" speed="1000000000"/>
1385 <host id="AS_2_host3" speed="1000000000"/>
1386 <link id="AS_2_link1" bandwidth="1250000000" latency="5E-4"/>
1387 <link id="AS_2_link2" bandwidth="1250000000" latency="5E-4"/>
1388 <link id="AS_2_link3" bandwidth="1250000000" latency="5E-4"/>
1389 <link id="AS_2_link4" bandwidth="1250000000" latency="5E-4"/>
1390 <router id="central_router"/>
1391 <router id="AS_2_gateway"/>
1392 <!-- routes providing topological information -->
1393 <route src="central_router" dst="AS_2_host1"><link_ctn id="AS_2_link1"/></route>
1394 <route src="central_router" dst="AS_2_host2"><link_ctn id="AS_2_link2"/></route>
1395 <route src="central_router" dst="AS_2_host3"><link_ctn id="AS_2_link3"/></route>
1396 <route src="central_router" dst="AS_2_gateway"><link_ctn id="AS_2_link4"/></route>
1398 <link id="backbone" bandwidth="1250000000" latency="5E-4"/>
1400 <ASroute src="AS_1" dst="AS_2"
1401 gw_src="AS_1_gateway"
1402 gw_dst="AS_2_gateway">
1403 <link_ctn id="backbone"/>
1408 \section pf_other Other tags
1410 The following tags can be used inside a \<platform\> tag even if they are not
1411 directly describing the platform:
1413 - @ref pf_tag_config passes configuration options, e.g. to change the network model;
1414 - @ref pf_tag_prop gives user-defined properties to various elements
1416 \subsection pf_tag_config <config>
1418 Adding configuration flags into the platform file is particularly
1419 useful when the described platform is best used with specific
1420 flags. For example, you could finely tune SMPI in your platform file directly.
1422 | Attribute | Values | Description |
1423 | ---------- | ------------------- | ---------------------------------------------- |
1424 | id | String (optional) | This optional identifier is ignored by SimGrid |
1426 * **Included tags:** @ref pf_tag_prop to specify a given configuration item (see @ref options).
1428 Any such configuration must be given at the very top of the platform file.
1433 <?xml version='1.0'?>
1434 <!DOCTYPE platform SYSTEM "http://simgrid.gforge.inria.fr/simgrid.dtd">
1435 <platform version="4">
1437 <prop id="maxmin/precision" value="0.000010" />
1438 <prop id="cpu/optim" value="TI" />
1439 <prop id="network/model" value="SMPI" />
1440 <prop id="smpi/bw-factor" value="65472:0.940694;15424:0.697866;9376:0.58729" />
1443 <AS id="AS0" routing="Full">
1447 \subsection pf_tag_prop <prop>
1449 Defines a user-defined property, identified with a name and having a
1450 value. You can specify such properties to most kind of resources:
1451 @ref pf_tag_zone, @ref pf_tag_host, @ref pf_tag_storage,
1452 @ref pf_tag_cluster and @ref pf_tag_link. These values can be retrieved
1453 at runtime with MSG_zone_property() or simgrid::s4u::NetZone::property(),
1454 or similar functions.
1456 | Attribute | Values | Description |
1457 | --------- | ---------------------- | ----------------------------------------------------------------------------------------- |
1458 | id | String (mandatory) | Identifier of this property. Must be unique for a given property holder, eg host or link. |
1459 | value | String (mandatory) | Value of this property; The semantic is completely up to you. |
1461 * **Included tags:** none.
1466 <prop id="Operating System" value="Linux" />
1470 \subsection pf_trace trace and trace_connect
1472 Both tags are an alternate way to pass files containing information on
1473 availability, state etc. to an entity. (See also, for instance, Section \ref
1474 pf_host_churn "Churn", as described for the host entity.) Instead of referring
1475 to the file directly in the host, link, or cluster tag, you proceed by defining
1476 a trace with an id corresponding to a file, later a host/link/cluster, and
1477 finally using trace_connect you say that the file trace must be used by the
1484 <AS id="AS0" routing="Full">
1485 <host id="bob" speed="1000000000"/>
1487 <trace id="myTrace" file="bob.trace" periodicity="1.0"/>
1488 <trace_connect trace="myTrace" element="bob" kind="POWER"/>
1492 The order here is important. \c trace_connect must come
1493 after the elements \c trace and \c host, as both the host
1494 and the trace definition must be known when \c trace_connect
1495 is parsed; the order of \c trace and \c host is arbitrary.
1498 #### \c trace attributes ####
1501 | Attribute name | Mandatory | Values | Description |
1502 | --------------- | --------- | ---------------------- | ----------- |
1503 | id | yes | String | Identifier of this trace; this is the name you pass on to \c trace_connect. |
1504 | file | no | String | Filename of the file that contains the information - the path must follow the style of your OS. You can omit this, but then you must specifiy the values inside of <trace> and </trace> - see the example below. |
1505 | trace_periodicity | yes | String | This is the same as for \ref pf_tag_host "hosts" (see there for details) |
1507 Here is an example of trace when no file name is provided:
1510 <trace id="myTrace" periodicity="1.0">
1517 #### \c trace_connect attributes ####
1519 | Attribute name | Mandatory | Values | Description |
1520 | --------------- | --------- | ---------------------- | ----------- |
1521 | kind | no | HOST_AVAIL\|POWER\|<br/>LINK_AVAIL\|BANDWIDTH\|LATENCY (Default: HOST_AVAIL) | Describes the kind of trace. |
1522 | trace | yes | String | Identifier of the referenced trace (specified of the trace's \c id attribute) |
1523 | element | yes | String | The identifier of the referenced entity as given by its \c id attribute |
1525 \section pf_hints Hints, tips and frequently requested features
1527 Now you should know at least the syntax and be able to create a
1528 platform by your own. However, after having ourselves wrote some platforms, there
1529 are some best practices you should pay attention to in order to
1530 produce good platform and some choices you can make in order to have
1531 faster simulations. Here's some hints and tips, then.
1533 @subsection pf_hints_search Finding the platform example that you need
1535 Most platform files that we ship are in the @c examples/platforms
1536 folder. The good old @c grep tool can find the examples you need when
1537 wondering on a specific XML tag. Here is an example session searching
1538 for @ref pf_trace "trace_connect":
1541 % cd examples/platforms
1542 % grep -R -i -n --include="*.xml" "trace_connect" .
1543 ./two_hosts_platform_with_availability_included.xml:26:<trace_connect kind="SPEED" trace="A" element="Cpu A"/>
1544 ./two_hosts_platform_with_availability_included.xml:27:<trace_connect kind="HOST_AVAIL" trace="A_failure" element="Cpu A"/>
1545 ./two_hosts_platform_with_availability_included.xml:28:<trace_connect kind="SPEED" trace="B" element="Cpu B"/>
1546 ./two_hosts.xml:17: <trace_connect trace="Tremblay_power" element="Tremblay" kind="SPEED"/>
1549 \subsection pf_as_h AS Hierarchy
1550 The AS design allows SimGrid to go fast, because computing route is
1551 done only for the set of resources defined in this AS. If you're using
1552 only a big AS containing all resource with no AS into it and you're
1553 using Full model, then ... you'll loose all interest into it. On the
1554 other hand, designing a binary tree of AS with, at the lower level,
1555 only one host, then you'll also loose all the good AS hierarchy can
1556 give you. Remind you should always be "reasonable" in your platform
1557 definition when choosing the hierarchy. A good choice if you try to
1558 describe a real life platform is to follow the AS described in
1559 reality, since this kind of trade-off works well for real life
1562 \subsection pf_exit_as Exit AS: why and how
1563 Users that have looked at some of our platforms may have notice a
1564 non-intuitive schema ... Something like that :
1568 <AS id="AS_4" routing="Full">
1569 <AS id="exitAS_4" routing="Full">
1570 <router id="router_4"/>
1572 <cluster id="cl_4_1" prefix="c_4_1-" suffix="" radical="1-20" speed="1000000000" bw="125000000" lat="5E-5" bb_bw="2250000000" bb_lat="5E-4"/>
1573 <cluster id="cl_4_2" prefix="c_4_2-" suffix="" radical="1-20" speed="1000000000" bw="125000000" lat="5E-5" bb_bw="2250000000" bb_lat="5E-4"/>
1574 <link id="4_1" bandwidth="2250000000" latency="5E-5"/>
1575 <link id="4_2" bandwidth="2250000000" latency="5E-5"/>
1576 <link id="bb_4" bandwidth="2250000000" latency="5E-4"/>
1577 <ASroute src="cl_4_1"
1579 gw_src="c_4_1-cl_4_1_router"
1580 gw_dst="c_4_2-cl_4_2_router">
1581 <link_ctn id="4_1"/>
1582 <link_ctn id="bb_4"/>
1583 <link_ctn id="4_2"/>
1585 <ASroute src="cl_4_1"
1587 gw_src="c_4_1-cl_4_1_router"
1589 <link_ctn id="4_1"/>
1590 <link_ctn id="bb_4"/>
1592 <ASroute src="cl_4_2"
1594 gw_src="c_4_2-cl_4_2_router"
1596 <link_ctn id="4_2"/>
1597 <link_ctn id="bb_4"/>
1602 In the AS_4, you have an exitAS_4 defined, containing only one router,
1603 and routes defined to that AS from all other AS (as cluster is only a
1604 shortcut for an AS, see cluster description for details). If there was
1605 an upper AS, it would define routes to and from AS_4 with the gateway
1606 router_4. It's just because, as we did not allowed (for performances
1607 issues) to have routes from an AS to a single host/router, you have to
1608 enclose your gateway, when you have AS included in your AS, within an
1609 AS to define routes to it.
1611 \subsection pf_P2P_tags P2P or how to use coordinates
1612 SimGrid allows you to use some coordinated-based system, like vivaldi,
1613 to describe a platform. The main concept is that you have some peers
1614 that are located somewhere: this is the function of the
1615 <b>coordinates</b> of the \<peer\> or \<host\> tag. There's nothing
1616 complicated in using it, here is an example:
1619 <?xml version='1.0'?>
1620 <!DOCTYPE platform SYSTEM "http://simgrid.gforge.inria.fr/simgrid.dtd">
1621 <platform version="4">
1623 <AS id="AS0" routing="Vivaldi">
1624 <host id="100030591" coordinates="25.5 9.4 1.4" speed="1.5Gf" />
1625 <host id="100036570" coordinates="-12.7 -9.9 2.1" speed="7.3Gf" />
1627 <host id="100429957" coordinates="17.5 6.7 18.8" speed="8.3Gf" />
1632 Coordinates are then used to calculate latency (in microseconds)
1633 between two hosts by calculating the distance between the two hosts
1634 coordinates with the following formula: distance( (x1, y1, z1), (x2,
1635 y2, z2) ) = euclidian( (x1,y1), (x2,y2) ) + abs(z1) + abs(z2)
1637 In other words, we take the euclidian distance on the two first
1638 dimensions, and then add the absolute values found on the third
1639 dimension. This may seem strange, but it was found to allow better
1640 approximations of the latency matrices (see the paper describing
1643 Note that the previous example defines a routing directly between hosts but it could be also used to define a routing between AS.
1644 That is for example what is commonly done when using peers (see Section \ref pf_peer).
1646 <?xml version='1.0'?>
1647 <!DOCTYPE platform SYSTEM "http://simgrid.gforge.inria.fr/simgrid.dtd">
1648 <platform version="4">
1650 <AS id="AS0" routing="Vivaldi">
1651 <peer id="peer-0" coordinates="173.0 96.8 0.1" speed="730Mf" bw_in="13.38MBps" bw_out="1.024MBps" lat="500us"/>
1652 <peer id="peer-1" coordinates="247.0 57.3 0.6" speed="730Mf" bw_in="13.38MBps" bw_out="1.024MBps" lat="500us" />
1653 <peer id="peer-2" coordinates="243.4 58.8 1.4" speed="730Mf" bw_in="13.38MBps" bw_out="1.024MBps" lat="500us" />
1657 In such a case though, we connect the AS created by the <b>peer</b> tag with the Vivaldi routing mechanism.
1658 This means that to route between AS1 and AS2, it will use the coordinates of router_AS1 and router_AS2.
1659 This is currently a convention and we may offer to change this convention in the DTD later if needed.
1660 You may have noted that conveniently, a peer named FOO defines an AS named FOO and a router named router_FOO, which is why it works seamlessly with the <b>peer</b> tag.
1663 \subsection pf_routing_howto_choose_wisely Choosing wisely the routing model to use
1666 Choosing wisely the routing model to use can significantly fasten your
1667 simulation/save your time when writing the platform/save tremendous
1668 disk space. Here is the list of available model and their
1669 characteristics (lookup : time to resolve a route):
1671 \li <b>Full</b>: Full routing data (fast, large memory requirements,
1673 \li <b>Floyd</b>: Floyd routing data (slow initialization, fast
1674 lookup, lesser memory requirements, shortest path routing only).
1675 Calculates all routes at once at the beginning.
1676 \li <b>Dijkstra</b>: Dijkstra routing data (fast initialization, slow
1677 lookup, small memory requirements, shortest path routing only).
1678 Calculates a route when necessary.
1679 \li <b>DijkstraCache</b>: Dijkstra routing data (fast initialization,
1680 fast lookup, small memory requirements, shortest path routing
1681 only). Same as Dijkstra, except it handles a cache for latest used
1683 \li <b>None</b>: No routing (usable with Constant network only).
1684 Defines that there is no routes, so if you try to determine a
1685 route without constant network within this AS, SimGrid will raise
1687 \li <b>Vivaldi</b>: Vivaldi routing, so when you want to use coordinates
1688 \li <b>Cluster</b>: Cluster routing, specific to cluster tag, should
1691 \subsection pf_switch I want to describe a switch but there is no switch tag!
1693 Actually we did not include switch tag. But when you're trying to
1694 simulate a switch, assuming
1695 fluid bandwidth models are used (which SimGrid uses by default unless
1696 ns-3 or constant network models are activated), the limiting factor is
1697 switch backplane bandwidth. So, essentially, at least from
1698 the simulation perspective, a switch is similar to a
1699 link: some device that is traversed by flows and with some latency and
1700 so,e maximum bandwidth. Thus, you can simply simulate a switch as a
1702 can be connected to this "switch", which is then included in routes just
1706 \subsection pf_multicabinets I want to describe multi-cabinets clusters!
1708 You have several possibilities, as usual when modeling things. If your
1709 cabinets are homogeneous and the intercabinet network negligible for
1710 your study, you should just create a larger cluster with all hosts at
1713 In the rare case where your hosts are not homogeneous between the
1714 cabinets, you can create your cluster completely manually. For that,
1715 create an As using the Cluster routing, and then use one
1716 <cabinet> for each cabinet. This cabinet tag can only be used an
1717 As using the Cluster routing schema, and creating
1719 Be warned that creating a cluster manually from the XML with
1720 <cabinet>, <backbone> and friends is rather tedious. The
1721 easiest way to retrieve some control of your model without diving into
1722 the <cluster> internals is certainly to create one separate
1723 <cluster> per cabinet and interconnect them together. This is
1724 what we did in the G5K example platform for the Graphen cluster.
1726 \subsection pf_platform_multipath I want to express multipath routing in platform files!
1728 It is unfortunately impossible to express the fact that there is more
1729 than one routing path between two given hosts. Let's consider the
1730 following platform file:
1733 <route src="A" dst="B">
1736 <route src="B" dst="C">
1739 <route src="A" dst="C">
1744 Although it is perfectly valid, it does not mean that data traveling
1745 from A to C can either go directly (using link 3) or through B (using
1746 links 1 and 2). It simply means that the routing on the graph is not
1747 trivial, and that data do not following the shortest path in number of
1748 hops on this graph. Another way to say it is that there is no implicit
1749 in these routing descriptions. The system will only use the routes you
1750 declare (such as <route src="A" dst="C"><link_ctn
1751 id="3"/></route>), without trying to build new routes by aggregating
1754 You are also free to declare platform where the routing is not
1755 symmetrical. For example, add the following to the previous file:
1758 <route src="C" dst="A">
1764 This makes sure that data from C to A go through B where data from A
1765 to C go directly. Don't worry about realism of such settings since
1766 we've seen ways more weird situation in real settings (in fact, that's
1767 the realism of very regular platforms which is questionable, but
1768 that's another story).
