/** @brief Autonomous Systems
*
* An AS is a network container, in charge of routing information between elements (hosts) and to the nearby ASes.
- * In SimGrid, there is a hierarchy of ASes, with a unique root AS (that you can retrieve from the s4u::Engine).
+ * In SimGrid, there is a hierarchy of ASes, ie a tree with a unique root AS, that you can retrieve from the
+ * s4u::Engine.
+ *
+ * The purpose of the kernel::routing module is to retrieve the routing path between two points in a time- and
+ * space-efficient manner. This is done by AsImpl::getGlobalRoute(), called when creating a communication to
+ * retrieve both the list of links that the create communication will use, and the summed latency that these
+ * links represent.
+ *
+ * The network could recompute the latency by itself from the list, but it would require an additional link
+ * set traversal. This operation being on the critical path of SimGrid, the routing computes the latency on the
+ * behalf of the network.
+ *
+ * Finding the path between two nodes is rather complex because we navigate a hierarchy of ASes, each of them
+ * being a full network. In addition, the routing can declare shortcuts (called bypasses), either within an AS
+ * at the route level or directly between ASes. Also, each AS can use a differing routing algorithm, depending
+ * on its class. @ref{AsFull} have a full matrix giving explicitly the path between any pair of their
+ * contained nodes, while @ref{AsDijkstra} or @ref{AsFloyd} rely on a shortest path algorithm. @ref{AsVivaldi}
+ * does not even have any link but only use only coordinate information to compute the latency.
+ *
+ * So AsImpl::getGlobalRoute builds the path recursively asking its specific information to each traversed AS with
+ * AsImpl::getLocalRoute, that is redefined in each sub-class.
+ * The algorithm for that is explained in http://hal.inria.fr/hal-00650233/
+ *
*/
XBT_PUBLIC_CLASS AsImpl : public s4u::As
{
protected:
explicit AsImpl(As * father, const char* name);
- ~AsImpl() override;
public:
/** @brief Make an host within that AS */
simgrid::s4u::Host* createHost(const char* name, std::vector<double>* speedPerPstate, int coreAmount);
+ protected:
/**
- * @brief Probe the routing path between two points
- *
- * The networking model uses this function when creating a communication
- * to retrieve both the list of links that the create communication will use,
- * and the summed latency that these links represent.
- *
- * The network could recompute the latency by itself from the list, but it would
- * require an additional link set traversal. This operation being on the critical
- * path of SimGrid, the routing computes the latency in behalf of the network.
+ * @brief Probe the routing path between two points that are local to the called AS.
*
- * Things are rather complex here because we have to find the path from ASes to ASes, and within each.
- * In addition, the different ASes may use differing routing models.
- * Some ASes may be routed in full, others may have only some connection information and use a shortest path on top
- * of that, and so on.
- * Some ASes may even not have any predefined links and use only coordinate information to compute the latency.
- *
- * So, the path is constructed recursively, with each traversed AS adding its information to the set.
- * The algorithm for that is explained in http://hal.inria.fr/hal-00650233/
- *
- * @param src Initial point of the routing path
- * @param dst Final point of the routing path
- * @param into Container into which the traversed links should be pushed
+ * @param src where from
+ * @param dst where to
+ * @param into Container into which the traversed links and gateway informations should be pushed
* @param latency Accumulator in which the latencies should be added (caller must set it to 0)
*/
- protected:
- virtual void getRouteAndLatency(NetCard * src, NetCard * dst, sg_platf_route_cbarg_t into, double* latency) = 0;
+ virtual void getLocalRoute(NetCard * src, NetCard * dst, sg_platf_route_cbarg_t into, double* latency) = 0;
/** @brief retrieves the list of all routes of size 1 (of type src x dst x Link) */
/* returns whether we found a bypass path */
bool getBypassRoute(routing::NetCard * src, routing::NetCard * dst,
/* OUT */ std::vector<surf::Link*> * links, double* latency);
public:
+ /* @brief get the route between two nodes in the full platform
+ *
+ * @param src where from
+ * @param dst where to
+ * @param links Accumulator in which all traversed links should be pushed (caller must empty it)
+ * @param latency Accumulator in which the latencies should be added (caller must set it to 0)
+ */
+ static void getGlobalRoute(routing::NetCard * src, routing::NetCard * dst,
+ /* OUT */ std::vector<surf::Link*> * links, double* latency);
+
virtual void getOneLinkRoutes(std::vector<Onelink*> * accumulator);
virtual void getGraph(xbt_graph_t graph, xbt_dict_t nodes, xbt_dict_t edges) = 0;
- static void getRouteRecursive(routing::NetCard * src, routing::NetCard * dst,
- /* OUT */ std::vector<surf::Link*> * links, double* latency);
enum class RoutingMode {
unset = 0, /**< Undefined type */
base, /**< Base case: use simple link lists for routing */
