boinc/ssim/ssim.cpp

331 lines
8.5 KiB
C++
Raw Normal View History

// This file is part of BOINC.
// http://boinc.berkeley.edu
// Copyright (C) 2011 University of California
//
// BOINC is free software; you can redistribute it and/or modify it
// under the terms of the GNU Lesser General Public License
// as published by the Free Software Foundation,
// either version 3 of the License, or (at your option) any later version.
//
// BOINC is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
// See the GNU Lesser General Public License for more details.
//
// You should have received a copy of the GNU Lesser General Public License
// along with BOINC. If not, see <http://www.gnu.org/licenses/>.
// ssim - simulator for distributed storage
//
// Simulates the storage of files on a dynamic set of hosts.
//
// The model of the host population is:
// - hosts arrival is a Poisson process
// - host lifetime is exponentially distributed
// - the time needed to upload or download n bytes of data
// to/from a host is
// MTD*U() + n/XFER_BYTES_SEC
// where C1 and C2 are parameters and U() is a uniform random var
#define HOSTS_PER_DAY 10.
#define HOST_LIFE_MEAN 100.*86400
#define MAX_TRANSFER_DELAY 86400
#define UPLOAD_BYTES_SEC 1e6
#define DOWNLOAD_BYTES_SEC 5e6
// We simulate policies based on coding and replication.
//
// Coding means that data is divided into M = N+K units,
// of which any N are sufficient to reconstruct the data.
//
// The units in an encoding can themselves be encoded.
// In general we model C levels of encoding.
//
// The bottom-level data units ("chunks") are stored on hosts,
// with R-fold replication
#define ENCODING_N 10
#define ENCODING_K 5
#define ENCODING_M 15
#define ENCODING_LEVELS 1
#define REPLICATION_LEVEL 2
// When we need to reconstruct an encoded unit on the server,
// we try to upload N_UPLOAD subunits,
// where N <= N_UPLOAD <= M
#define N_UPLOAD 12
// Terminology:
//
// A chunk may or may not be "present_on_server".
// An encoded data unit is "present_on_server" if at least N
// of its subunits are present_on_server (recursive definition).
// A chunk is "recoverable" if it is present on at least 1 host.
// An encoded data unit is "recoverable" if at least N
// of its subunits are recoverable.
// A chunk is "uploading" if at least one of its instances
// is being uploaded to the server.
// An encoded data unit is "uploading" if at least
// 1 of its subunits is uploading,
// and at least N of its subunits are either present_on_server or uploading
// The scheduling policy can be briefly described as:
// 1) distribute chunks to hosts when possible, up to the replication level
// Put at most 1 chunk of a file on a given host.
// 2) if a data unit becomes unrecoverable,
// upload its parent unit, reconstruct the data, then do 1)
#include <stdio.h>
#include <set>
#include "des.h"
using std::set;
#define K 15
// this many packets per meta-packet
#define N 10
// need this many to reconstruct the meta-packet
#define META_K 15
// similar, meta-packets per file
#define META_N 10
SIMULATOR sim;
inline double drand() {
return (double)rand()/(double)RAND_MAX;
}
// place-holder
double ran_exp(double mean) {
// gsl_ran_exponential(mean);
return (drand() + .5)*mean;
}
struct CHUNK;
struct META_CHUNK;
struct DFILE;
struct HOST;
struct CHUNK_ON_HOST : public EVENT {
HOST* host;
CHUNK* chunk;
bool present_on_host;
bool transfer_in_progress; // upload if present_on_host, else download
virtual void handle();
};
struct HOST : public EVENT {
set<CHUNK_ON_HOST*> chunks;
virtual void handle();
};
set<HOST*> hosts;
struct HOST_ARRIVAL : public EVENT {
virtual void handle() {
HOST* h = new HOST;
h->t = t + ran_exp(HOST_LIFE_MEAN);
hosts.insert(h);
sim.insert(h);
t += ran_exp(86400./HOSTS_PER_DAY);
sim.insert(this);
}
};
struct REPORT_STATS : public EVENT {
virtual void handle() {
printf("%f: %d hosts\n", t, hosts.size());
t += 86400;
sim.insert(this);
}
};
// base class for chunks and meta-chunks
//
struct DATA_UNIT {
virtual bool recoverable(){};
// can be reconstructed w/o reconstructing parent,
// assuming that current downloads succeed
virtual void start_upload(){};
virtual void assign(){};
bool present_on_server;
bool is_uploading;
};
struct CHUNK : DATA_UNIT {
set<CHUNK_ON_HOST*> hosts;
META_CHUNK* parent;
double size;
CHUNK(META_CHUNK* mc, double s) {
parent = mc;
size = s;
}
virtual void assign();
void host_failed(CHUNK_ON_HOST* p);
void upload_complete();
};
struct META_CHUNK : DATA_UNIT {
vector<DATA_UNIT*> children;
META_CHUNK* parent;
int n_children_present;
DFILE* dfile;
bool uploading;
META_CHUNK(DFILE* d, META_CHUNK* par, double size, int encoding_level) {
dfile = d;
parent = par;
if (encoding_level) {
for (int j=0; j<ENCODING_M; j++) {
children.push_back(new META_CHUNK(
d,
this,
size/ENCODING_N,
encoding_level-1
));
}
} else {
for (int j=0; j<ENCODING_M; j++) {
children.push_back(new CHUNK(this, size/ENCODING_N));
}
}
}
int n_recoverable_children() {
int n = 0;
for (int i=0; i<ENCODING_N; i++) {
if (children[i]->recoverable()) {
n++;
}
}
}
// a child has become unrecoverable.
// reconstruct this data unit if we still can.
//
void child_unrecoverable() {
if (n_recoverable_children() >= ENCODING_N) {
for (int i=0; i<ENCODING_N; i++) {
DATA_UNIT* c = children[i];
if (c->recoverable()) {
c->start_upload();
}
}
}
}
virtual void assign() {
for (unsigned int i=0; i<children.size(); i++) {
children[i]->assign();
}
}
void child_upload_complete() {
}
void upload_complete() {
}
};
struct DFILE : EVENT {
META_CHUNK* meta_chunk;
double size;
set<HOST*> unused_hosts;
// hosts that don't have any packets of this file
// the creation of a file
//
virtual void handle() {
meta_chunk = new META_CHUNK(this, NULL, size, ENCODING_LEVELS);
meta_chunk->assign();
}
};
//////////////////// method defs ////////////////////
// transfer has finished
//
void CHUNK_ON_HOST::handle() {
transfer_in_progress = false;
if (present_on_host) {
// it was an upload
chunk->upload_complete(); // create new replicas if needed
} else {
present_on_host = true;
}
}
// the host has departed
//
void HOST:: handle() {
set<HOST*>::iterator i = hosts.find(this);
hosts.erase(i);
set<CHUNK_ON_HOST*>::iterator p;
for (p = chunks.begin(); p != chunks.end(); p++) {
CHUNK_ON_HOST* c = *p;
c->chunk->host_failed(c);
delete c;
}
}
void CHUNK::host_failed(CHUNK_ON_HOST* p) {
set<CHUNK_ON_HOST*>::iterator i = hosts.find(p);
hosts.erase(i);
if (present_on_server) {
// if data is on server, make a new replica
//
assign();
} else if (!hosts.empty()) {
// if there's another replica, start upload of 1st instance
// NOTE: all instances are inherently present_on_host
//
CHUNK_ON_HOST *c = *(hosts.begin());
c->transfer_in_progress = true;
c->t = sim.now + size/UPLOAD_BYTES_SEC;
sim.insert(c);
} else {
parent->child_unrecoverable();
}
}
void CHUNK::upload_complete() {
assign();
if (parent->uploading) {
parent->child_upload_complete();
}
}
void CHUNK::assign() {
while (hosts.size() < REPLICATION_LEVEL) {
set<HOST*>::iterator i = parent->dfile->unused_hosts.begin();
HOST* h = *i;
parent->dfile->unused_hosts.erase(i);
CHUNK_ON_HOST *c = new CHUNK_ON_HOST();
c->host = h;
c->chunk = this;
c->t = sim.now + size/DOWNLOAD_BYTES_SEC;
sim.insert(c);
}
}
set<DFILE*> dfiles;
int main() {
HOST_ARRIVAL *h = new HOST_ARRIVAL;
h->t = 0;
sim.insert(h);
REPORT_STATS* r = new REPORT_STATS;
r->t = 0;
sim.insert(r);
sim.simulate(200*86400);
}