From fd0e99e618319974320ed3fb49535aec501be1fb Mon Sep 17 00:00:00 2001
From: Douglas Rumbaugh <dbr4@psu.edu>
Date: Thu, 6 Feb 2025 15:56:33 -0500
Subject: Background compaction stuff

---
 benchmarks/include/standard_benchmarks.h          |  12 ++-
 benchmarks/tail-latency/isam_insert_dist.cpp      | 107 ++++++++++++++++++++
 benchmarks/tail-latency/standard_latency_dist.cpp | 114 ++++++++++++++++++++++
 3 files changed, 231 insertions(+), 2 deletions(-)
 create mode 100644 benchmarks/tail-latency/isam_insert_dist.cpp
 create mode 100644 benchmarks/tail-latency/standard_latency_dist.cpp

(limited to 'benchmarks')

diff --git a/benchmarks/include/standard_benchmarks.h b/benchmarks/include/standard_benchmarks.h
index 76423ab..2cbe1a8 100644
--- a/benchmarks/include/standard_benchmarks.h
+++ b/benchmarks/include/standard_benchmarks.h
@@ -14,6 +14,8 @@
 
 #include "framework/DynamicExtension.h"
 #include "framework/interface/Query.h"
+#include "framework/reconstruction/FixedShardCountPolicy.h"
+#include "framework/reconstruction/ReconstructionPolicy.h"
 #include "query/irs.h"
 #include "psu-util/progress.h"
 #include "benchmark_types.h"
@@ -23,13 +25,14 @@
 #include "framework/reconstruction/TieringPolicy.h"
 #include "framework/reconstruction/BSMPolicy.h"
 #include "framework/reconstruction/FloodL0Policy.h"
+#include "framework/reconstruction/BackgroundTieringPolicy.h"
 
 constexpr double delete_proportion = 0.05;
 static size_t g_deleted_records = 0;
 static size_t total = 0;
 
 template<de::ShardInterface S, de::QueryInterface<S> Q>
-de::ReconstructionPolicy<S, Q> *get_policy(size_t scale_factor, size_t buffer_size, int policy=0) {
+std::unique_ptr<de::ReconstructionPolicy<S, Q>> get_policy(size_t scale_factor, size_t buffer_size, int policy=0, size_t reccnt=0) {
 
     de::ReconstructionPolicy<S, Q> *recon = nullptr;
     
@@ -41,9 +44,14 @@ de::ReconstructionPolicy<S, Q> *get_policy(size_t scale_factor, size_t buffer_si
         recon = new de::BSMPolicy<S, Q>(buffer_size);
     } else if (policy == 3) {
         recon = new de::FloodL0Policy<S, Q>(buffer_size);
+    } else if (policy == 4) {
+        assert(reccnt > 0);
+        recon = new de::FixedShardCountPolicy<S, Q>(buffer_size, scale_factor, reccnt);
+    } else if (policy == 5) {
+        recon = new de::BackgroundTieringPolicy<S, Q>(scale_factor, buffer_size);
     }
 
-    return recon;
+    return std::unique_ptr<de::ReconstructionPolicy<S, Q>>(recon);
 }
 
 template<typename DE, typename Q, bool BSM=false>
diff --git a/benchmarks/tail-latency/isam_insert_dist.cpp b/benchmarks/tail-latency/isam_insert_dist.cpp
new file mode 100644
index 0000000..88d37c5
--- /dev/null
+++ b/benchmarks/tail-latency/isam_insert_dist.cpp
@@ -0,0 +1,107 @@
+/*
+ *
+ */
+
+#include "framework/scheduling/FIFOScheduler.h"
+#define ENABLE_TIMER
+#define TS_TEST
+
+#include <thread>
+
+#include "framework/DynamicExtension.h"
+#include "shard/ISAMTree.h"
+#include "query/rangecount.h"
+#include "framework/interface/Record.h"
+#include "file_util.h"
+#include "standard_benchmarks.h"
+
+#include <gsl/gsl_rng.h>
+
+#include "psu-util/timer.h"
+
+
+typedef de::Record<uint64_t, uint64_t> Rec;
+typedef de::ISAMTree<Rec>Shard;
+typedef de::rc::Query<Shard> Q;
+typedef de::DynamicExtension<Shard, Q, de::DeletePolicy::TOMBSTONE, de::FIFOScheduler> Ext;
+typedef Q::Parameters QP;
+
+void usage(char *progname) {
+    fprintf(stderr, "%s reccnt datafile queryfile\n", progname);
+}
+
+int main(int argc, char **argv) {
+
+    if (argc < 4) {
+        usage(argv[0]);
+        exit(EXIT_FAILURE);
+    }
+
+    size_t n = atol(argv[1]);
+    std::string d_fname = std::string(argv[2]);
+    std::string q_fname = std::string(argv[3]);
+
+    
+    auto data = read_sosd_file<Rec>(d_fname, n);
+    auto queries = read_range_queries<QP>(q_fname, .0001);
+
+    std::vector<int> policies = {3};
+    std::vector<size_t> buffers = {8000, 16000, 32000};
+    std::vector<size_t> sfs = {8};
+
+    for (size_t l=0; l<policies.size(); l++) {
+    for (size_t j=0; j<buffers.size(); j++) {
+    for (size_t k=0; k<sfs.size(); k++) {
+        auto policy = get_policy<Shard, Q>(sfs[k], buffers[j], policies[l]);
+        auto extension = new Ext(policy, buffers[j]/4, buffers[j]);
+
+        /* warmup structure w/ 10% of records */
+        size_t warmup = .1 * n;
+        for (size_t i=0; i<warmup; i++) {
+            while (!extension->insert(data[i])) {
+                usleep(1);
+            }
+        }
+
+        extension->await_next_epoch();
+
+        TIMER_INIT();
+
+        for (size_t i=warmup; i<data.size(); i++) {
+            TIMER_START();
+            while (!extension->insert(data[i])) {
+                usleep(1);
+            }
+            TIMER_STOP();
+
+            //fprintf(stdout, "%ld\t%ld\t%d\t%ld\n", sfs[k], buffers[j], policies[l], TIMER_RESULT());
+        }
+
+        extension->await_next_epoch();
+        
+        /* repeat the queries a bunch of times */
+        for (size_t l=0; l<10; l++) {
+        for (size_t i=0; i<queries.size(); i++) {
+            TIMER_START();
+            auto q = queries[i];
+            auto res = extension->query(std::move(q));
+            res.get();
+            TIMER_STOP();
+
+            fprintf(stdout, "%ld\t%ld\t%d\t%ld\n", sfs[k], buffers[j], policies[l], TIMER_RESULT());
+        }
+        }
+
+
+        QP p = {0, 10000};
+        auto res =extension->query(std::move(p));
+
+        fprintf(stderr, "%ld\n", res.get());
+        extension->await_next_epoch();
+        delete extension;
+    }}}
+
+
+    fflush(stderr);
+}
+
diff --git a/benchmarks/tail-latency/standard_latency_dist.cpp b/benchmarks/tail-latency/standard_latency_dist.cpp
new file mode 100644
index 0000000..2e800fc
--- /dev/null
+++ b/benchmarks/tail-latency/standard_latency_dist.cpp
@@ -0,0 +1,114 @@
+/*
+ *
+ */
+
+#include "framework/scheduling/SerialScheduler.h"
+#include "framework/util/Configuration.h"
+#include "util/types.h"
+#define ENABLE_TIMER
+#define TS_TEST
+
+#include <thread>
+
+#include "framework/DynamicExtension.h"
+#include "framework/scheduling/FIFOScheduler.h"
+#include "shard/TrieSpline.h"
+#include "query/rangecount.h"
+#include "framework/interface/Record.h"
+#include "file_util.h"
+#include "standard_benchmarks.h"
+
+#include "framework/reconstruction/FixedShardCountPolicy.h"
+
+#include <gsl/gsl_rng.h>
+
+#include "psu-util/timer.h"
+
+
+typedef de::Record<uint64_t, uint64_t> Rec;
+typedef de::TrieSpline<Rec> Shard;
+typedef de::rc::Query<Shard> Q;
+typedef de::DynamicExtension<Shard, Q, de::DeletePolicy::TOMBSTONE, de::FIFOScheduler> Ext;
+typedef Q::Parameters QP;
+typedef de::DEConfiguration<Shard, Q, de::DeletePolicy::TOMBSTONE, de::FIFOScheduler> Conf;
+
+void usage(char *progname) {
+    fprintf(stderr, "%s reccnt datafile queryfile\n", progname);
+}
+
+int main(int argc, char **argv) {
+
+    if (argc < 4) {
+        usage(argv[0]);
+        exit(EXIT_FAILURE);
+    }
+
+    size_t n = atol(argv[1]);
+    std::string d_fname = std::string(argv[2]);
+    std::string q_fname = std::string(argv[3]);
+    
+    auto data = read_sosd_file<Rec>(d_fname, n);
+    auto queries = read_range_queries<QP>(q_fname, .0001);
+
+    std::vector<size_t> sfs = {8}; //, 4, 8, 16, 32, 64, 128, 256, 512, 1024}; 
+    size_t buffer_size = 8000;
+    std::vector<size_t> policies = {5,};
+
+    for (auto pol: policies) {
+    for (size_t i=0; i<sfs.size(); i++) {
+        auto policy = get_policy<Shard, Q>(sfs[i], buffer_size, pol, n);
+        auto config = Conf(std::move(policy));
+        config.recon_enable_maint_on_flush = true;
+        config.recon_maint_disabled = false;
+        config.buffer_flush_trigger = 4000;
+        
+        auto extension = new Ext(std::move(config));
+
+        /* warmup structure w/ 10% of records */
+        size_t warmup = .1 * n;
+        for (size_t j=0; j<warmup; j++) {
+            while (!extension->insert(data[j])) {
+                usleep(1);
+            }
+        }
+
+        extension->await_version();
+
+        TIMER_INIT();
+
+        TIMER_START();
+        for (size_t j=warmup; j<data.size(); j++) {
+            while (!extension->insert(data[j])) {
+                usleep(1);
+                fprintf(stderr, "%ld\n", j);
+            }
+        }
+        TIMER_STOP();
+
+        size_t insert_tput = (double) (n - warmup) / (double) (TIMER_RESULT()) * 1e9;
+
+        extension->await_version();
+        
+        size_t total = 0;
+        TIMER_START();
+        /* repeat the queries a bunch of times */
+        for (size_t l=0; l<10; l++) {
+            for (size_t j=0; j<queries.size(); j++) {
+                auto q = queries[j];
+                auto res = extension->query(std::move(q));
+                total += res.get();
+            }
+        }
+        TIMER_STOP();
+
+        size_t query_lat = (double) TIMER_RESULT() / (10*queries.size());
+
+        fprintf(stdout, "S\t%ld\t%ld\t%ld\t%ld\t%ld\t%ld\t%ld\t%ld\n", pol, sfs[i], extension->get_height(), extension->get_shard_count(), extension->get_record_count(), total, insert_tput, query_lat);
+        extension->print_structure();
+        delete extension;
+    }
+    }
+
+    fflush(stderr);
+}
+
-- 
cgit v1.2.3