diff --git a/database.cc b/database.cc
index 558fbe5b9a..6ea2d9ac03 100644
--- a/database.cc
+++ b/database.cc
@@ -162,6 +162,7 @@ database::database(const db::config& cfg, database_config dbcfg)
     , _enable_incremental_backups(cfg.incremental_backups())
     , _querier_cache(_read_concurrency_sem, dbcfg.available_memory * 0.04)
     , _large_partition_handler(std::make_unique<db::cql_table_large_partition_handler>(_cfg->compaction_large_partition_warning_threshold_mb()*1024*1024))
+    , _nop_large_partition_handler(std::make_unique<db::nop_large_partition_handler>())
     , _result_memory_limiter(dbcfg.available_memory / 10)
     , _data_listeners(std::make_unique<db::data_listeners>(*this))
 {
@@ -817,7 +818,14 @@ keyspace::make_column_family_config(const schema& s, const database& db) const {
     cfg.streaming_scheduling_group = _config.streaming_scheduling_group;
     cfg.statement_scheduling_group = _config.statement_scheduling_group;
     cfg.enable_metrics_reporting = db_config.enable_keyspace_column_family_metrics();
-    cfg.large_partition_handler = db.get_large_partition_handler();
+
+    // avoid self-reporting
+    if (s.ks_name() == "system" && s.cf_name() == db::system_keyspace::LARGE_PARTITIONS) {
+        cfg.large_partition_handler = db.get_nop_large_partition_handler();
+    } else {
+        cfg.large_partition_handler = db.get_large_partition_handler();
+    }
+
     cfg.view_update_concurrency_semaphore = _config.view_update_concurrency_semaphore;
     cfg.view_update_concurrency_semaphore_limit = _config.view_update_concurrency_semaphore_limit;
     cfg.data_listeners = &db.data_listeners();
diff --git a/database.hh b/database.hh
index 2c8ae0843b..20b10deaee 100644
--- a/database.hh
+++ b/database.hh
@@ -1272,6 +1272,7 @@ private:
     query::querier_cache _querier_cache;
 
     std::unique_ptr<db::large_partition_handler> _large_partition_handler;
+    std::unique_ptr<db::large_partition_handler> _nop_large_partition_handler;
 
     query::result_memory_limiter _result_memory_limiter;
 
@@ -1423,6 +1424,10 @@ public:
         return _large_partition_handler.get();
     }
 
+    db::large_partition_handler* get_nop_large_partition_handler() const {
+        return _nop_large_partition_handler.get();
+    }
+
     future<> flush_all_memtables();
 
     // See #937. Truncation now requires a callback to get a time stamp
diff --git a/db/large_partition_handler.cc b/db/large_partition_handler.cc
index 255f96e9e0..56c9735135 100644
--- a/db/large_partition_handler.cc
+++ b/db/large_partition_handler.cc
@@ -55,10 +55,6 @@ logging::logger cql_table_large_partition_handler::large_partition_logger("large
 future<> cql_table_large_partition_handler::update_large_partitions(const schema& s, const sstring& sstable_name, const sstables::key& key, uint64_t partition_size) const {
     static const sstring req = format("INSERT INTO system.{} (keyspace_name, table_name, sstable_name, partition_size, partition_key, compaction_time) VALUES (?, ?, ?, ?, ?, ?) USING TTL 2592000",
             db::system_keyspace::LARGE_PARTITIONS);
-    // avoid self-reporting
-    if (s.ks_name() == "system" && s.cf_name() == db::system_keyspace::LARGE_PARTITIONS) {
-        return make_ready_future<>();
-    }
     auto ks_name = s.ks_name();
     auto cf_name = s.cf_name();
     std::ostringstream oss;