feat: add a new ratio config instead of table_data_deserialized_data_bytes

src/query/config/src/config.rs

@@ -2803,18 +2803,18 @@ pub struct CacheConfig {
     #[serde(rename = "disk")]
     pub disk_cache_config: DiskCacheConfig,
 
-    /// Max size of in memory table column object cache. By default it is 0 (disabled)
+    /// Max percentage of in memory table column object cache relative to whole memory. By default it is 0 (disabled)
     ///
-    /// CAUTION: The cached items are deserialized table column objects, may take a lot of memory.
+    /// CAUTION: The cached items are deserialized table column objects, ma take a lot of memory.
     ///
     /// Only if query nodes have plenty of un-utilized memory, the working set can be fitted into,
     /// and the access pattern will benefit from caching, consider enabled this cache.
     #[clap(
-        long = "cache-table-data-deserialized-data-bytes",
+        long = "cache-table-data-deserialized-memory-ratio",
         value_name = "VALUE",
         default_value = "0"
     )]
-    pub table_data_deserialized_data_bytes: u64,
+    pub table_data_deserialized_memory_ratio: u64,
 
     // ----- the following options/args are all deprecated               ----
     /// Max number of cached table segment
@@ -2946,7 +2946,7 @@ mod cache_config_converters {
                 table_data_cache_population_queue_size: value
                     .table_data_cache_population_queue_size,
                 disk_cache_config: value.disk_cache_config.try_into()?,
-                table_data_deserialized_data_bytes: value.table_data_deserialized_data_bytes,
+                table_data_deserialized_memory_ratio: value.table_data_deserialized_memory_ratio,
             })
         }
     }
@@ -2967,7 +2967,7 @@ mod cache_config_converters {
                 table_data_cache_population_queue_size: value
                     .table_data_cache_population_queue_size,
                 disk_cache_config: value.disk_cache_config.into(),
-                table_data_deserialized_data_bytes: value.table_data_deserialized_data_bytes,
+                table_data_deserialized_memory_ratio: value.table_data_deserialized_memory_ratio,
                 table_meta_segment_count: None,
             }
         }

src/query/config/src/inner.rs

@@ -557,13 +557,13 @@ pub struct CacheConfig {
     /// Storage that hold the raw data caches
     pub disk_cache_config: DiskCacheConfig,
 
-    /// Max size of in memory table column object cache. By default it is 0 (disabled)
+    /// Max percentage of in memory table column object cache relative to whole memory. By default it is 0 (disabled)
     ///
-    /// CAUTION: The cache items are deserialized table column objects, may take a lot of memory.
+    /// CAUTION: The cached items are deserialized table column objects, ma take a lot of memory.
     ///
     /// Only if query nodes have plenty of un-utilized memory, the working set can be fitted into,
     /// and the access pattern will benefit from caching, consider enabled this cache.
-    pub table_data_deserialized_data_bytes: u64,
+    pub table_data_deserialized_memory_ratio: u64,
 }
 
 #[derive(Clone, Debug, PartialEq, Eq)]
@@ -621,7 +621,7 @@ impl Default for CacheConfig {
             data_cache_storage: Default::default(),
             table_data_cache_population_queue_size: 0,
             disk_cache_config: Default::default(),
-            table_data_deserialized_data_bytes: 0,
+            table_data_deserialized_memory_ratio: 0,
         }
     }
 }

src/query/service/src/global_services.rs

@@ -116,7 +116,11 @@ impl GlobalServices {
             &config.query.share_endpoint_auth_token_file,
             config.query.tenant_id.to_string(),
         )?;
-        CacheManager::init(&config.cache, config.query.tenant_id.to_string())?;
+        CacheManager::init(
+            &config.cache,
+            &config.query.max_server_memory_usage,
+            config.query.tenant_id.to_string(),
+        )?;
 
         if let Some(addr) = config.query.cloud_control_grpc_server_address.clone() {
             CloudControlApiProvider::init(addr, config.query.cloud_control_grpc_timeout).await?;

src/query/service/tests/it/storages/fuse/io.rs

@@ -48,7 +48,7 @@ async fn test_array_cache_of_nested_column_iusse_14502() -> Result<()> {
 
     let mut config = InnerConfig::default();
     // memory cache is not enabled by default, let's enable it
-    config.cache.table_data_deserialized_data_bytes = 1024 * 1024 * 10;
+    config.cache.table_data_deserialized_memory_ratio = 10;
     let fixture = TestFixture::setup_with_config(&config).await?;
 
     fixture.create_default_database().await?;

src/query/service/tests/it/storages/testdata/configs_table_basic.txt

@@ -13,7 +13,7 @@ DB.Table: 'system'.'configs', Table: configs-table_id:1, ver:0, Engine: SystemCo
 | 'cache'   | 'table_bloom_index_filter_size'            | '2147483648'                                                   | ''       |
 | 'cache'   | 'table_bloom_index_meta_count'             | '3000'                                                         | ''       |
 | 'cache'   | 'table_data_cache_population_queue_size'   | '0'                                                            | ''       |
-| 'cache'   | 'table_data_deserialized_data_bytes'       | '0'                                                            | ''       |
+| 'cache'   | 'table_data_deserialized_memory_ratio'     | '0'                                                            | ''       |
 | 'cache'   | 'table_meta_segment_bytes'                 | '1073741824'                                                   | ''       |
 | 'cache'   | 'table_meta_segment_count'                 | 'null'                                                         | ''       |
 | 'cache'   | 'table_meta_snapshot_count'                | '256'                                                          | ''       |

src/query/storages/common/cache_manager/src/cache_manager.rs

@@ -58,7 +58,11 @@ pub struct CacheManager {
 
 impl CacheManager {
     /// Initialize the caches according to the relevant configurations.
-    pub fn init(config: &CacheConfig, tenant_id: impl Into<String>) -> Result<()> {
+    pub fn init(
+        config: &CacheConfig,
+        max_server_memory_usage: &u64,
+        tenant_id: impl Into<String>,
+    ) -> Result<()> {
         // setup table data cache
         let table_data_cache = {
             match config.data_cache_storage {
@@ -94,8 +98,10 @@ impl CacheManager {
         };
 
         // setup in-memory table column cache
+        let memory_cache_capacity =
+            max_server_memory_usage * config.table_data_deserialized_memory_ratio / 100;
         let table_column_array_cache = Self::new_in_memory_cache(
-            config.table_data_deserialized_data_bytes,
+            memory_cache_capacity,
             ColumnArrayMeter,
             "table_data_column_array",
         );