背景
sqlserver 单表数据过多,数据量已经过亿,查询缓慢,之前是采取手动分表的处理,我是半路接手,现在想对其进行改造,但是项目比较老,无法使用现成的分表插件,我们分表是根据项目来,这里记录下操作过程以及具体代码,希望能给小伙伴提供一些思路
//这里在配置文件中设置分表的阈值
///
/// 根据原表名创建新表
///
///
///
public static string CreateSubmeterTable(string sourceTableName)
{
string newTableName = "";
try
{
string DateStr = DateTime.Now.ToString("yyyyMMddHHmmss");
string connectionString =Settings.DBHerperSqlConnection; // 替换为你的数据库连接字符串
newTableName = sourceTableName+"_" + DateStr; // 替换为你的新表名称
using (SqlConnection connection = new SqlConnection(connectionString))
{
connection.Open();
// 复制表结构
string createTableQuery = "SELECT TOP 0 * INTO "+newTableName+" FROM "+sourceTableName+" ";
using (SqlCommand command = new SqlCommand(createTableQuery, connection))
{
command.ExecuteNonQuery();
}
// 复制主键,这里通过语句只能获取到主键名称,和主键字段不一致,所以这里自己写死,且主键名是数据库唯一,所以自行定义
string pkkey = "PK_" + newTableName;
string pkvalue = "主键字段id"
string alterTableQuery = "ALTER TABLE " + newTableName + " ADD CONSTRAINT " + pkkey + " PRIMARY KEY CLUSTERED (" + pkvalue + ")";
using (SqlCommand alterTableCommand = new SqlCommand(alterTableQuery, connection))
{
alterTableCommand.ExecuteNonQuery();
}
// 复制索引
string indexQuery = "SELECT name, type_desc, is_unique, is_primary_key,index_id FROM sys.indexes WHERE OBJECT_NAME(object_id) = '" + sourceTableName + "'";
List modelindex = new List();
using (SqlCommand command = new SqlCommand(indexQuery, connection))
{
using (SqlDataReader reader = command.ExecuteReader())
{
while (reader.Read()) {
bool isPrimaryKey = reader.GetBoolean(3);
if (!isPrimaryKey)
{
ModelIndex entity = new ModelIndex();
entity.indexName = reader.GetString(0) + "_" + DateStr;
entity.indexType = reader.GetString(1);
entity.isUnique = reader.GetBoolean(2);
entity.index_id = reader.GetInt32(4);
modelindex.Add(entity);
}
}
}
}
//复制索引字段
foreach (var item in modelindex)
{
string createIndexQuery = "CREATE " + (item.isUnique ? "UNIQUE " : "") + item.indexType + " INDEX " + item.indexName + " ON " + newTableName + " (";
// 获取索引关联的字段
string indexColumnsQuery = "SELECT name FROM sys.index_columns ic JOIN sys.columns c ON ic.column_id = c.column_id AND ic.object_id = c.object_id WHERE ic.object_id = OBJECT_ID('" + sourceTableName + "') AND ic.index_id = " + item.index_id;
using (SqlCommand indexColumnsCommand = new SqlCommand(indexColumnsQuery, connection))
{
using (SqlDataReader indexColumnsReader = indexColumnsCommand.ExecuteReader())
{
bool firstColumn = true;
while (indexColumnsReader.Read())
{
if (!firstColumn)
{
createIndexQuery += ", ";
}
createIndexQuery += indexColumnsReader.GetString(0);
firstColumn = false;
}
}
}
createIndexQuery += ")";
using (SqlCommand createIndexCommand = new SqlCommand(createIndexQuery, connection))
{
createIndexCommand.ExecuteNonQuery();
}
}
}
}
catch (Exception ex)
{
LogHelper.WriteLog("分表失败:" + ex.Message);
}
return newTableName;
}
思路提供:
每次创建项目,都获取下最新分表的数据总量,如果记录数超过设置的阈值,则进行分表处理,且将分表的表名,项目id进行记录
目的是下次创建项目的时候,根据项目id以及存储的分表表名,查询新表数据是否超出阈值
手动分表的缺点就是需要重写插入、查询方法,因为没有引用orm,这里 插入数据需要进行改造
//部分代码如下
db.CommandTimeout = 600000;
// 打开数据库连接
db.Connection.Open();
//开启事务
var transaction = db.Connection.BeginTransaction();
db.Transaction = transaction;
try
{
//获取字段属性
var tabledata = SubmeterHerper.GetEntityData(存储数据的实体, true, 1);
string answerSql = string.Format(@"INSERT INTO {0} ({1}) OUTPUT INSERTED.answerid VALUES ({2})", "分表表名", tabledata.Item1, tabledata.Item2);
var result = db.ExecuteQuery(answerSql).FirstOrDefault();
db.Transaction.Commit();
}
catch (Exception ex)
{
// 发生异常时回滚事务
db.Transaction.Rollback();
throw;
}
finally
{
// 关闭数据库连接
db.Connection.Close();
}
///
/// 获取字段属性
///
///
/// 值
/// 是否存在关联表
/// 如果存在关联,末尾跳过的数量
///
public static Tuple GetEntityData(T entity, bool association = false, int takenum = 1)
{
// 用于拼接字段名
StringBuilder fieldNameBuilder = new StringBuilder();
// 用于拼接字段值
StringBuilder fieldValueBuilder = new StringBuilder();
// 获取entity的类型
Type entityType = entity.GetType();
// 获取entity的所有属性
PropertyInfo[] properties = entityType.GetProperties();
//判断是否存在关联表,如果存在需要跳过关联字段
if (association)
{
properties = properties.Skip(1).Take(properties.Length - (1 + takenum)).ToArray();
}
else
{
properties = properties.Skip(1).ToArray();
}
// 遍历属性
foreach (PropertyInfo property in properties)
{
//主键自增无法通过反射获取,这里去除第一个字段(--默认第一个字段为主键)
// 获取属性名
string fieldName = property.Name;
// 获取属性值
object fieldValue = property.GetValue(entity, null);
// 将属性名拼接到字段名的字符串中
fieldNameBuilder.Append("[").Append(fieldName).Append("],");
// 将属性值拼接到字段值的字符串中
fieldValueBuilder.Append("'").Append(fieldValue).Append("',");
}
// 去除最后一个逗号
fieldNameBuilder.Length--;
fieldValueBuilder.Length--;
// 返回拼接的结果
string fieldNameString = fieldNameBuilder.ToString();
string fieldValueString = fieldValueBuilder.ToString();
return new Tuple(fieldNameString, fieldValueString);
}
///
/// 获取字段属性--被关联专用
///
///
/// 值
/// 被关联字段
/// 需赋的值
///
public static Tuple GetEntityDataGL(T entity,string colname,int value)
{
// 用于拼接字段名
StringBuilder fieldNameBuilder = new StringBuilder();
// 用于拼接字段值
StringBuilder fieldValueBuilder = new StringBuilder();
// 获取entity的类型
Type entityType = entity.GetType();
// 获取entity的所有属性
PropertyInfo[] properties = entityType.GetProperties();
//这因为是关联表,所以去掉首尾,如果存在多个关联,则下面应该是properties.Length -(1+n)
properties = properties.Skip(1).Take(properties.Length - (2)).ToArray();
// 遍历属性
foreach (PropertyInfo property in properties)
{
//主键自增无法通过反射获取,这里去除第一个字段(--默认第一个字段为主键)
// 获取属性名
string fieldName = property.Name;
// 获取属性值
object fieldValue = property.GetValue(entity, null);
if (fieldName==colname)
{
fieldValue = value;
}
// 将属性名拼接到字段名的字符串中
fieldNameBuilder.Append("[").Append(fieldName).Append("],");
// 将属性值拼接到字段值的字符串中
fieldValueBuilder.Append("'").Append(fieldValue).Append("',");
}
// 去除最后一个逗号
fieldNameBuilder.Length--;
fieldValueBuilder.Length--;
// 返回拼接的结果
string fieldNameString = fieldNameBuilder.ToString();
string fieldValueString = fieldValueBuilder.ToString();
return new Tuple(fieldNameString, fieldValueString);
}
}