这是我第一次写博客，由于水平有限，写不出什么好东西，还望见谅。

我现在参与的这个项目采用的是EF框架，方便了数据库的访问。但在实际中，发现项目中导入市县Excel数据耗时太长，于是趁这段时间专门研究了一下大数据的导入。

Excel数据如图：

项目数据库里已经有了省市数据，任务是导入县数据。其中省市县之间有外键关系。项目里已经有了读取Excel表的方法，而且效率不低，耗时主要是在导入到Sql Server2008中

public void ImportCounties(List<County> counties)

        {

            List<City> cities = counties.Select(d => d.City).ToList();

            MyTestEntities db = new MyTestEntities();

            foreach (var city in cities)

            {

                PRovince province =

                    db.Provinces.Where(d => d.Name == city.Province.Name).First();

                city.Province = province;

            }

            foreach (var county in counties)

            {

                db.Counties.AddObject(county);

            }

            db.SaveChanges();

        }

这是项目里的源代码，耗时40秒左右。我测试之后发现不仅效率低，而且是有bug的。测试时第一次录入数据没问题，如果删除Counties表里的数据再导入数据是有问题的，我不知道为什么要为Province赋值而且第一次可以成功，如果有大神看见，请帮我解惑。我只能先尝试以我的方法去做。

public void ImportCountiesSecond(List<County> counties)

        {

            MyTestEntities db = new MyTestEntities();

            List<City> cities = db.Cities.ToList();

            foreach (var county in counties)

            {

                county.City = cities.Where(d => d.Name == county.City.Name).FirstOrDefault();

                db.Counties.AddObject(county);

            }

            db.SaveChanges();

        }

速度大概缩减到20秒，但这也不是我所能忍受的。于是上网又查了资料，了解到SqlBulkCopy。

代码如下：

public void ImportCountiesThird(List<County> counties)

        {

            MyTestEntities db = new MyTestEntities();

            string conStr = ConfigurationManager.ConnectionStrings["dbConnStr"].ConnectionString;

            List<City> cities = db.Cities.ToList();

            DataTable table = new DataTable();

            table.Columns.Add("CityId", typeof(int));

            table.Columns.Add("Name");

            table.Columns.Add("OrderNum", typeof(int));

            table.Columns.Add("IsCity", typeof(bool));

            for (int i = 0; i < counties.Count; i++)

            {

                County line = counties[i];

                int cityId = cities.Where(d => d.Name == counties[i].City.Name).FirstOrDefault().CityId;

                string name = line.Name;

                int orderNum = line.OrderNum;

                bool isCity = line.IsCity;

                DataRow row = table.NewRow();

                row["CityId"] = cityId;

                row["Name"] = name;

                row["OrderNum"] = orderNum;

                row["IsCity"] = isCity;

                table.Rows.Add(row);

            }

            using (SqlBulkCopy bulkCopy = new SqlBulkCopy(conStr))

            {

                bulkCopy.DestinationTableName = "Counties";

                bulkCopy.ColumnMappings.Add("CityId", "CityId");

                bulkCopy.ColumnMappings.Add("Name", "Name");

                bulkCopy.ColumnMappings.Add("OrderNum", "OrderNum");

                bulkCopy.ColumnMappings.Add("IsCity", "IsCity");

                bulkCopy.WriteToServer(table);

            }

        }

时间缩减到4秒左右，其中主要是在EF查询City表并为每一个Counties赋值时耗时占了大半。但这个的优化我觉得暂时做不到了，先这样吧。