[转]SSIS ADO.NET vs OLEDB
ADO.NET vs OLEDB:
I'll try to answer this question by slicing it into 4 main areas:
1. support across SSIS components:
a. SSIS is not welcoming ADO.NET and OLE DB connectors equally across the board. Although this is more so in SQL Server 2005, we are making steps to close the gap in Katmai in between the two. For instance, there is no ADO.NET destination component in Data Flow, and the Lookup component only supports OLE DB. OLE DB is by far the most supported connectivity option across the board. For a full list of supported connectivity options and SSIS components, see my article on the subject: http://ssis.wik.is/Connectivity_Libraries
2. performance
a. Performance is very much dependent on the computing environment and the data and network load levels in your scenario. We highly recommend slicing your end-end scenario into isolated sub scenarios to see where the performance bottleneck is. RowCount component comes particularly handy where you can replace source and destination components with. By replacing the connectors with the RowCount component, which does not incur any I/O and frees memory as an inline component, you are indeed collecting benchmark performance numbers with and without the connector. This will give you the time the connector spends reading/writing the data.
b. That being said, OLE DB, for most cases will have better performance than ADO.NET. This is due to ADO.NET being a managed façade, and providing more abstraction with a little more performance overhead.
3. 64bit considerations
a. Since ADO.NET is managed, the deployment of packages and moving them around 32bit and 64bit boxes will be easier compared to the experience with OLE DB providers. SSIS design time is a 32bit application because of the dependency on Visual Studio, and has to work with 32bit connectors. However, during runtime, you have the option to choose a native 64bit or a 32bit connector. Please read my article on 64bit and why things are the way they are here: http://ssis.wik.is/64-bit_Story
b. Some OLE DB providers are 32bit only (i.e. Office connectivity, JET and ACE providers), whereas others support both architectures (i.e. Oracle’s own OLE DB provider for Oracle). To work in a seamless 64bit environment, make sure your provider has both binaries.
c. For an OLE DB provider to be available only in 32bit (i.e. JET), does not necessarily mean that you cannot run it on 64bit machines. You can, but it will be in WOW64 emulation mode. The only downside to that is if you need the 64bit address space of huge memory (i.e. greater than 4GB). Otherwise, it’s not a problem.
4. target data source & data type supportability
a. Connectors, whether they are ADO.NET or OLE DB providers, don’t support a specific target database/application system equally. For instance, the OLE DB providers for DB2 are all different in the way they support different flavors of DB2. Some support DB2 on Linux, others support DB2 on AS/400, etc…For a good list of connectors and what actually they support, please see the table we are compiling in our connectivity wiki : http://ssis.wik.is/Data_Sources
b. Another key support question is whether these connectors support a specific version of the target database and how good they are in dealing with a recently introduced data type. For instance, Microsoft’s Oracle OLE DB provider was not recently updated to fully support new data types in Oracle 10g +…However, Microsoft’s ADO.NET provider for Oracle does a better job with Oracle data types. Again, our connectivity wiki is a good place to start and contribute to this very organic structure of the connectivity realm.
We also have a white paper on connectivity which covers other aspects of this comparison, it’s a good read: http://ssis.wik.is/Connectivity_White_Paper
Hope this helps,
Deniz
[转]SSIS ADO.NET vs OLEDB的更多相关文章
- 数据库,ADO.NET(ADO),Oledb(Odbc)和编程语言关系框架图
---恢复内容开始--- ---恢复内容结束---
- ODBC, OLEDB, ADO, ADO.NET
在SSIS中,选择Connction Manager时,有很多的类型.其中会发现如下4个连接类型ODBC, OLEDB, ADO, ADO.NET.这4个东东到底是什么,他们有什么关联,什么区别,如何 ...
- ATL模板库中的OLEDB与ADO
上次将OLEDB的所有内容基本上都说完了,从之前的示例上来看OLEDB中有许多变量的定义,什么结果集对象.session对象.命令对象,还有各种缓冲等等,总体上来说直接使用OLEDB写程序很麻烦,用很 ...
- ADO.NET(OleDb)读取Excel表格时的一个BUG
如果我们有例如以下一个Excel表格: 如今要使用C#程序读取其内容: using System; using System.Data.OleDb; namespace Skyiv.Be ...
- ADO和DAO的区别
ADO(ACTIVEX DATA OBJECTS)应用层的数据访问接口ODBC 数据库驱动接口OLE DB 系统级数据访问接口DAO (DATA ACCESS OBJECTS) 对象的数据访问接口AD ...
- OLEDB和ODBC的区别(优缺点)
ODBC是一种连接数据库的开放标准,OLEDB(对象链接和嵌入数据库)位于ODBC层与应用程序之间. 在你的ASP页面里,ADO是位于OLEDB之上的应用程序. 你的ADO调用先被送到OLEDB,然后 ...
- OLEDB简介
OLE DB(OLEDB)是微软的战略性的通向不同的数据源的低级应用程序接口.OLE DB不仅包括微软资助的标准数据接口开放数据库连通性(ODBC)的结构化查询语言(SQL)能力,还具有面向其他非SQ ...
- OLEDB和ODBC的区别
ODBC(开放数据库互连):是Microsoft引进的一种早期数据库接口技术.它实际上是ADO的前身.早期的数据库连接是非常困难的. 每个数据库的格式都不一样,开发者得对他们所开发的每种数据库的底层A ...
- ADO.Net的发展史
1.演变历史: 它们是按照这个时间先后的顺序逐步出现的,史前->ODBC->OLEDB->ADO->ADO.Net. 2.下面分别介绍一下这几个. a. 史前的数据访问是什么样 ...
随机推荐
- jQuery实现,动态自动定位弹窗。JS分页,Ajax请求
工作中碰到一个问题,一个页面中碰到多个地方需要弹窗数据. 网上找了一圈,没有找到合适的,所以自己写了一个. 兼容IE7+,chrome.其它未测试. 需求:点击任意的输入框(也可其它元素,代码中有注释 ...
- 【python】时间戳、字典列表排序
记录一下昨天学到的知识: 一.文件相关 文件追加:f = open("fname","a") 文件不存在时创建 二.时间戳相关 http://www.jb ...
- LeetCode解题报告—— N-Queens && Edit Distance
1. N-Queens The n-queens puzzle is the problem of placing n queens on an n×n chessboard such that no ...
- 机器学习方法(七):Kmeans聚类K值如何选,以及数据重抽样方法Bootstrapping
欢迎转载,转载请注明:本文出自Bin的专栏blog.csdn.net/xbinworld. 技术交流QQ群:433250724,欢迎对算法.技术感兴趣的同学加入.我的博客写一些自己用得到东西,并分享给 ...
- 【JBPM4】完成任务
示例代码: ProcessEngine processEngine = Configuration.getProcessEngine(); TaskService taskService = proc ...
- Linux和Windows双系统安装要点
这里主要说下CentOS和Windows7的双系统安装遇到的一些问题 一.磁盘分区 这个不得不说下在Windows下坑爹的设定了,如果是用默认的磁盘划分,系统会自己先划出一个100M左右的区,关键还是 ...
- CentOS7用yum安装MySQL与启动
首先CentOS7 已经不支持mysql,因为收费了你懂得,所以内部集成了mariadb,而安装mysql的话会和mariadb的文件冲突,所以需要先卸载掉mariadb,以下为卸载mariadb,安 ...
- 在class中获取web资源
背景介绍 项目中用jasperreport做报表,模板文件为web资源,不在classpath之中.class又需要获取模板文件,结合数据源,生成pdf格式的报表. 之前的做法是定义一个public ...
- HZAU 1205 Sequence Number(双指针)
题目链接:http://acm.hzau.edu.cn/problem.php?id=1205 [题意]给你一串数,要求你找到两个数a[i],a[j],使得a[i]<=a[j]且j>=i且 ...
- 【UOJ #131】【NOI 2015】品酒大会
http://uoj.ac/problem/131 求出后缀数组和height数组后,从大到小扫相似度进行合并,每次相当于合并两个紧挨着的区间. 合并区间可以用并查集来实现,每个区间的信息都记录在这个 ...