ASP. NET performance best practices

Sort by msdn webcast

Logical/physical design practices

Recommendation: Use a logical three-tier model

1. Page (. aspx) and user control (. ASCs)
2. Reusable types (components) are placed in the \ bin directory
3. Store data in SQL database

Recommendation: designed for Web farm

1. Do not assume that the visitor will always return to the same server
2. Pay attention to static variables and application status
3. Web farm session Status

Recommendation: use the same process

Avoid (when possible ):

1. Synchronous calls to XML Web Services
2. Remote Call through DCOM

Use XML Web Services:

1. Communication between applications on the Internet
2. Do not use for intra-Application Communication

Recommendation: use an ISA Server. When a DMZ security domain is required:

1. Only access through ISA is allowed
2. ISA tunnel reaches ASP. NET through DMZ
3. Reduce process Conversion

Language/CodeSuggestions

1. Avoid late binding
1. <% @ Page Language = "VB" Explicit = "true" %> declare all variable types and allow late binding
2. <% @ Page Language = "VB" strict = "true" %> Any late binding is prohibited.

Com interaction suggestions

1. Use managed components instead of COM

Data recommendations

1. Store data in SQL Server
2. Select an appropriate data access method: oledb/SQL client/ODBC
3. Properly Select dataset and datareader for data access

Status Management

1. Select an appropriate method to store sessions: inproc/StateServer/Database
2. Use simple data types
3. Avoid saving the Stas COM object in the session.
4. If you do not need a session, disable it: enablesessionstate = "false"
5. If possible, you can use the readonly feature: enablesessionstate = "readonly"
6. If viewstate is not required, disable it. enableviewstate = false
7. Minimize data in viewstate
8. View viewstate size view HTML output source code or enable trace

Use Cache

1. Page Cache
2. Partial page Cache
3. Cache object
4. Data Cache
5. WebService Cache

How to quantify web performance:
Machine Throughput
Response time
Test by adding load to the server

Load with multiple client machines

Free Microsoft Web application Stress

Test multiple scenarios

1. Simulate end-to-end scenario traversal of the site
2. Test single-page performance

Value to be calculated

1. Request/sex under different loads (100,250,500,750,100 0 and other clients simultaneously accessing)
2. Identify the maximum customer load within the acceptable response time of ttfb/TTLB

Key performance counters

CPU, CPU % usage | low value = overcast or locked

ASP. NET, number of incoming column requests | linear growth means the server is fully loaded

ASP. NET applications, requests per second | dynamic throughput (Consistent)

ASP. NET application, total error count | indicates a function error (it should be 0)

ASP. NET application, Working Process restart | indicates a serious function Error