分布式部署与性能优化

合集下载
  1. 1、下载文档前请自行甄别文档内容的完整性,平台不提供额外的编辑、内容补充、找答案等附加服务。
  2. 2、"仅部分预览"的文档,不可在线预览部分如存在完整性等问题,可反馈申请退款(可完整预览的文档不适用该条件!)。
  3. 3、如文档侵犯您的权益,请联系客服反馈,我们会尽快为您处理(人工客服工作时间:9:00-18:30)。

ArcGIS Server分布式部署和性能优化
分布式部署
•ArcGIS Server应用部署中总共有这个几个部分:WebServer,SOM,SOC,DataServer
•单层结构
•双层结构
•三层结构
具体部署中注意的问题
•SOM和SOC机器上的用户名和密码相同•Mxd以及数据源所在目录的权限
•输出图片的目录的权限
•发布服务时候的参数要使用UNC目录
性能优化
ArcGIS Server: Response time factors
•Four main factors of response time
•Multiple tiers
•Performance bottleneck can occur in each tier
Browser
Web Server
SOM
SOC
ArcSDE Transmission
Time
Wait Time
Search &
Retrieval Time
Usage Time
SDE
数据方面
•数据量
对数据进行分析,在可能的情况下,对数据进行一些简化操作
•数据组织角度
属性字段索引,空间索引大小的调整
Mxd数据的组织(对动态服务而言)•较少数据图层
•图层按比例控制
•使用annotation来代替label
•减少复杂符号
General Guidelines
•Show relevant information
–Start simple (additional layers can be toggled on by user)
–Use field visibility (hide unnecessary attributes)
•Use scale dependencies
–Use data appropriate for the given scale (generalize if
necessary)
–Display similar number of features at all scales for
consistent user experience
Point, Line & Polygon rendering For non-cached, dynamic layers •Points
–Use single layer Simple or Character markers for best performance
–Use EMF instead of bitmaps
–Use Integer (vs. character) fields for symbol values
–Avoid halos, complex shapes, masking
•Line & Polygons
–Use ESRI_Optimized style
–Avoid cartographic lines (also includes polygon outline!)
ESRI_Optimized
•For lines & polygons
•Outlines for all fills
are simple lines
instead of
cartographic lines
•Picture fills are emf-based instead of bmp-based •Improves drawing performance by > 50%
Best Practices
Text and labeling
•Use annotation instead of labels
•Use indexed fields (reduce label SQL query
number and complexity when possible)
•Use label and feature conflict weights sparingly •Avoid special effects (fill patterns, halos,
callouts, backgrounds)
•Avoid very large text size (60+ pts)
•Avoid Maplex for dynamic labeling
•Avoid overuse
Symbolizing data effectively: Labels / Text •Use larger size
•Use wider line and character spacing
•Use sans serif, open-letter fonts
•Minimize special effects (e.g. italics)•Minimize text backgrounds (e.g. callout boxes)•Don’t overuse Labels
–ID tool can be used
–Use scale dependency
服务的组织
•不要把所有的图层都放在一个服务中,图层很多的情况下可以分放在不同的服务中,这样可以使用多进程来完成所需请求,从而提高相应速度
服务的设置(一)
•池化的服务性能会超过非池化,因此在功能许可
的情况下,尽可能使用池化服务
•在使用编辑等一定要使用非池化的服务的情况下,要尽可能的把非池化的服务中的图层减少,而把
他们增加到池化服务中
服务的设置(二)
•使用MapCache,这是一个非常有效的提高客户端的用户感受的方法
•在创建cache的时候,如果第一次出现错误,可以使用updatecache来进行弥补
瓶颈•CPU •Memory
Cpu 瓶颈:动态服务的池化实例的最优个数•Set instances to level where maximum throughput occurs (usually between 2 to 4 instances per CPU)
CPU bottleneck
number of instances
T h r o u g h p u t (m a p s /h o u r )
N
Memory 瓶颈
Memory Bottleneck –Solutions •Increase physical memory
•Limit capacity on host machines
Setting Capacity
•Limits number of service instances running on a specific host machine.
•Once this limit is reached, Server starts replacing least recently used instances instead of creating new ones.
•Serve large number of services only part of which are used at any point in time
–Usage sliding from European to U.S. map services over time.
–Supports “limited resource” scenario
•Serving a large library of maps
•Individual services rarely used
Building Map Service Cache •Renders 4K x 4K images
–1 GB consumed by just 3 processes •Options to resolve
–Increase memory to 2 or 3 GB
–Reduce number of instances per server
Memory and CPU usage by Service Type
–Dynamic map service
•CPU Usage = High
•Memory = Low*
–Static Map and Globe services
•CPU Usage = Low
•Memory = Low
–Editing map service
•CPU Usage = High
•Memory = High
•Database connections = High
–Geoprocessing service
•CPU = High
•Memory = High
•I/O = High
Detecting Bottlenecks
•Single user testing is inadequate •Simulating multiple users
–Low Tech:
•Recruit others in the office and synchronize over the phone –High Tech: Load simulation tool
Load Simulation Tool
•Build
–Code your own test harness
•Buy
–Price: (free/bundled to $100K or more)
–Technical vs. Functional
–Microsoft
•Web Application Stress Tool (WAST)
•Application Center Test (ACT)
•Visual Studio 2005 Team Edition for Testers
Measuring Web Performance •Client Response Time
–How “fast” does web application appear
to remote browser hitting the site
–Response Time (average, 75th percentile, max)•Machine Throughput
–Measured in requests/sec
–Determines number of servers you need to buy
Visual Studio 2005 Team System
•Tester tools are now on
par with Developer
Tools
•Unit, Web, Load, and
Manual tests
•Measures code
coverage
VS 2005 –Recording a Web Test
• A series of HTTP URLs that can be created or recorded from a browser session.
Monitor Statistics and Log Files •Creation Time
•Wait Time
•Usage Time
Log File Behavior
•Modify Log Level in Server.cfg
–0 (None): No logging
– 1 (Error): Serious problems
– 2 (Warning): Additional errors.
– 3 (Normal): Common administrative messages of the server, warnings, and errors are logged.
– 4 (Detailed): Common messages from user use of the server, including server objects, normal messages, warnings, and errors are
logged.
– 5 (Debug): Verbose messages to aid in troubleshooting
Questions?
High Availability Configuration。

相关文档
最新文档