3. Agenda: System Architecture Desing
• System Design Process
• GIS Technology
• Data Administration
• Network Communications
• Esri Architecture
• Performance
- Software
- Hardware
• Virtualization
• Example of CPT
4. System Architecture Design
User
needs
Infrastructure specifications
Data resources
Network
design
Hardware
People
(information products)
Applications
5. Plannning
Identify business requirements
User Workflows / Services
Peak system design loads
Data center hardware specifications
Network infrastructure needs
Define project requirements
Project budget
Project schedule
System performance needs
Return on investment/funding
Reduce Implementation Risk
Identify key performance parameters
Establish system performance targets
Define system performance validation plan
6. Why is System Architecture Design Important?
Reduce cost Improve productivity
Balanced system design
Servers Network DBMS Geodatabase
Storage Clients ArcSDE Application
Hardware infrastructure Database design User workflow
System architecture design
Framework for productive operations
7. Stages Support a successful implementation
Application
development
Prototype
test
Data
Construction
needs
Design
System
design Implementation
Requirements System
delivery
Building a GIS
User needs is an Incremental Process
8. Capacity Planning Tool: System Design
• Design tab
User workflow requirements Network Display
suitability performance
Platform Selection Platform sizing
and utilization
9. Agenda: System Architecture Desing
• System Design Process
• GIS Technology
• Data Administration
• Network Communications
• Esri Architecture
• Performance
- Software
- Hardware
• Virtualization
• Example of CPT
10. GIS Software Evolution
What are the lessons learned?
Platform Performance Change
>> 10x Faster Hardware >> 10x Faster Hardware >> 10x Faster Hardware
> 2x Process Load > 2x Process Load
>> 10x Faster Technology Change >> 10x Faster Technology Change
Service Oriented Architecture
Tightly Scripted Software Code Object Relational Software Hardware Virtualization ArcGIS Server AMI
1982–1999
Cloud Computing GIS iPhone Apps
ArcGIS Desktop ArcGIS 2008 -2009 ArcGIS 2010
ARC/INFO 1982 - ArcInfo 1999 - REST API -ArcGIS.com
- ArcEditor 2000 - WEB 2.0 Interoperability -Online Community Basemap
ArcView GIS 1992 - ArcView - User Publishing / Collaboration -Intelligent Maps, Web Mapping
MapObjects 1996 - ArcGIS Server 2004 - ArcGIS Online Resources -Packaging toolset (layers, maps)
- Rich Internet Clients -Mosaic Dataset, Image Processing
ArcIMS 1997 - ArcGIS Engine
- Javascript, Flex API, Silverlight
Javascript, -Crowd Computing
Coverages ArcSDE 1999 ArcGIS Server 2009 New search tools
Librarian - Geodatabase Versioning -Faster Maps New parcel fabric
Shapefiles (ArcSDE) -Map Cache Temporal Analysis
- Distributed Geodatabase -Map Service Description (MSD)
ArcStorm -Optimized Mapping tools
Replication
Spatial Database Engine (SDE)
- Simple non-versioned file types
non-
Scripts Objects Services
11. GIS Is Deployed in Many Ways
Departmental Enterprise
Traditional
Emerging
12. Esri Product Family
Desktop GIS
ArcInfo
ArcGIS Extensions
ArcEditor
Mobile GIS ArcView
GIS Web Services
ArcPad
ArcReader ArcGIS.com
ArcGIS Mobile Online Resource Centers
ArcGIS Engine Templates, Forums, Collaboration
Global Map Cache, Community Basemap,
3D Viewers Intelligent maps, Web mapping
Google Virtual Earth
ArcGIS
Explorer
Lightweight Network Developer GIS
Viewers ESRI Developer Network
(EDN)
Web Browsers
XML/SOAP/OGC REST (Javascript, FLASH, Silverlight)
Server GIS ArcGIS Server ArcSDE Image Server ArcIMS
Blending into a single server product line
GIS Data
Files Geodatabase Imagery Map Cache
Distributed Geodatabase Management
13. Desktop Operations
Stand alone Desktop Connected Desktop Centralized Desktop Management
ICA RDP
ArcGIS Desktop ArcGIS Desktop
Direct Connect*
SDE Citrix Microsoft
Personal ArcSDE DBMS Client Windows Terminal Server
ArcGIS Desktop
File GDB Map Cache Direct Connect*
SDE
ArcGIS DBMS Client
Server
ArcGIS.com
World Basemaps
Map Cache
Community Basemaps Geodatabase
Files Imagery Map Cache
Web Maps and Services
Physical or Virtual Servers
14. Esri Mobile Operations
Stand alone Desktop
ArcGIS Mobile
ArcGIS Desktop
ArcPad (smart clients)
Protected Data Cache
Personal ArcSDE 4 or 10 GB SDE GDB
Data
Synchronization
Exchange
File GDB Map Cache
1 TB
Geodatabase
iPhone
Replication
application
Web Server Web Services
ArcGIS.com ArcGIS Server
World Basemaps
Map Cache
Community Basemaps
Web Maps and Services
Files Geodatabase Imagery Map Cache
Physical or Virtual Servers
15. ArcGIS Server Web Operations
Google Microsoft HTML Browser
ArcGIS Desktop ArcGIS Explorer ADF Client
3D Analyst Online
ArcGIS Desktop ArcGIS Explorer Web Maps HTML Browser
ArcGIS Engine JavaScript
2D and 3D Services
Data Services Rich Browser Clients
Geoprocessing Services Flash / Silverlight
Amazon Cloud
ArcGIS iPhone / Android (soon)
Server
Enterprise Data Center
ArcGIS.com .NET/JAVA WAS SOAP REST
Open
App Dev Framework Standards
ArcGIS Server
World Basemaps
Map Cache
Community Basemaps
Web Maps and Services
Files Geodatabase Imagery Map Cache
Physical or Virtual Servers
16. Software Technology Selection
GIS Data Source
Remote Data Sources
Amazon Cloud Internal Data Sources
ArcGIS
Server
Files Geodatabase Imagery Map Cache
Internet Service
Physical or Virtual Servers
WTS
(Citrix) Network Services
LAN Web Server Published Network Services
ArcIMS ArcGIS Distributed Batch Processing
Server
Web Services
Published Map Products
Workstations Published Geoprocessing Services
Browser WAN
Web Editing / Crowd Computing
Desktop Applications Integrated Business Workflows
Professional GIS Users
Database Maintenance
Heavy Business Workflows
GIS Project Efforts
Data Conversion
Terminals Browser
Mobile Applications
Disconnected Operations
Loosely Connected Workflows
17. Agenda: System Architecture Desing
• System Design Process
• GIS Technology
• Data Administration
• Network Communications
• Esri Architecture
• Performance
- Software
- Hardware
• Virtualization
• Example of CPT
18. Geodatabase
• Geographic Data Store
- Scalable, multi-user
• Core ArcGIS Data model Networks
Survey
Addresses
Vectors Annotations
• Transactional model for 3D Objects Attributes
GIS workflows
Dimensions
Topology
• COM Components for Terrain
Parcels
data accesibility
Cartography CAD
Raster
19. Data management approach
• Geodatabase built on a table structure
- Spatial Data Types & SQL
- Relational Integrity
- Scalability, Reliability, Flexibility
- Support for continous, large datasets
• Based on the simple feature model
- Points, Lines, Polygons
- Open Access
20. Complex behavior in the Geodatabase
• Validation Rules
- Attribute, connectivity and
relationships
• Domains
- Ranged or coded values
• Subtypes
- Defined as the class level
• Relationship classes
- Association between objects
on different classes
21. Complex behavior in the Geodatabase
• Geometric Networks
- Connectivity rules
• Network datasets
- Multi-modal networks
- Turns, restrictions, costs
• Topology
- Set of rules that define spatial behavior
- Ensure data integrity
• Parcel fabrics
22. Versioning
• Technology that allows multiple users to edit and
view data at the same time
- Appears to users as if they have their own copy of a
table
- Does not apply locks or duplicate data
Default
Readers
Version1 Version2
Editor 1 Editor 2
23. Replicas
ArcSDE GDB
• Geodatabase replication extends DFLT
the concept of versioning to
Parent VR1 VR2
multiple databases. Child
check out
check in
DFLT
ArcSDE GDB
V1 V2 File GDB
DFLT
Parent
Child ArcSDE GDB
VR1 VR2
Two
One
way
way - Versions in parent and child geodatabases are
used as replica versions.
File GDB
DFLT
DFLT
- Changes in this replica versions are transferred
during synchronization
V1 V2
ArcSDE GDB
ArcSDE GDB
24. Replication types
• Checkout/Check-in replication: One time
- Edits in child replica can be syncronized one single time Parent Child
with parent replica.
SDE
- Uses: Disconected editing; File GDB as a requisite. SDE PGDB
FGDB
• One-way Replication:
- Allows data changes to be sent multiple times in a single Development Production
direction (parent-to-child or child-to-parent)
- Data in origin GDB are editable; data in destination GDB are Persist
read-only. Edit Read
- Changes in destination GDB are overwritten during SDE (child-to-parent)
syncronization if edits in origin GDB exist. SDE PGDB
- Uses: development-staging-production models; data FGDB
publishing
• Two-way replication: Office A Office B
- Allows data changes to be sent multiple times, both ways. Persist
- If the same row is edited in both replica geodatabases, it is Edit A Edit B
detected as a conflict when the replicas are syncronized.
SDE SDE
Conflict
resolution
25. Historical versions
• Geodatabase archiving.
- Built on Geodatabase versions.
- Enables recording and accessing changes made to a subset of
data.
- No need for storing snapshots for the entire database.
- Archiving lives as long as the Version is alive.
- Introduces a historical version.
- Creation of historical markers possible.
26. Geodatabase compress
• Maintain database performance and health.
- Delete all states that do not participate within a version's
lineage.
- Collapse any candidate lineage of states into one state.
- Move rows from the delta tables into the business tables.
Before After
0 0
v2
v2
v1 DEFAULT
v1
v4
v4 DEFAULT
27. Geodatabase administration in 10.1
• New Geodatabase administration dialog box
- Manage versions
- View and remove connections
- View and manage locks
- Block new connections
• For system tables:
- Rebuild indexes
- Update statistics
28. Managing data in ArcGIS 10.1
• View and manage privileges
• View locks on a version
or specific dataset
• Rebuild indexes on multiple datasets
• Update database statistics
on multiple datasets
• Enable editor tracking
• One-click spatial table registering
29. Enterprise Geodata Management
Providing options for distributing geodata management activities
Geodatabase Versions
•Direct editing
•Basis for replication Production Publication
Geodatabase Geodatabase
Connected Replicas
•Workgroups
•Two-way exchange
One-Way
Replica
Disconnected Replicas
•Ideal for outsourcing work
•XML-based transfer
•Two-way exchange
•Read-Write Access •Read-Only Access
•Tuned for Editing •Tuned for Services
Check Out Replicas •Authoritative Content •Approved Content
•File-based GDB
•Inhouse or Outsourced
•One-time exchange
31. Agenda: System Architecture Desing
• System Design Process
• GIS Technology
• Data Administration
• Network Communications
• Esri Architecture
• Performance
- Software
- Hardware
• Virtualization
• Example of CPT
32. GIS Network Impact
What GIS Does ...
- Graphic Data Representation (Maps)
- Large Quantity of Data Analysis
- Lots of Network Traffic
Dagwood Sandwich
Network Communications
Distributed Clients Central Data Repository
34. Network Latency can make a difference
Network latency is minimum communication packet
travel time between client and server platform
Rule of thumb:
GIS application with 20 layers
makes 200 round trips to the
ArcSDE Server.
35. Network Latency
Example: 200 trips to server for single map display
(i.e. 10 sequential round trips for each map layer – 40 display layers)
Local User
Local Network (LAN)
Latency Transport Time
CPU Time 0.001 sec (10 Mb / 10 Mbps)
CPU Time
200 trips DBMS
0.66 sec 0.2 sec 1.0 sec 0.05 sec
GIS User
2.00 sec per display, maximum 5.00 Mbps traffic (10 Mb/2.00 sec)
Remote User 500 miles (WAN)
Latency Transport Time
CPU Time 0.03 sec (10 Mb / 10 Mbps) DBMS CPU Time
0.66 sec 200 trips 0.05 sec
6.0 sec 1.0 sec
GIS User
7.80 sec per display, maximum 1.28 Mbps traffic (10 Mb/7.80 sec)
(4 times slower than local environment)
Latency reduces peak single user network utilization.
Note: Message protocols (minimum chatter) perform best over high latency.
37. Network Traffic Adjustments
Data source adjustments
SDE and File GDB about the same Shapefile traffic can be significant
38. Network Suitability Analysis
1. Identify user workflows
Identify average display traffic (Mbpd) Utilization
100% 60%
2. Identify user locations
80%
3. Compute workflow display traffic 60%
Traffic = productivity (DPM) x Mbpd / 60 seconds
40%
4. Compute site traffic 20%
Site Traffic = Sum (workflow traffic for that site) 0%
Traffic
5. Identify Site Bandwidth
6. Compute network utilization
Utilization = traffic / bandwidth
39. Network Suitability Analysis
7. Calculate Workflow response times Response time
Calculate network queue times Transport time Queue time
Calculate network transport times
Network response time = transport + queue times + latency
Calculate workflow display response times (Platform + Network response
times)
Validate user productivity Workflow Performance Summary
1.6
Client
(Adjust productivity as needed) 1.4 Latency
NWQ
1.2
Network
Performance (sec)
1.0 WTSQ
Response time Cycle time
Computed Think time WTS
0.8
WebQ
Margin Minimum 0.6 Web
think time MapQ
0.4
Map
0.2 DBMSQ
0.0 DBMS
8. Show Workflow Performance Summary
40. Application Types and Network Load
•Rich Client applications – Heavy Load (10 Mb per display)
•RCA via Citrix(image) – Medium (5 Mb)
•RCA via Citrix(Vector) – Light (3 Mb)
•Web Apps (Dynamic) – Medium (2 Mb)
•Web Apps (Cached) – Medium (1 Mb)
•Services (REST) – Medium (2 Mb)
•Services (SOAP) – Medium (3 Mb)
•Mobile Apps – Light (0.05 Mb)
41. Agenda: System Architecture Desing
• System Design Process
• GIS Technology
• Data Administration
• Network Communications
• Esri Architecture
• Performance
- Software
- Hardware
• Virtualization
• Example of CPT
42. ArcGIS System
------- ArcGIS Desktop ------- ArcGIS Web ArcGIS
Terminals ArcInfo ArcEditor ArcView
--------- ArcGIS Engine --------- Explorer Browser Mobile
Terminal Services Web Services
ArcGIS Desktop ArcGIS Server
Web GIS
ArcGIS Online Files Geodatabase Imagery Map Cache
43. Single-tier Platform Configuration
Standard Configuration High-Availability Configuration
Network Load Balancing
GIS Web Server GIS Web Server GIS Web Server
WAS WAS1 WAS2
SOM
SOM1 SOM1 Load Balancing SOM2
SOC1 SOC1 SOC1 SOC2 SOC1 SOC2
SDE DC SDE DC SDE DC SDE DC SDE DC SDE DC
DBMS Client DBMS Client DBMS Client
DBMS DBMS1 DBMS2
DATA DATA DATA
Workgroup Server with SQL Express Enterprise Server with SQL Express
or other Supported DBMS Platforms
44. ArcGIS Server Image Extension. Single-tier
Traditional Configuration High-Availability Configuration
Single Server Multiple Servers
Network Load Balancing
AGS Image Service AGS Image Service AGS Image Service
AGS SOM AGS SOM AGS SOM
AGS SOC
AGS SOC AGS SOC
Mosaic Data Set Mosaic Data Set Mosaic Data Set
Imagery Imagery Imagery
AGS Image Extension AGS Image Extension AGS Image Extension
Data Source Data Source Data Source
45. Two-tier Platform Configuration
Standard Configuration High-Availability Configuration
GIS Web Server
Network Load Balancing
WAS
SOM1 GIS Web Server GIS Web Server
WAS1 SOM WAS2
SOC1 SOC1
SOM1 Load Balancing SOM2
SDE DC SDE DC
DBMS Client SOC1 SOC2 SOC1 SOC2
SDE DC SDE DC SDE DC SDE DC
DBMS Client DBMS Client
Cluster Address
DBMS
DATA
DBMS DBMS
Geodatabase Server DATA
Primary Data Server Failover Data Server
Common Storage Array
Enterprise Server with Supported Database
46. Three-tier Platform Configurations
Standard Configuration High-Availability Configuration
Network Load Balancing
GIS Web Server
GIS Web Server GIS Web Server
WAS1
WAS1 WAS2
SOM1
SOM1 SOM2
SOM Load Balancing SOM Load Balancing
SOC1 SOC1 SOC1 SOC1 SOC2 SOC1 SOC2 SOC1 SOC2
SDE DC SDE DC SDE DC SDE DC SDE DC SDE DC SDE DC SDE DC SDE DC
DBMS Client DBMS Client DBMS Client DBMS Client DBMS Client DBMS Client
Container Machine Container Machine Container Machine Container Machine Container Machine Container Machine
Enterprise Database Server DS Enterprise Database Server DS
48. Votos y Cifras
•Elections & data Web site
-Elections Results
-Elections historic results
-Population Data (economic + demographic)
•Cloud Amazon Server
•Esri technology
-ArcGIS Server workstation
-JavaScript API
-Tiled Maps
50. votos y cifras: Results
•331.065 Esri España brand impacts
-Direct presence in 18 regional newspapers website (Heraldo,
Provincias, Verdad, Ideal,….)
-More than 10.000 social networks impacts (Twitter, Facebook,
eskup, Nolesvotes,…
-Daily record: 107.920 visits the 22-M.
•An average of 4 minutes remaining in the application
•Direct presence in the website of 4 national media (El
Mundo, ABC, Publico y Telecinco)
•321.085 different visitors to our website from 29-04 to 23-
05
51.
52. Architecture Description
Instance m1.xlarge
Extra Large Instance JavaScript Web App
15 GB memory
8 EC2 Compute Units (4 virtual cores
with 2 EC2 Compute Units each) Map Service
64-bit platform
ArcGIS Server + FGDB
Tiled Map Service
Internet
AWS Esri Instance
Basic Unit
…
54. Agenda: System Architecture Desing
• System Design Process
• GIS Technology
• Data Administration
• Network Communications
• Esri Architecture
• Performance
- Software
- Hardware
• Virtualization
• Example of CPT
55. User Expectations – a closer look
4.0
Web
IMS – ArcIMS
ADF – ArcGIS Server ADF SS/SOC
3.5 REST – ArcGIS Server REST MXD ArcSDE
MSD – ArcGIS Server REST MSD
DBMS
3.0
Light complexity display
Display Performance (sec)
2.5
Medium complexity display
2.0
Minimum user performance expectations.
1.5 When is the software fast enough?
1.0
0.5
0.0
IMS IMS IMS IMS ADF IMS ADF IMS ADF IMS ADF IMS ADF IMS ADF REST IMS ADF REST MSD IMS ADF REST MSD
...2001 ...2002 ...2003 ...2004 ...2005 ...2006 ...2007 ...2008 ...2009 ...2010 ...2011
Calender Year
56. Map Display Complexity
• Only show relevant data
- Start simple
- Use field visibility
• Use scale dependencies
- Appropriate data for given scale
AGD10 wkstn MXD 100%Dyn 10x7 Feature
- Same number of features at all scales Local FGDB Data Source
• Select the right point representation
Use appropriate text and labeling
- Use single layer simple or character markers
Use annotation instead of labels
- Use EMF instead of bitmaps Use indexed fields
Use label and feature conflict weights
- Use integer fields for symbol values sparingly
Avoid special effects (fill patterns, halos,
- Avoid halos, complex shapes, masking
callouts, backgrounds)
Avoid very large text size (60+ pts)
• Select the right lines and polygons
Avoid Maplex for dynamic labeling (avoid
- Use ESRI Optimized style overuse)
- Avoid cartographic lines and polygon
outline Things to do to keep the map service time within performance
budgets
58. ESRI Optimized lines and polygons
• Outlines for all fills are simple lines instead of
cartographic lines
• Picture fills are
EMF-based instead of
BMP-based
• ESRI Optimized
improves drawing
performance by 50%+
ESRI Optimized
improves performance
59. Use ArcGIS Optimize tool
• Analyze map directly from ArcMap (ArcGIS
9.3.1+)
- Analyze function included in Map Service
Publishing tools
- Provides performance tuning recommendations
- Preview allows for visual analysis of performance
and graphic quality
- Provides optimized display rendering time
60. Display complexity (Mxdperfstat)
• Measuring MXD Desktop display rendering time (http://arcscripts.esri.com/details.asp?dbid=16931)
Key performance metrics
Display refresh times at multiple
scales
Layer refresh times for each map
scale
Performance tuning
recommendations
61. Design Phase—Performance Factors
Performance Test Cache vs. MSD vs. MXD
2D Cache vs. Dynamic Data Response Time Comparison
6
When possible, use
5 Optimized Services
for
4 dynamic data.
Single user
Response Time (sec)
response times
3
are similar.
2
If data is static, use
1
cache map
Services.
0
1 6 11 16 21 26 31 36 41 46 51
User Count
2D Cache 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01
MSD 0.38 0.39 0.41 0.45 0.48 0.54 0.62 0.76 1.17 2.12 2.79
Dynamic 0.57 0.58 0.62 0.67 0.78 0.87 1.22 1.74 2.72 3.8 4.98
Cache map services use the least of hardware resources.
61
62. Web Mapping output image size
Smaller maps
travel faster
PNG24 = 49 KB
Traffic (56Kbps / 1.5Mbps)
0.5 Mbpd (10 sec / 0.6 sec)
PNG24 = 267 KB
Traffic (56Kbps / 1.5Mbps)
• Based on REST MXD service 2.7 Mbpd (151 sec / 2.3 sec)
- 600 x 400 resolution
- 1280 x 1024 resolution (same display)
• Recommendation: Use reasonable output image size
to support application while minimizing impact to
network
63. Web Mapping output image type
Output Selection Response
Image Size Traffic 56Kbps 1.5Mbps
RASTER JPEG 60 KB 0.6 Mbpd 13 sec 0.8 sec
PNG24 130 KB 1.3 Mbpd 36 sec 1.4 sec
PDF 180 KB 1.8 Mbpd 59 sec 1.9 sec
Output Selection Response
VECTOR Image Size Traffic 56Kbps 1.5Mbps
JPEG 60 KB 0.6 Mbpd 13 sec 0.8 sec
PNG24 50 KB 0.5 Mbpd 11 sec 0.7 sec
PNG8 30 KB 0.3 Mbpd 6 sec 0.5 sec
PDF 180 KB 1.8 Mbpd 59 sec 1.9 sec
High network traffic
Based on REST MXD Service
reduces performance
600 x 400 resolution
Recommendation: Use appropriate output type to
support application while minimizing network traffic
64. Concurrent User Loads
• Service Instance Configuration (concurrent users)
- Random arrival distribution reduces peak throughput
- 3-5 instances per core provides optimum throughput
- Display response time increases with max instance configuration
Random arrival queuing delays
Note: 4 core platform – 16 (4x4) instances provide peak throughput
What is optimum capacity setting?
65. Batch Process Loads
Service Instance Configuration (batch processing)
Batch process consumes a processor core
N+1 Instances provide optimum throughput
Batch run time increases with max instance configuration
Workflow Performance Summary
1.20
DPM/TPM 621 622 622 622 622 622 Client Latency
546 NWQ Network
1.00 100% 100% 100% 100% 100% 100%
Performance (sec)
Tier 01Q Tier 01
88% Tier 02Q Tier 02
0.80 410
Tier 03Q Tier 03
0.60 66% Tier 04Q Tier 04
273 No random arrival queuing Tier05Q Tier05
0.40 Tier06Q Tier06
44%
137 Tier07Q Tier07
0.20 Tier08Q Tier08
22% Tier09Q Tier09
0.00 Tier10Q Tier10
1 2 3 4 5 6 7 8 9 10
1 2 3 4 5 6 7 8 9 10
Batch Processes
Note: 4 core platform – 5 (4+1) instances provide peak throughput
What is optimum capacity setting?
66. Memory recommendations
• Too many instances per server can exhaust memory
- Increased paging when not enough memory
- Slower processing due to shared compute resources
- Maximum of 3-5 instances per core recommended
• Too few instances per server
- Can limit utilization of host
hardware
- Minimum of 3 instances per
core recommended
• Provide sufficient
memory to support
optimum performance
- Minimum of 3 GB memory
per core recommended
- More memory may be
required when using large
file data sources
67. Why should I care about map caches?
• Performance, performance, performance
• Scalability: Industry standard
• Cartographic quality
• ArcGIS Explorer Online
68. What should you cache?
• Base maps (always)
• Operational layers that satisfy one of the following:
- High volumes of traffic
- Don’t change often
- Cover small scales only
69. What about optimized map services?
• Optimized drawing format
- Based on MSD file
- Enhanced map drawing engine
- Use to generated map tiles faster
• Does NOT replace cache map service
- Requires dynamic rendering
- Web services are optimized for cached tiles
70. Tile size
• Pixel dimensions of each image
• 256x256 is the web standard
- 512 X 512 : legacy ArcGIS Online
256
• Larger dimensions are faster to
build, but tiles take longer to
download
256
71. Choosing an image format
• Image format effects
- Tile storage space requirements
- Web application performance
(speed and supported browsers)
- Tile image quality and transparency
.JPG
• JPEG
- Great compression for many colors but
not transparency
• PNG
- Best compression for less colors
.PNG
• Mixed
72. Cached imagery workflow
• Prepare imagery
Prepare
- Build raster pyramids
- Create mosaic dataset with overviews
• Author map document Author
- Add imagery to map
- Save as MSD
• Publish as a map service Publish
• Create and manage the map cache
- Configure cache definition
Create
- Generate cache tiles
- Update cached tiles
74. Standard Performance Evaluation Corporation
How do we measure relative
performance?
SPEC mission:
To develop technically credible and objective benchmarks so
that both computer designers and purchasers can make
decisions on the basis of realistic workloads
ESRI system design history
1992–1996, SPEC92 benchmark results
1996–2000, SPEC95 benchmark results
2000–2006, SPEC2000 benchmark results
2006+, SPEC2006 benchmark results
SPEC benchmarks have served ESRI customers well
78. Agenda: System Architecture Desing
• System Design Process
• GIS Technology
• Data Administration
• Network Communications
• Esri Architecture
• Performance
- Software
- Hardware
• Virtualization
• Example of CPT
79. Desktop Virtualization
Customer Experiencie and Motivations
• Simplified/Streamlined Administration
• Release control; Managing Upgrades
• WAN bandwidth and latency tolerance
• High availability for ArcGIS Desktop solutions
• Reduced client side requirements
• Specialized skill requirements for server
administration and configurations to support
printing, working with local data, etc.
80. Enhancing Desktopo Deployment - Infraestructure
• App/Session Virtualization
+ High Availability
+ Simplified Administration
+ Network efficiency / WAN
- 3D
- System admin skill level
~ User Experience
84. Recommendations for VMWare
Penalty of virtualization (Physical vs. various VM configs)
45000,00 32% Degradation 11% Degradation
40000,00
35000,00
Throughput (Tr/Hr)
30000,00
25000,00
20000,00
15000,00
10000,00
5000,00
0,00
1VM,4C,16R 2VM,2C,8R 4VM,1C,4R 4VM,1C,2R 1P,4C,16R
Virtual Machine Configuration
85. Example: Virtual Server peaks of Works
Web Browser
Internet
Intranet
SOM SOM
SOC SOC SOC SOC SOC
Direct Connects Direct Connects Virtual Server
Physical Servers
DBMS
86. Agenda: System Architecture Desing
• System Design Process
• GIS Technology
• Data Administration
• Network Communications
• Esri Architecture
• Performance
- Software
- Hardware
• Virtualization
• Example of CPT
88. City of Portland - Solution Architecture
City Hall
Internet
T-1
15 Mbps
1 Gbps LAN
T-1
15 Mbps
Sandy River
Maintenance Facility
T-1 T-1
15 Mbps 15 Mbps
89. City of Mini-Portland - Solution Architecture
City Hall
Internet
T-1
15 Mbps
1 Gbps LAN
T-1
15 Mbps
Maintenance Facility
T-1
15 Mbps
City of Minnie-Portland
90. User Requirements
ArcGIS Desktop (Power, Editor, Viewer)
Desktop (LAN), Citrix (remote sites)
ArcGIS Server (Internal and Public web services)
91. User Workflow Requirements
• Suggested DESIGN procedure to build the city of Mini-
Portland capacity planning model.
- ArcGIS Desktop ArcInfo (AI) Power users
- AGD10 wkstn MXD Med 100%Dyn 10x7 Feature
- AGD10 WTS Citrix MXD Med 100%Dyn 10x7 V ICA
- ArcGIS Desktop Editors
- AGD10 wkstn MXD Lite100%Dyn 10x7 Feature
- AGD10 WTS Citrix MXD Lite 100%Dyn 10x7 ICA
- ArcGIS Desktop Viewer
- AGD10 wkstn MXD Lite100%Dyn 10x7 Feature
- AGD10 WTS Citrix MXD Lite 100%Dyn 10x7 R ICA
- ArcGIS Server internal Web mapping application
- AGS10 REST MXD Med 100%Dyn 10x7 R JPEG
- ArcGIS Server REST MSD Public Web mapping application
- AGS10 REST MSD Lite 100%Dyn 10x7 V PNG8
92. User Workflow Requirements
- ArcGIS Desktop ArcInfo (AI) Power
users
- AGD10 wkstn MXD Med 100%Dyn 10x7
Feature
- AGD10 WTS Citrix MXD Med 100%Dyn
10x7 V ICA
- ArcGIS Desktop Editors
- AGD10 wkstn MXD Lite100%Dyn 10x7
Feature
- AGD10 WTS Citrix MXD Lite 100%Dyn
10x7 ICA
- ArcGIS Desktop Viewer
- AGD10 wkstn MXD Lite100%Dyn 10x7
Feature
- AGD10 WTS Citrix MXD Lite 100%Dyn
10x7 R ICA
ArcGIS Server internal and public Web mapping application
93. User Workflow Requirements
AGS10 REST MXD Med 100%Dyn 10x7 R JPEG
and... Where is AGS10 REST MSD
Lite 100%Dyn 10x7 V PNG8?
96. City of Mini-Portland - Solution Architecture
City Hall
Internet
T-1
15 Mbps
1 Gbps LAN
T-1
15 Mbps
Maintenance Facility
T-1
15 Mbps
City of Minnie-Portland
97. Additional Resources
• System Design Process
- http://wiki.gis.com/wiki/index.php/System_Design_Process
• Esri Press. Building a GIS
- http://esripress.esri.com/display/index.cfm?fuseaction=display
&websiteID=141
• On-line resource center
- http://esripress.esri.com/display/index.cfm?fuseaction=display
&websiteID=141&moduleID=27
98. Additional Resources
• ESRI project methodology
- www.esri.com/services/professional-
services/methodology.html
• Business case resources
- The Business Benefits of GIS: an ROI Approach—Outlines case
studies and general methodology for doing cost-benefit
analysis
- Thinking About GIS—Roger Tomlinson
- www.esri.com/getting_started/executives/success.html
• Project Management Body of Knowledge (PMBOK)
- www.pmi.org