GIS Hardware/Software System Architecture and Design Parts of this lecture draw upon: Dave Peters, ESRI User Conference 2002 And Ronald Briggs, U Texas, Dallas GIS Software system components User interface Tools and functions Data manager System architecture configurations Desktop Client-server Centralized desktop Centralized server GIS software categories Desktop Server (Internet) Developer Hand-held Other ESRI’s GIS Components c:\ ArcGIS Workstation Clients ArcInfo ArcEditor ArcView ArcMap ArcCatalog ArcToolbox ArcMap ArcCatalog ArcToolbox ArcMap ArcCatalog ArcToolbox ArcEngine/ ArcObjects Application Development & Customization $ Consistent interface Increasing capability ArcExplorer Browser ArcServer Services ArcIMS Services ArcSDE Services Internet Source: ESRI with mods. GIS Characteristics Detailed, high resolution graphic displays Large data files Intensive data analysis/computation Need powerful workstations Need powerful servers for data retrieval and batch processing Need high capacity networks These requirements differ significantly from standard IT environments. Information System Components Computer hardware: physical machinery, boxes, cables, connectors Software: instructions which make hardware perform as desired by user – Operating System to control the basic functions of the hardware and networks – Applications to provide users with desired results Data and databases to store information required by users Networks to distribute information between different computers and users Orgware: people, procedures & organizational structures to make all of the above function – 85% of cost of IT? The Computer: Hardware Components Central processing unit (CPU): – microprocessor (control unit and arithmetic/logic unit) – primary storage -RAM (main memory or simply memory) Storage -disks -tapes Secondary storage Central – disks » magnetic » optical – tapes Input/Output Devices (I/O) communications devices/ network connections Input devices cpu Main Memory Processor Networks Output devices (hardcopy) Some Measurement Concepts for CPU and Storage capacity – – – – – – bits and bytes: 8bits=1 byte kilobytes (KB), megabytes(MB), gigabytes(GB), terabytes (TB),petabytes(PT) (x1,000) 1KB=1,000bytes 1TB=1,000,000,000,000bytes (approx.) 5MB: the text of all of Shakespeare’s works 1TB: 2001 digital orthos for one large city 10TB: text of the library of Congress speed – – – – hertz: number (frequency) of electrical pulses per second clock cycle (megahertz: MHz): (8088: 4.7MHz; p5: 133MHz; p5II 450MHz) MIPS: million instructions (integer) per second (very~ = 1 MHz) Gigaflops: Billions of floating point operations per second Central Processing Unit: performance Moore’s Law: Performance doubles every 18 months processor speed – clock cycle (megahertz: MHz): 8088-4.7MHz PIII-750MHz – word length (8,16,32,64) at single address (8088-16; PIII32) – data bus width (processor to main memory) (8088-8; PIII-64) Examples – 4.7Mhz - 2.2Ghz (pc/ws) – Servers substantially more main memory capacity (RAM): (size of program &/or data file) – 256MB-1GB (pc) – Servers substantially more Processors & their organization – single v. multi processing » 1 v 2,4,8 processors – massively parallel processing » 1,000 processors Ability of multi-processors to improve performance depends on operating system capabilities and application software design Moore’s Law (Wikipedia) Moore's law is the empirical observation that at our rate of technological development, the complexity of an integrated circuit, with respect to minimum component cost, will double in about 18 months. Storage: disks and tapes – seek time (to find data) » disk: random access » tape: sequential access capacity – 4,000 (80x50) bytes per printed page – 350 pages per 1.44MB diskette – 50,000 pages per 200MB 1/2 inch reel/cartridge – 150,000 typed pages per 640MB CD-ROM – 1-7 Gigabytes per 4mm or 8mm cassette – 40-80GB per DEC DLT (digital linear tape) – DVD (4.7-17GB) performance (speed) – transfer rate – cache size and performance – SCSI (fast) v. IDE (cheap) bus magnetic disk cost – $15,000 per gigabyte in 1985 (mainframe) – $500 per gigabyte in 1995(pc) – $30 per GB in 1999 – $2 per GB in 2002 for IDE – $7 per GB in 2002 for SCSI traditional trade-off: disk---optical---tape Speed & cost capacity IBM Hollerith Card Input/Output: User Interface “IBM” punched card – Herman Hollerith, 1884 Graphical user interface (GUI) – bit-mapped display: – each picture element (pixel) manipulable Resolution – CGA: 320x300 pixels with 4 colors (96,000 pixels) – VGA: 640x480 with 16 colors – SVGA: 800x600 or 1024x768 with 256 colors – XGA: 16,777,216 colors at 1024x768 (786,432 pixels) 24 bit color – RGB: red/green/blue primary colors – 1x8bit register per color =256 intensity values – 3 primary colors @ 256 each = 16,777,216 combinations (256x256x256) Xerox Star 8010 with mousecontrolled GUI April 1981 Issues: Performance bottlenecks CPU performance Memory (RAM) I/O: to disk and/or to network Network performance: speed and/or load Balanced system critical Issues: Capacity Planning peak load versus standard load Issues: CPU Selection & configuration Servers: database, geoprocessing, web, terminal Client: Workstation, terminal client, browser client Mobile: portable, palm, pocket (laptop, PDA, phone) Example ArcIMS (Web) Server Configurations ArcIMS Components Single Tier Config. Two Tier with load balancing Three Tier Three Tier with Load balancing Biggest processing requirement Note: data is duplicated Note: data is from database server Example ArcIMS configurations relative to security Least secure One of many intermediate solutions Most secure Client Configurations Web server Terminal Server GISWeb Software runs on web server Can hand-off some work to GP server Data stays on servers Image to Client Low end PC with standard web browser Low network bandwidth Limited GIS functionality (depends on GIS Web software) Web Link Web Browser Geoproc. Server GIS Software runs on terminal server Can hand-off some work to GP server Data stays on servers Image to Client Low end PC with special TS software Low network bandwidth Full GIS functionality (Citrix is primary vendor) LAN link Or VPN* Terminal Client *VPN Virtual Private Network Database Server Servers GIS Software runs on PC Data to PC High end PC High network bandwidth Full GIS functionality LAN link PC Workstation Mobile Client Considerations Device type: Portable, tablet, palm or pocket? – Palm and pocket likely to merge – How much capability is required » In field editing or info look up? Is Connectivity required? – Wireless data connections slow and expensive How to keep databases in synch? Software and Programs Software: instructions to the computer Program: a set of instructions which accomplish a specific task. All computers use the stored program concept for program execution: – instructions (program) reside permanently on secondary storage (disk) – program is copied and stored (loaded) into main memory to be run (executed). to be executed, programs must be in binary machine language – compilers convert a programmer’s source code to binary code » » » » once converted, can be run many times most purchased packages delivered in binary form will only run on the OS for which binary was created cannot be “understood” or modified by the user Software Layers HARDWARE Firmware, Device drivers Operating System Utilities Data Base Software Application Software User Interface USER Operating Systems: Primary Current Systems Portable – Windows CE – Palm Desktop: Advanced (professional) – Windows XP Professional – UNIX of various flavors Windows Server – Server – Advanced server – Data center server Basic (home) – Windows 95/98/Me/XP home – MAC System 9 and X – Linux (Red Hat, Caldera) Server/Network/Enterprise UNIX: Solaris (SUN), HP-UX, IBM AIX Linux Legacy – IBM MVS & OS/390 (mainframe) – IBM AS/400 (minicomputer) – Digital Equipment Corp (DEC): Open VMS (1999 Compaq announced 5 yr support) – UNIX niche (Unisys, NCR, Sequent) Significance of Different Operating Systems Upside to Variety: advantages for particular groups of users or applications – legacy for hard core data processing » extreme reliability » maximized for I/O and data access » binary compatibility between versions – UNIX for scientific processing, Internet applications and, increasingly, database applications » extreme flexibility » horsepower for number crunching – Windows » user oriented, but with power left for processing – MAC » user friendly – Wireless » Access from the field Downside to Variety – – – – – user training user interface development different software binaries support staff expertise communications and networking complexities Web browsers and JAVA help address some of these issues, but really only the first two GIS may have to access legacy systems for data Software for GIS: ArcGIS--ARC/INFO (ESRI, Inc., Redlands, CA) – – – – – Trying very hard to move from Desktop/Present. to Enterprise/Industrial with newer MapX, MapXtreme and MapInsight products Strong presence in telecom and business, especially site selection & marketing Intergraph (Huntsville, AL) – – – – privately owned by Jack Dangermond, a legend in the field and former Harvard student The only two originated commercial GIS and still clear market leader with about a third of the market “pure GIS” companies. Strong in gov., education, utilities and business logistics MapInfo – The Main Vector Players origins in proprietary CAD hardware/software strong in design and FM (facilities management), and running hard to match ESRI in GIS UNIX-based MGE (Modular GIS Environment) and FRAMME evolved from CAD after split with Bentley, courageously committed to NT and object technology in 1995 and first released “new generation” GeoMedia product in March 1997 Serious threat to ESRI until ArcInfo 8 released. Bentley Systems (Exton, PA) – Bentley brothers (Barry & Ray) originally developed the PC-based Micro-Station product under contract with Intergraph, but split in 1995 – – have very successfully continued to develop and sell MicroStation GeoGraphics since then. Strong in engineering; advertises itself as “geoengineering” Autodesk’s AutoCAD Map – – – – dominant CAD supplier and world’s 4th largest software company fully topological AutoCAD Map introduced in 1996 (but no raster) High initial expectation (huge installed AutoCad base) but hasn’t lived up to earlier expectations Primarily small business/small city customer base Software for GIS: Remote Sensing/Raster – Different players – origins in remote sensing thus raster oriented – Some now include vector support, but has proven insufficient for them to really compete with vector-based GIS – Convergence of raster and vector GIS software has not occurred to the degree expected 5 years ago – Need one of these products if you are heavy into remote sensing, image analysis, ororthophotography production – new satellite data products enhance their utility – Will never compete in the enterprise/management market – ERDAS/Imagine long established leader, purchased by Leica Systems of Switzerland in 2001 – ER MAPPER very agressive newcomer originating in Australia – Envi, another relative newcomer, radar specialization, now (2000) Kodak owned – PCI long-term Canadian player – CARIS another newer Canadian entry Software for GIS: other professional/analytical Also…. Other Commercial Players – Manifold (CDA International Corp) one of the best deals around – Maptitude (Caliper Corp, MA): another low cost one – CadCorp SIS (Spatial Information System) recent entry from UK – WinGIS (Progis Corp, Austria): European entry – Smallworld Systems (Englewood, CO) first to use OO (in early 1990s) but never broke out of its FM utilities base and bought by GE in 2000 GRASS – – – raster oriented, but some vector routines originally developed by US Army Construction Engineering Research Lab(CERL) army ended dev. & support in 1996 but assumed by Baylor University. IDRISI (Clark Univ) – – – one of the pioneering, university developed packages, limited commercial penetration but still trying raster orientation with some vector capabilities GRASS: Unix Tkl/Tk Manifold GIS Release 6.5 Software for GIS: Niche products Business Targeted – BusinessMAP (from ESRI): $99 choropleth and pin mapping (originally Maplinx) – NDS (National Decision Systems): marketing data supplier; I-Mark software (released 1999) combines earlier Infomark Express (reporting) and I-Map (maps) – Tactician: specialist product for market analysis, site selection, etc. – GeoQuery (Naperville, IL): a business mapping product based on Intergraph GeoMedia (originally an independent product) Other Industry Targeted – TransCAD (from Caliper): specialized GIS for transportation – EDX, Planet (from Mobile Systems International, now part of Marconi), Network Engineer (from Mesa Solutions), Cell CAD II: wireless telephony planning GIS specialized functions – Surfer (Golden Software): contour creation & spatial interpolation – Map Viewer (Golden Software): thematic map creation – FreeGIS (www.freegis.org) – Xmap Geographic (DeLorme): 3-D TopoQuads display, image registration, coordinate geometry – Blue Marble Geographic Calculator: projections, conversions, etc. Extensions to standard GIS Products – Add-ons to standard GIS packages to meet niche needs Estimated Market Shares GIS Vendor Marketshare 2001 Other 14% Logica 3% ESRI 34% SICAD 5% Total Revenue Estimates 2001 1,073 M IBM 5% 2004 2, 020 M MapInfo 6% Leica Geosystems (ERDAS) 6% Autodesk 7% GE Netw ork Solutions 7% Intergraph 13% 2003 Sector share Utilities Public sector Private sector 44% 29% 24% 2003 Type share Software Services Data Hardware 64% 24% 8% 4% ESRI Product Line-up: ArcGIS client products ArcReader (“adobe acrobat” for maps) & ArcExplorer (spatial data viewer) – Free viewers for geographic data. ArcGIS 9.x Desktop: two primary modules (MS NT/2000/XP only) 1. ArcMap: for data display, map production, spatial analysis, data editing 2. ArcCatalog: for data management and preview ArcToolbox, for specialized data conversions and analyses, available as a window in both Available capabilities within these modules are “tiered” » » » ArcView: viewing, map production, spatial analysis, basic editing ArcEditor: ArcView, plus specialized editing ArcInfo: ArcView & ArcEditor plus special analyses and conversions Extensions: for special apps.: Spatial Analyst, 3D Analyst, Geostatistics, Business Analyst,etc. ArcObjects: build specialized capabilities within ArcMap or ArcCatalog using VB for Applications ArcGIS Workstation (for UNIX and MS NT/2000/XP) – the old command line ArcInfo 7.1 ArcGIS Engine (MS NT/2000/XP) – – Set of embeddable GIS components (ArcObjects software objects) for building custom applications Runs under Windows, Unix and Linux, with support for Java, C++, COM and .NET – Replaces MapObjects which were based upon a previous generation of GIS objects Notes: ArcGIS 8 released 2000 to integrate two previous standalone products: ArcView and ArcInfo ArcGIS 9 released 2004 providing the full capability that should have been in ArcGIS 8!!! --full support for all data types (coverages, shapefiles, geodatabases) --full support for all previous geoprocessing analyses --Modelbuilder for scripting and repetitive processing --ArcEngine for building custom applications ArcView 3.3 (the predecessor to ArcGIS 8.x) the only GUI option for UNIX. ESRI Product Line-up: ArcGIS server products SDE (Spatial Database Engine) – middleware to support spatial data storage in standard DBMS – Supports all major industry databases: » Oracle, SQL-Server, IBM DB2, Ingres ArcGIS Server – Permits the creation of server-based GIS services using any ArcGIS capability – Provides GIS capabilities to a user without a desktop GIS system: » inward focus—user goes to server ArcIMS – Software to develop Internet server-based mapping and basic analysis – Provides maps to the user without a desktop GIS system : » outward focus—gives user a map ArcGIS Services – Server based applications built and operated by ESRI or its partners and made available on the Internet for subscription – Normally charged on a “per transaction” basis, but can be flat fee – presumably built using ArcGIS Server Decisions/Actions Required for Software Operating systems for clients and servers Interfacing with existing non-GIS application systems Selection of GIS software appropriate for task required GIS Software Selection Council, get off my back! Planner Needs faster, easier, jazzier way of answering council member queries. – Need professional analysis package – use ArcView 9, since city has extensive ArcInfo shapefiles & coverages; – Uses ModelBuilder and Python scripts to automate most common queries Appraisal District needs to accurately maintain parcel files – Replaces current CAD system with ArcEditor for accurate topological editing and easier interface with GIS files Emergency Preparedness Manager for coastal county needs application to track potential spread of oil spills – Uses ArcObjects and VB for Applications within ArcInfo version of ArcGIS to develop customized model City Planner wants easier way for citizens to know location of city facilities, property zoning, roads, etc.. – Developes ArcIMS application on city server which includes layers for roads, zoning, parcels, schools, other city facilities, etc.. which citizens can access from their home PCs using any browser Contd on next slide GIS Software Selection Bus service manager. Wants application to display real time location of buses on touch-screen terminal to passengers waiting at transit centers – Uses ArcServer to develop central application to track buses – Uses internet browser at transit station to query ArcServer Developer of software to support operations within a real estate office, including payroll, client tracking, billing, etc.., requires way to map location of houses sold by office – Uses ArcEngine to incorporate mapping capabilities within his software system Taxation aqnd Assessment. Must tie all my data together--land ownership, tax rolls, utility lines, roads, 911 calls etc. – Needs enterprise solution. – ArcSDE with ORACLE data base, accessed with ArcExplorer (free), ArcView 9 (for read only & analysis), ArcEditor 9 (for data maintenance) & ArcInfo 9 (for specialized analyses) Functional Types of Data strategic: long term direction, goals (us versus them; Ford v. GM; Plano v. Richardson Tactical/procedural: comparisons for short term decision making (is it selling?, focus policing at high accident/crime sites; identify roads) operational: daily activities (inventory replenishment; repair orders) transactional: flows through the system ( water meter readings, billings, 911 calls). Database Choices Workstation ArcSDE DB Middleware ArcSDE supports Oracle Microsoft SQL Server IBM’s DBII Informix Some database vendors have ability to support spatial data directly without ArcSDE (e.g. Oracle Spatial) Selection often a function of what already exists in house for business applications may conform to ISO standard Better security May cost less than ArcSDE More limited capability (e.g. no “geodatabase”) Does not protect against disastrous destruction! (WTC 9/11, tornado, etc.)