| GISdevelopment.net ---> GITA 1999 ---> Enterprise Resource Planning |
From GIS to Spatial Resource Planning: The Benefits of
Integration to the Customer
Dick Newell
Chairman
Smallworld
Elizabeth House
1, Chesterton High Street
Cambridge, CB4 lWR, UK
Email: Dick.Newell@Smallworld.co.uk
Dick Newell
Chairman
Smallworld
Elizabeth House
1, Chesterton High Street
Cambridge, CB4 lWR, UK
Email: Dick.Newell@Smallworld.co.uk
Introduction
That utilities and communications companies need to become competitive commercial organisations that deliver shareholder value is now well known. That this has become a worldwide phenomenon is also well known. Until these major business drivers became prevalent, GIS was always going to be a Cinderella technology addressing automated mapping and facilities management applications confined to the bounds of the mapping department, or alternatively a small-time project-based departmental application used on a one-off basis to solve a particular problem. The GIS vendor industry has had to go through 2 transitions in order to meet the challenges faced by their customers in these industries. The first transition, from digital mapping to GIS, is far advanced for a few vendors. The second transition from GIS to something beyond GIS that we call Spatial Resource Planning (SRP) has only just begun.
From Digital Mapping to GIS - The importance of architecture
Digital mapping systems, often based on CAD principles, or principles more appropriate to solving gee-relational problems suffered many drawbacks when applied to the problem of managing large continuous networks.
The shift to GIS meant the following changes of emphasis:
Modelling a network should focus on the objects in the network. Treating spatial attributes (features) in some way different from other object attributes misses the point. Spatial is not special.
Object orientation is the natural way to model object behaviour. It reduces project implementation costs and greatly reduces the risk of costly failures or project over-runs. Beware object-oriented wrappers around defective technology - no matter how you wrap a mackerel, it always smells like a mackerel.
The relational model is well proven. Why use a special format for spatial data? Everything should be managed in a single seamless environment - a RDBMS.
Analysis is a small, but important, part of the problem. Systems that get the modelling right make the analysis much easier. The gee-relational model only solves a limited set of problems, and is not particularly appropriate for networks.
One vendor recently announced that they had just realised that their customers want to access their databases more than one user at a time - a remarkable observation in 1998. The traditional approach of using a large number of checked-out single-user database subsets ran out of steam years ago. A multi-user version-managed database is the way to go.
Systems that demo well on small project-based databases do not scale to the petiorrnance requirements of an enterprise-wide, mission-critical network-GIS.
All of these have provided, and in many instances continue to provide, a challenge to the vendor community. That is not to say that one cannot deploy an operational system by starting with a defective architecture, there are some limited successes. These include systems now in operational use where the vendor has gone out of business. In all cases these “successes” have been achieved at huge expense over many years, and are a testament totheresolve of the people involved ingesting thejob done. Costs of $100 million dollars and more have been run up by these early pioneers, resulting in systems of limited capability or even worse, failure. It is also true to say that fi.uther large expenditures are required to maintain these systems and even more time and money is needed to move firther into the realms of Spatial Resource Planning.
This section could be summed up by saying that there has been a change of emphasis from a hodge-podge of badly integrated, incompatible pieces of system to an emphasis on overall system architecture. Such an architecture, comprising compatible components and a single consistent data model manipulated by appropriate languages, significantly lowers implementation, operational and maintenance costs and reduces the risks of failure or cost overrun. It is remarkable that those systems that had their roots in the technologies of the 70’s and early 80’s may not have accommodated many of these concepts until the next millennium, even though they have been well understood and available in some systems for the whole of the nineties.
The Spatial Resource Planning Market Place
The analysts, such as Dataquest, tell us that the worldwide market for “GIS software” is of the order of 1 billion dollars, that another 4 to 5 billion is spent on “services” (not including data capture) and another 1.5 billion on hardware. What “GIS software” includes is not precisely defined, but may be as broad as any system that is capable of putting a map on a screen. It certainly includes desktop mapping, CAD systems that are used for mapping, as well as filly functional, professional, multi-user GIS. Here lies part of the reason why there have been so many disappointing results from GIS projects in the early years - the considerable risk in the spending of that 4 to 5 billion dollars. The analysts also tell us that this market is growing at a rather tame 110/0per annum. This is absurd when one considers the amount that is still to be done in the utility and communications companies of the world to get operational systems into place to satisfy the business drivers of competition, customer satisfaction and shareholder value.
The Enterprise Resource Planning (ERP) space occupied by giants such as SAP, Oracle, PeopleSoft and Baan is growing at a considerably faster pace, yet the fimdamental business drivers of their success are essentially the same.
This is truly strange, given the old clich~ that 80% or more of data relates to location. In some work that was undertaken in Germany recently, a group of utility companies was asked to list their most important business processes and then to note those that required GIS. Their list is shown in Table 1.
Table 1. Electric and GAS Business Processes (* denotes requires GIS)
Although a small number of these processes can be addressed by products that can be bought off-the-shelf, the tradition has been that these processes have been satisfied by custom development or specialist applications. Apart from the GIS market there are also markets for systems that address SCADA, Works and Asset Management, Hydraulic Analysis, Power Simulation, Outage Management, Network Design and Documentation and Circuit Provisioning in the communications industry.
Figure 1: Shared Network Model (Maintained by GIS)
All of these require a model of the network, which is the province of the GIS to manage and maintain. In an ideal world, there would only be one representation of the network in the organisation from which all of these systems could work. Although it may not be practical to achieve this physically, it is essential to achieve at least a logically equivalent state if serious problems of inconsistency and integrity are to be avoided.
The GIS is the system that owns, manages and maintains that single network model. It is asking for trouble if two systems are allowed to update different representations of the same network. Such architectures based on round-trip propagation of change can be very difficult to maintain, unless a robust 2-way replication scheme is put in place. One-way-trip replication from the GIS out to these other systems is the way to go, if the ideal of these other systems directly accessing the GIS model cannot be achieved. The huge benefit of this is that all of these systems that need to use a network do not themselves need sophisticated systems to build and maintain that network. That is what the R&D dollars of the GIS vendors have been spent on these past years. That these systems now have a robust model with integrity to rely on increases their own value-added contribution. It is a “one plus one equals three” argument. Not only does one model avoid unnecessary duplication, but backup and recovery procedures become easier and one development environment reduces development costs.
From GIS to Spatial Resource Planning - The Importance of an Enterprise Strategic Approach
Unlike a few years ago, proven, state-of-the-art GIS technology is now capable of modelling large networks and handling enormous databases accessed concurrently by large numbers of users. One of the biggest deployments in the world is at the National Land Survey of Finland who have deployed 700 concurrent update GIS seats, in 44 sites distributed across Finland, accessing a single seamless GIS database in Helsinki containing 5 million land parcels managed in 7000 database versions, all with good performance.
To use an automotive analogy, the brakes have been removed; privatisation and deregulation have put the foot on the accelerator. Even the fuel is there; data availability and data conversion costs, are greatly improved compared to just a few years ago. So now, nothing should stop a major deployment of the technology. Many European, American and Asian Water, Gas, Electric and Communications companies have now achieved major deployments resulting in large benefits for their organisations. But just as in other parts of the software industry, what was achieved by bespoke development will in the fiture be achieved by suites of standard applications - Spatial Resource Planning applications. Some of these applications will be achieved by interfacing or embedding. Other applications will be achieved by a development on top of a GIS. Either way, the customer will be able to see what he is getting ahead of his project and this will reduce implementation costs and time-scales and lower risks (see Figure 2).
Figure 2: Reducing costs with improved approaches
There could be a merger of some niche areas, which were formerly addressed by separate companies, for example GIS and Outage Management. The process of Outage Management is addressed in the main by integrating a Customer System, Network Management System and SCADA (See Figure 3).
Figure 3: The Network-Premise-Customer Link
Simply put, this provides complete connectivity from a piece of equipment in the real world to a model of that equipment. Then, via a model of a connected network to a model of a customer premise, from the premise to a record of the person and hence to the real customer that depends on the equipment. This complete end-to-end connectivity is the fundamental basis of business processes that restore outages, plan outages, plan network maintenance and so on.
There are many benefits from addressing business processes by integrating application suites with GIS in an SRP framework giving a common operating environment:
Conclusion
In moving from digital mapping to GIS, the focus was a mainly technical one to provide core technology which had adequate virility to do the job. Now that this phase is over, the focus moves from one of technical virility of the GIS to a different list, more relevant to satisfying the demand for deployable enterprise-wide processes:
So we will see the following trends:
Figure 4: Increased benefit with move into SW applications
Spatial Resource Planning applications are a way of providing leveraged access to the network data by a much larger community of users. The benefit increases out of all proportion to the additional investment. If implementing an initial GIS may have seemed like pushing a boulder up a hill, the payoff from implementing SRP applications to exploit the database is rather like rolling down the other side.
That utilities and communications companies need to become competitive commercial organisations that deliver shareholder value is now well known. That this has become a worldwide phenomenon is also well known. Until these major business drivers became prevalent, GIS was always going to be a Cinderella technology addressing automated mapping and facilities management applications confined to the bounds of the mapping department, or alternatively a small-time project-based departmental application used on a one-off basis to solve a particular problem. The GIS vendor industry has had to go through 2 transitions in order to meet the challenges faced by their customers in these industries. The first transition, from digital mapping to GIS, is far advanced for a few vendors. The second transition from GIS to something beyond GIS that we call Spatial Resource Planning (SRP) has only just begun.
From Digital Mapping to GIS - The importance of architecture
Digital mapping systems, often based on CAD principles, or principles more appropriate to solving gee-relational problems suffered many drawbacks when applied to the problem of managing large continuous networks.
The shift to GIS meant the following changes of emphasis:
-
From tiles & layers to seamless spatial databases
From spatial feature-centred to object-centred
From procedural programming to object-orientation
From proprietary spatial formats to RDBMS
From analysis focus to modelling focus
From single user to many concurrent users
From short transactions to long transactions
From check-out to version management
From small database to enormous database
Modelling a network should focus on the objects in the network. Treating spatial attributes (features) in some way different from other object attributes misses the point. Spatial is not special.
Object orientation is the natural way to model object behaviour. It reduces project implementation costs and greatly reduces the risk of costly failures or project over-runs. Beware object-oriented wrappers around defective technology - no matter how you wrap a mackerel, it always smells like a mackerel.
The relational model is well proven. Why use a special format for spatial data? Everything should be managed in a single seamless environment - a RDBMS.
Analysis is a small, but important, part of the problem. Systems that get the modelling right make the analysis much easier. The gee-relational model only solves a limited set of problems, and is not particularly appropriate for networks.
One vendor recently announced that they had just realised that their customers want to access their databases more than one user at a time - a remarkable observation in 1998. The traditional approach of using a large number of checked-out single-user database subsets ran out of steam years ago. A multi-user version-managed database is the way to go.
Systems that demo well on small project-based databases do not scale to the petiorrnance requirements of an enterprise-wide, mission-critical network-GIS.
All of these have provided, and in many instances continue to provide, a challenge to the vendor community. That is not to say that one cannot deploy an operational system by starting with a defective architecture, there are some limited successes. These include systems now in operational use where the vendor has gone out of business. In all cases these “successes” have been achieved at huge expense over many years, and are a testament totheresolve of the people involved ingesting thejob done. Costs of $100 million dollars and more have been run up by these early pioneers, resulting in systems of limited capability or even worse, failure. It is also true to say that fi.uther large expenditures are required to maintain these systems and even more time and money is needed to move firther into the realms of Spatial Resource Planning.
This section could be summed up by saying that there has been a change of emphasis from a hodge-podge of badly integrated, incompatible pieces of system to an emphasis on overall system architecture. Such an architecture, comprising compatible components and a single consistent data model manipulated by appropriate languages, significantly lowers implementation, operational and maintenance costs and reduces the risks of failure or cost overrun. It is remarkable that those systems that had their roots in the technologies of the 70’s and early 80’s may not have accommodated many of these concepts until the next millennium, even though they have been well understood and available in some systems for the whole of the nineties.
The Spatial Resource Planning Market Place
The analysts, such as Dataquest, tell us that the worldwide market for “GIS software” is of the order of 1 billion dollars, that another 4 to 5 billion is spent on “services” (not including data capture) and another 1.5 billion on hardware. What “GIS software” includes is not precisely defined, but may be as broad as any system that is capable of putting a map on a screen. It certainly includes desktop mapping, CAD systems that are used for mapping, as well as filly functional, professional, multi-user GIS. Here lies part of the reason why there have been so many disappointing results from GIS projects in the early years - the considerable risk in the spending of that 4 to 5 billion dollars. The analysts also tell us that this market is growing at a rather tame 110/0per annum. This is absurd when one considers the amount that is still to be done in the utility and communications companies of the world to get operational systems into place to satisfy the business drivers of competition, customer satisfaction and shareholder value.
The Enterprise Resource Planning (ERP) space occupied by giants such as SAP, Oracle, PeopleSoft and Baan is growing at a considerably faster pace, yet the fimdamental business drivers of their success are essentially the same.
This is truly strange, given the old clich~ that 80% or more of data relates to location. In some work that was undertaken in Germany recently, a group of utility companies was asked to list their most important business processes and then to note those that required GIS. Their list is shown in Table 1.
| Customer-oriented Processes: |
| * Meter Reading |
| Contract Management |
| Track Accounts Receivable |
| * Handle Customer Inquiries I |
| * Provide Customer Service I |
| Technical Processes: |
| * Plan and Coordinate Network Construction |
| * Perform Maintenance |
| * Fault Management |
| * Generate House Connection I |
| Supporting Processes: |
| * Acquire and Manage Materials |
| Provide new IT |
Although a small number of these processes can be addressed by products that can be bought off-the-shelf, the tradition has been that these processes have been satisfied by custom development or specialist applications. Apart from the GIS market there are also markets for systems that address SCADA, Works and Asset Management, Hydraulic Analysis, Power Simulation, Outage Management, Network Design and Documentation and Circuit Provisioning in the communications industry.
Figure 1: Shared Network Model (Maintained by GIS)
All of these require a model of the network, which is the province of the GIS to manage and maintain. In an ideal world, there would only be one representation of the network in the organisation from which all of these systems could work. Although it may not be practical to achieve this physically, it is essential to achieve at least a logically equivalent state if serious problems of inconsistency and integrity are to be avoided.
The GIS is the system that owns, manages and maintains that single network model. It is asking for trouble if two systems are allowed to update different representations of the same network. Such architectures based on round-trip propagation of change can be very difficult to maintain, unless a robust 2-way replication scheme is put in place. One-way-trip replication from the GIS out to these other systems is the way to go, if the ideal of these other systems directly accessing the GIS model cannot be achieved. The huge benefit of this is that all of these systems that need to use a network do not themselves need sophisticated systems to build and maintain that network. That is what the R&D dollars of the GIS vendors have been spent on these past years. That these systems now have a robust model with integrity to rely on increases their own value-added contribution. It is a “one plus one equals three” argument. Not only does one model avoid unnecessary duplication, but backup and recovery procedures become easier and one development environment reduces development costs.
From GIS to Spatial Resource Planning - The Importance of an Enterprise Strategic Approach
Unlike a few years ago, proven, state-of-the-art GIS technology is now capable of modelling large networks and handling enormous databases accessed concurrently by large numbers of users. One of the biggest deployments in the world is at the National Land Survey of Finland who have deployed 700 concurrent update GIS seats, in 44 sites distributed across Finland, accessing a single seamless GIS database in Helsinki containing 5 million land parcels managed in 7000 database versions, all with good performance.
To use an automotive analogy, the brakes have been removed; privatisation and deregulation have put the foot on the accelerator. Even the fuel is there; data availability and data conversion costs, are greatly improved compared to just a few years ago. So now, nothing should stop a major deployment of the technology. Many European, American and Asian Water, Gas, Electric and Communications companies have now achieved major deployments resulting in large benefits for their organisations. But just as in other parts of the software industry, what was achieved by bespoke development will in the fiture be achieved by suites of standard applications - Spatial Resource Planning applications. Some of these applications will be achieved by interfacing or embedding. Other applications will be achieved by a development on top of a GIS. Either way, the customer will be able to see what he is getting ahead of his project and this will reduce implementation costs and time-scales and lower risks (see Figure 2).
Figure 2: Reducing costs with improved approaches
There could be a merger of some niche areas, which were formerly addressed by separate companies, for example GIS and Outage Management. The process of Outage Management is addressed in the main by integrating a Customer System, Network Management System and SCADA (See Figure 3).
Figure 3: The Network-Premise-Customer Link
Simply put, this provides complete connectivity from a piece of equipment in the real world to a model of that equipment. Then, via a model of a connected network to a model of a customer premise, from the premise to a record of the person and hence to the real customer that depends on the equipment. This complete end-to-end connectivity is the fundamental basis of business processes that restore outages, plan outages, plan network maintenance and so on.
There are many benefits from addressing business processes by integrating application suites with GIS in an SRP framework giving a common operating environment:
-
A database with more detail
Development environment
System administration
Consistent user interface
Functionality
Conclusion
In moving from digital mapping to GIS, the focus was a mainly technical one to provide core technology which had adequate virility to do the job. Now that this phase is over, the focus moves from one of technical virility of the GIS to a different list, more relevant to satisfying the demand for deployable enterprise-wide processes:
So we will see the following trends:
-
From isolated to integrated system
From departmental to enterprise system
From operations to IT/MIS
From 2 tier to 3 tier architectures
From proprietary interfaces to CORBA and COM
From monolithic systems to components
From operational system to business process
From single vendor to partnerships
Figure 4: Increased benefit with move into SW applications
Spatial Resource Planning applications are a way of providing leveraged access to the network data by a much larger community of users. The benefit increases out of all proportion to the additional investment. If implementing an initial GIS may have seemed like pushing a boulder up a hill, the payoff from implementing SRP applications to exploit the database is rather like rolling down the other side.