FLOSS strategy for Pakistan
Posted: Mon Nov 10, 2003 8:14 pm
This is a paper brought to our attention by Mr. Frederick Noronha <firstname.lastname@example.org>, a journalist from Goa India and a veteran participant on the PLUC mailing list. It was presented on the Asia Opensource Symposium. A copy in PDF format is also available. A discussion of the paper either here and/or [url=mailto:email@example.com]the PLUC list[/url] will probably be very useful.
Open Source Strategy
Proposals for Pakistan
Asia Open Source Symposium
Advisor to the Ministry of IT &Telecommunications
Chief Operations Officer
With the rapid growth of IT and Telecommunications in Pakistan, there has been a growing awareness that it is imperative for Pakistan to come to grips with some key issues which relate to IPR, cost of deploying IT on a massive scale, growing the know-how in software technologies and integrating applications in the local languages.
The Ministry of IT has been very active in the last few years and to address this issue formed a Task Force for Linux which has been very active since (see www.tremu.gov.pk) has taken a few key decisions of which three are fundamental to the strategy:
1.In order to develop a clear direction, about US$ 1 Million (to grow to US$ 3 million) was provided in the annual budget of the Government budget this year to help fund development, research, coordinating centre, education and training in open source with an initial emphasis on Linux and Open Office deployment.
2.Officially mandating the use of Linux on all Government Servers - in force since 2 years.
3.In parallel negotiate with proprietary software vendors to get the correct pricing levels for Pakistan and use this as long as the alternate is not available, so that there is no disruption in the system. It is estimated that large vendors of software will overcome their 'reluctance' and will start providing support for OSS.
The programs and suggestions that are contained in this document could lead to jump starting a sustainable OSS/Linux movement in Pakistan. This is a preliminary document and will be modified as the knowledge and user base grows and the requisite evangelists, professionals and Organisations take ownership of the tasks.
This document has been drafted by three people who have been involved in this activity. It comprises of information about the current state and some suggestions for the way forward. This has been submitted to the Government of Pakistan and is still not an official policy document though many aspects of this have already been implemented.
We are at the starting line and have a lot to learn and emulate from our peers across the world.
The Asia OSS Forum objectives and Pakistan
Picking up from Asia OSS's agenda ...
- To encourage knowledge/data/software sharing
We need a central Database of Tried, Tested and Supported OSS software. Each country can pool resources together and Test new OSS software for ALL ASIA OSSes. These resources can be housed in places like non-profit Mimos of Malaysia and a similar outfit in Pakistan.
The objective of this team should be to bring out "stable" versions of BUSINESS SOFTWARE. The Server side is fine. The Kernel, Proxy, File Sharing and Mail software is well known and tested. What we need is ERPs and Small Accounting applications that can WORK in our environments.
A similar stance is needed for an Anti-virus database. We have Anti-virus software(the checking engine), BUT NOT a fully supported Database. Asia could pool resources and put together a Database. (Something that a PAKISTANI CERT can support). In view of the growing fear of the intelligence agencies of western governments planting Trojans into brand name US/western company based antivirus software, national security and personal privacy of countries is deemed to be at risk. This can be best addressed by such an open source initiative.
- To establish permanent infrastructure for OSS development and support as well as other related activities.
Asia OSS can form a Central BODY to become "online" support for all Asian Countries. Each Country can contribute a few resources to this body to help maintain and support OSS software. Support is the KEY to brining Linux to the mainstream.
This body can function like the support line of any large software vendor like IBM, etc. Registered users can login, write a support problem and get an auto generated ticket and wait for fixes/support or advice.
- To update status of OSS development from each of the participating countries.
We need assistance with KDE/Gnome Urdu. Significant progress has been made for Chinese and Japanese. India has a very advanced Tamil Version. We can collaborate with India to bring out an Urdu version - India has an Urdu version on it's agenda.
As regards to Open Office, it has the ability to save files in .doc (MS) format however there is about 20% loss in editing. A better conversion needs to be made to address the reluctance of the end users.
The Asian Open Source Community has to become the Central Development house for Open Office. We never know when SUN will stop churning out improvements to its Free Office applications!
Objective and Goals
Reduce Government and Private Sector dependence on proprietary / Licensed Technologies by providing Free-License alternatives, effectively improving Pakistan's position on IPR related issues. Reduce the costs of deployment of IT across Pakistan and develop a technology base for core software development in Pakistan.
Increase awareness of Free Licensed Technologies by conducting sponsored Seminars and workshops.
Investing in Human Resource Capital and enabling IT professionals to work with Free-Licensed technologies by conducting sponsored classroom training.
Jumpstart development of "solutions" based on Free-Licensed alternatives by funding viable projects through Educational Institutions and other Free-licensed based consulting firms.
In the long-term, provide confidence to the Local Industry by migrating a government organisation end-to-end with Open Source technologies and use the organisation as an example for future migrations.
Current state of Linux in Pakistan
Pakistani software industry has yet to fully realise the potential of using GNU/Linux based technology. Currently, GNU / Linux penetration is at its infancy.
Although a number of Educational Institutions have incorporated Linux based courses in their regular programs, these courses have very little or no impact - primarily because the courses are based on theory. Having said this, some notable institutions have recently excelled in the field - 2 students, one from LUMS and the other from Sir Syed won the worldwide IBM/Linux Challenge in November of 2002.
Another institution that has covered considerable ground work is PIMSAT Karachi. Well over 80% of Masters' final year students produced "prototype" commercial solutions based on Open Source Technologies from August 2002 to May 2003.
Private Sector Firms
Considerable private sector penetration can be found, primarily due to personal research and enthusiasm of IT Managers themselves - who have chosen to take a Free-Licensed path in order to steer clear of IPR issues. The majority of private sector firms still rely on Licensed proprietary software for their technology needs.
A few knowledgeable players have deployed Linux (Planning Commission of the Government of Pakistan - the heart of the planning of all projects for the Government) and some provincial governments. Most Servers are using Linux as mandated by the Federal Government directive.
InterNet Service Providers
The first to truly realise the benefits of GNU / Linux were the ISPs. Super Net and Zoom were amongst the first few ISPs, to successfully experiment with the technology. To date over 80% of ISPs use Open Source technologies for their infrastructure requirements.
Local IT / software houses
The local software industry is very small. Most medium to large corporations rely primarily on their internal IT staff to develop and deploy solutions. Very rarely does one find software contracts being outsourced to local IT vendors.
When projects are outsourced to the local industry, most of this work is given to the larger Software Houses, which inherently develop software on Licensed Based technologies - one simple reason for this is the availability of manpower. More so, proprietary software vendors have very strong marketing and sales teams and are able to influence projects at the time of inception. By the time tenders are floated, the underlying technology is already decided.
PLUC (Pakistan Linux Users Community) is a body set-up by Linux enthusiasts that is quite active on the Internet. They started off as a group of hobbyist and have now grown considerable in size. Unfortunately, PLUC members are still at it for the "love of the tool" and lack focus on business solutions. To the common man they appear to be very intelligent but extremely technical-jargon-oriented gurus.
ATRC (Applied Technology Research Centre)
A privately held research company, by a FAST graduate, who loves to toy with new technology. ATRC was meant to be a high-end research company. It found varied success in selling consultancy on Linux. Solutions for the ISP industry was its major focus, and ATRC was successful in being part of the team that set-up ZOOM Net, Cyber Access and InstaNet.
ATRC remains a high-end research company with an extremely technical CEO. ATRC has recently moved into small training and support for the local industry.
A privately held solutions company which produces applications based on Windows platform using some Open Source tools. PING has recently started pitching GNU/ Linux and has made some headway in web-application development.
Infotech, Infosys, Mushko and others
With the movement of Linux into the mainstream in the US, business partners of International Vendors have also started to provide Linux solutions. Little is known of their expertise in this area.
LMKR is one of the largest terra-byte companies in the world, which provides solutions to the Petroleum Engineering Services companies. LMKR houses one of the biggest Linux Clustered Super Computers, born out of its requirements to compute large amounts of data. LMKR as yet does not offer Linux expertise as part of its repertoire.
The only 100% Open Source Business Solutions Company in Pakistan, managed by one of the authors of this document.
Cable Net Operators, Cyber Cafe
BSA's fear had induced much of the local Cyber Caf� and Cable Net Operators to move to a Free-Licensed based operation. Unfortunately, most of the change over was done by Linux enthusiasts who had very little or no support to offer. Coupled with a lack of knowledge transfer training, these Cyber Caf� s and Cable Nets rapidly reverted to "cracked" Licensed based solutions. They continue to operate with unlicensed software.
Tremu kicked-off a Linux Task Force in August of 2002. People were invited to register their names on a website and were called to Islamabad for meetings to decide on a course of action. Tremu was considerably successful in being able to find volunteers willing to work to help solve IPR issues.
The initial Task Force was very enthusiastic and set-up several sub-committees to steer the GNU / Linux movement in Pakistan. Lacking coordination and leadership within the group of volunteers, the Task force was unable to come up with any concrete plans. Privately, Ants Consulting in collaboration with PIMSAT Karachi and TReMU came together to launch the first official GNU / Linux seminar which was held in Karachi, sponsored by IBM and executed by COMSATS.
Notable GNU/Linux achievements throughout Pakistan
Tremu / IBM Awareness Seminar
GNU / Linux Seminar -managed by TreMu, hosted by PIMSAT, sponsored by IBM with technical assistance from Ants Consulting - 5th April 2003.
CSP Karachi Seminars
Linux in a Nutshell, Ants Consulting - 21st June 2003.
IntraNet Development on Linux, PING Systems - 9th August 2003.
Worldwide IBM/Linux Challenge winners- Sir Syed and LUMS, IBM - November 2002.
IP/ Telephone - Pakistani company COMCEPT, embedded RTOS Linux Solution
IP table firewall product developed by a KU student - being used in UK
KESC - Payroll PIMSAT student
EOBF - Election System PIMSAT Student.
Linux Course Development
MAJU, Ants Consulting - July 2003
FAST NU Karachi, Ants Consulting - May 2003
PIMSAT, Ants Consulting - July 2002
PAF Kiet, PING Systems - June 2003
FAST NU, Karachi, Ants Consulting - June 2003
Mohammad Ali Jinnah University - Ants Consulting - July 2003
Stable, Secure, Fast, Reliable and requires minimal computing power.
Low-cost of maintenance and zero price - Perfectly suited for developing countries.
Huge, Stable competitors who have deep pockets for "free" awareness seminars, excellent marketing and a very strong local following.
Difficult to understand and implement. Dearth of books and educationalists alike.
Most of the knowledge is scattered over the NET.
Proprietary desktop software has a stronghold in homes and offices alike - GNU/Linux support is little and expertise is thinly spread.
IPR issue resolution.
BSA's threat in Pakistan works in Linux's favour.
Software giants are already beginning to share their code with University Students. This will further enable students to become more and more proprietary software savvy.
Current available Man Power
Incorporation of GNU / Linux based courses in degree programs is a very recent phenomenon, thus the availability of knowledgeable manpower is minimal. Further more, since most courses in Universities / Institutes are primarily based on theory, a very small amount of recent graduates are practically able to use their knowledge.
In the local industry, leaving apart the ISPs, there is an apparent lack of knowledge, confidence and support for GNU / Linux. Solution providers, continue to penetrate the local market with training and support solutions, but primarily due to the market's lack of knowledge/confidence in Linux, sales cycles are long and arduous.
ISPs are also not the place to look for when searching for business related Linux solutions. ISPs house support staff that have limited knowledge about installation and configuration of GNU/Linux Server or Desktops.
LPI.org certification, which happens to be the only distribution-independent Linux certification body, advocates a minimum of one-year practical use of Linux before enrolment into a 1-week certification preparation training session.
Having said this, there still exists a lot of talent that can be drawn upon.
Linux is almost another Unix, thus, Unix experts are the easiest to re-train for support.
If a concise effort is made, PLUC and similar organisations could turn out the required standard of professionals that can be used for MoiTTs Open Source future programs.
Some of the graduates that have used Linux in their degree course could also become worthy Linux professionals.
All in all, a guesstimate of professionals in Karachi number in hundreds. A similar trend exists in Lahore, Islamabad and other main cities of Pakistan.
Move Forward Plan
The local industry, educational institutions and software houses need to be simulated alike. In order to kick-off GNU / Linux, the following plans were suggested by the TreMu Linux Task Force:
Awareness Seminars and workshops should be conducted that address the local industries' needs. These seminars should not just be commercial sales pitches, but be practical knowledge sharing and confidence building workshops.
HEC and University Support
Educational Institutions should be provided with concrete course guidelines, via the HEC, to formally induct GNU / Linux within their curriculum.
Sponsored Desktop and Server Courses
Desktop is the key to solving the IPR issue. Whereas Linux is traditionally a Server level performance tool, quite recently the desktop has become user friendly enough to use. Unfortunately, be it, the server or the desktop, the installation procedures remains equally cumbersome.
In order to overcome installation problems, sponsored courses should be conducted for Desktop and Servers alike. It would be advisable to have private sector firms, educational institutions and government institutions nominate 2 IT professionals each to attend these courses.
Purely from experience, 8 hours of Desktop(1 day) and 16 hours of Server (2 days) courses, coupled with Resource Centre Support, should be adequate for IT professionals to effectively migrate their organisations' non-proprietary based infrastructure/solutions. In some cases, re-learning of Networking technologies and hardware selection knowledge transfer could extend the Server course by an additional day taking the course to a total of 3 days.
If MoiTT plans for about 5,000 desktop courses and a similar number for server courses in Karachi alone, then manpower requirements to train these can be estimated to be:
1 class has 10 students
5000 candidates / 10 = 500 classes.
500 classes / 25 trainers(institutes) = 20 working days
1 class has 10 students
5000 candidates / 10 = 500 classes.
500 classes x 3 days per course = 1500 working days
1500 / 25 trainers(institutes) = 60 working days
In order to ensure that the courses are taught equally well across the city, MoiTT will have to invest in a train-the-trainer exercise.
In Karachi for example, MoiTT could select, via testing, and invest in train-the-trainer sessions for 50 desktop and 50 server trainers, and select the best 25 for these training courses.
To impart this training, Linux Ready institutes will be required.
An important consideration: Currently Linux knowledge is mostly held with very techno-savvy geeks, and most of the trainers that may emerge from the train-the-trainer sessions may have their reservations about teaching in certain institutes. Also, being techno-savvy, they may NOT be the best trainers and may lack negotiating skills to secure fair remuneration for their training efforts.
It is recommended that Institutes are awarded "separate" fees for conducting Linux Classes whereas trainers are given their own share, completely disjointed from the institutes income. In all fairness, the Lion's share should be given to the trainer.
Linux Course and Certification Standards - LPI.org
Linux Professional Institute (LPI.org) is a non-profit group of individuals from Canada who run a Linux Certification "Distribution Independent" program. They have offices in various countries all working on a non-profit basis.
To ensure the quality of sponsored Desktop and Server courses in Pakistan, MoiTT can look into engaging LPI.org to assist in the course development and certification thereof. LPI.org is distribution-independent and non-profit.
The trainers and the trained could be given a 6 month time period to get themselves LPI certified. MoiTT can look into refunding LPI certification costs, provided a candidate passes.
Linux Resource Centre
To assist the local industry, students, and professionals alike, a resource centre for Linux help will be required. The primary responsibilities of this resource centre will be to assist computer literate masses via telephone, Instant Messaging, Website forum and face-to-face meetings (if necessary). This is planned to be funded by the Government in the current year. It will also be a clearing house for housing the latest releases, drivers and other support materials.
Grants to Universities
Universities will be offered grants to develop Linux LABs and research centres. Linux is a very low resource requirement technology and existing LABs at Universities should house adequate "old" hardware to supplement a Linux Lab. Research on Linux, at this juncture, seems a bit premature, as we have yet to fully comprehend the technology on a large enough scale to steer research in the right direction. Perhaps in the coming years, our Universities and students alike may be competent enough to develop a Pakistani based distribution. The need for one is not essential.
Other areas of focus for Universities can be Industrial automation and business solutions. Taking examples of the US, Universities could cater to local business requirements, and in turn, local businesses should reciprocate with funding. To break the chicken-egg deadlock, MoiTT can "invest" in viable projects, for example:
Pakistani Textile units do not, as yet, have real-time data capturing ability from their spinning and weaving machines. Small electronic PCBs, coupled with software (perhaps Linux based) need to be developed.
Defence forces have large-scale automation requirements. Some of the non-sensitive information needs can be fulfilled by local Universities. Linux is also a real-time operating system of choice. The solutions for defence could be enormous.
The local cotton, leather and engineering units require artificially intelligent applications that can measure, scale and plan.
Securities and Exchange commission requires real-time web-applications.
There are many more projects that can be provided as examples, the justification should be to "solve" a problem that someone will be willing to pay for.
Whereas Educational Institutes can be coerced into running courses for students, and awareness seminars and sponsored courses will help in improving expertise/knowledge of GNU / Linux, MoiTT would still require to develop a strategy such that this investment in Open Source sets in motion a self-sustained growth of Open Source Development.
In order to achieve sustenance and also longer-term goals, MoiTT needs to simulate the local IT industry.
One suggestion is to supplement all the above efforts with announcing tenders for some "future" date for Linux based Business Solutions.
E.g. June 2004 Tenders could be:
1 - Rs. 5,000,000/- for Karachi University Automation
2 - Rs. 2,000,000/- for sponsored ERP for a private company
3 - Rs 1,000,000/- for PIA online booking application (see emirates.com)
Tender requirements could clearly state that the company that is awarded these tender(s) houses at least 5 LPI certified (or equivalent) professionals on their payroll.
To ensure that educational institutions remain on-track, a home-grown Linux challenge (perhaps at AQ Khan), like that of IBM, with cash prizes, will help provide incentives to students.
Another suggestion is to make the A. Q. Khan LABs challenge theme Linux for the next three years.
Over and above all this is a requirement to standardise, steer and manage all these efforts. Thus, a Pakistan Linux Steering Group is recommended.
Pakistan Linux Steering Group
It is recommended that MoiTT formally engage part-time Linux business consultant(s) to steer GNU / Linux in Pakistan. This Group of consultants can be engaged on contract basis for a period of 1 - 3 year(s) and be made responsible for realisation of the objectives laid out in this document.
It is proposed that this group consist of 3 experts (one from each major city - KHI, ISL and LHR) and report to an authority based in Islamabad.
With PLSG MoiTT can ensure it is not just spending money for the sake of spending, but also has control over monitoring these expenses and ensuring that the right results are being sought.
PLSG can be used to realise this document into a concrete move forward plan that can help MoiTT and Pakistan in the years to come.
Open Source Strategy Road Map for Pakistan
Note: This part of the document has subjective references like 'we..' since this has been written by the main players of the Open Source community as a part of their recommendations by key members of the Task Force on Linux of TReMU
The Case for Open Source
Open Source is:
?about creating a software platform which brings down costs of technology by a factor of 10, thus making it affordable for consumers and enterprises in the world's emerging markets.
?going to become the computing platform for the next 500 million consumers and the world's 25 million SMEs who have not been able to adopt technology because of its dollar-denominated high pricing.
?.targeted at the world's emerging markets, because they are where technology has not yet penetrated deeply, and yet, for whom, technology offers perhaps the last opportunity to better integrate into the world's value chain and improve the standard of living for their people.
?.a solution which can thus bring down the total cost of ownership of technology (hardware, software, training, support) to no more than USD 15-20 per person-month. More importantly, from an emerging market perspective, it ensures that most of the IT spend is recycled among local companies, thus providing a benefit to the domestic IT industry, especially the independent software vendors (ISVs).
The Opensource architecture provides 4 key benefits, which are unmatched in any existing solution in today's marketplace:
1. Brings down the cost of hardware on the desktop to USD 125-150
2. Brings down the cost of software on the desktop to USD 5-10 per person.
3. Creates a solution that is easy to manage and scale through server-based computing
4. Integrates with the Windows world, by supporting proprietary MS Office file formats, networking and executing many applications.
The key components that comprise Opensource:
Enterprise Server side: Thick Server
Enterprise Client side: Thin Client and Digital Dashboard
Enterprise Applications: eBusiness Suite and RosettaNet Basics
Thin Client-Thick Server
This is the Linux-based platform which can bring down computing costs dramatically by leveraging older computers (anything from 486 machines with 16 MB RAM will work) and combining it with a Linux-based desktop and set of open source applications (Evolution for email and calendaring, Mozilla for browsing, OpenOffice for the productivity applications ' word processing, spreadsheet and presentation, GAIM for instant messaging, providing a single-window login to Yahoo, MSN, AOL, ICQ and Jabber).
What is different about Thin Clients this time around' After all, they've been talked about ever since computing began.
The major difference is the re-use of older hardware. We use older cars, older manufacturing plants, older homes, but we don't tend to use older computers.
The price points of USD 700-900 for a new desktop and USD 500 for MS Windows and Office may be fine in countries like US, UK, Germany and Japan, but it pinches a lot (and is almost unaffordable for the masses) in countries like Pakistan, Indonesia, India, China, Brazil and Mexico. For these low-income markets, hardware and software need to cost a tenth to bring them in line with what their people are earning for technology to go mass market and become a utility. Older PCs can now be leveraged as Thin Clients, without sacrificing performance or the desktop look or the applications.
The world's developed markets have been saturated with technology. New PC sales now imply upgrades, creating a huge supply of older computers. The PCs that are being still have a lot of life left in them ' after all, they are no more than a few years old. These PCs, which incidentally have become an environmental hazard in the developed markets, can now be shipped to the emerging markets where the Opensource platform brings them to life once again on the desktops and in the homes of people who may never have tasted computing. These PCs can be available for USD 100 or so in large numbers or more USD 125-150 in smaller quantities. Add USD 50 if one only uses the old motherboard and goes in for a new keyboard, mouse and monitor.
This happens because of the use of Linux in a server-computing mode. All the applications are run off the server, with only the display happening on the client side. In that sense, the desktop becomes a Terminal. The difference this time around is that the Linux-based Thin Client has all the key applications that the majority of people need to use (email, browser, instant messaging, word processing, spreadsheet, presentation) and there is no performance penalty even though the applications run off the server. The most recent releases of the open source versions of the applications is now more than 'good enough' ' both from the point of view of stability and Windows compatibility.
Server-based computing using Linux (and built on the X protocol) is now possible because LAN speeds have gone up to 100 Mbps enabling the transfer of a lot more data over the same network. The result is that a Thick Server (which is actually a new desktop with 1 GB RAM and 2 hard disks in a software RAID configuration) can easily support 30-40 users. Such a Server would cost about USD 1,500-2,000, implying a per client cost of no more than USD 50.
Taken together, the Thin Client and Thick Server combination not only brings down the cost of both hardware and software by 90%, but also provides the IT manager complete control of the client desktop from the server. What every user sees on their Thin Client can be standardized and controlled from the Thick Server itself. In addition, the use of Linux does away with all the virus-related problems. Server-based computing also centralises management of data (backup and restore, for example).
Way forward -
The Thin Client-Thick Server solution is ideal for three environments:
1. Where there is software piracy (illegal copies of Microsoft Windows and Office). At one stroke, all users can now be given legal compatible software for a small incremental cost (10% or less of what MS Windows and Office would cost). Considering that Microsoft and the Business Software Alliance have been targeting businesses in emerging markets, the Opensource architecture provides an alternative which is not disruptive and can be implemented very rapidly.
2. Where there is an abundance of older PCs and the upgrade costs are proving to be prohibitive. This is where the same older PCs can now be converted into Thin Clients, thus protecting the investment that has already been made.
3. Where computing is not present. This means providing a computer on the desktop of every person in the organisation. Consider the economics: the Thin Client-Thick Server solution costs about USD 15-20 per person-month. This means that if a person earning USD 150-200 per month can become 10% more productive using a computer, the investment pays back from itself from day one! The computer is perhaps the most 'transformative' invention in the past century and the defining device for today's Information Age. It has significantly improved productivity wherever it has been utilised. The emerging markets of the world and the bottom of the enterprise pyramid are where computers have not yet penetrated because of their costs. The Thin Client-Thick Server solution now makes it possible to level the technology playing field for the SMEs in the emerging markets.
The limitation of course is that this solution will not support native Windows applications. There are some alternatives:
1. Try out Wine and other solutions on Linux (Win4Lin, Crossover Office) to see if they will support Windows applications.
2. See if the client application can work in a browser; the back-end can be Windows 2000 or anything else.
3. Use 'vncviewer' to provide a Windows desktop (on a Linux Thin Client) grabbed from a Windows machine (only one user at a time).
4. Use the Windows Terminal Server to provide multiple users access to Windows on the Thin Client.
This solution will co-exist with Windows. The approach should be to look at whom we can delight rather than whom we may disappoint.
The Thin Client-Thick Server as envisioned here leverages the R&D that has given in the world. No solution for the bottom of the pyramid can be cost-effective if it involves significant R&D efforts (and the time associated with it), especially in technology where the research costs are huge. By lagging hardware technology by a few years but creating software using the latest modules and standards, we create a good and potent mix. In fact, the operative phrase for much of what Opensource is about is 'value-added aggregation'. The innovation needs to be less in new technology, but in how one can best leverage the existing technologies which exist.
Enterprise Software Applications
Enterprise Software is what every business needs, and yet few can afford it. The penetration of ERP and CRM applications have so far been limited to perhaps the top 10,000 enterprises in the world. Of course, that's not bad ' because these enterprises have the millions that the enterprise software vendors and the consultants charge for putting the solution together. The enterprise software applications capture the business processes and embed the information. The problem is that different enterprise apps have their own formats and models, effectively creating silos of information.
In the world that we are targeting, most enterprises will have just 1-2 applications, with accounting being the most common. Other information may be on the computer, but is likely to be in Foxpro/Microsoft-SQL Server databases or even in MS-Excel.
What we want to do here is to make available the business software applications that the enterprises need for no more than USD 5-10 or so per person-month. Compare this to the hundreds or thousands of dollars that are paid per seat for enterprise applications today. Our price point will make the apps available to the bottom of the enterprise pyramid and if we are smart (and lucky), we may be able to move up the ladder (but that is not as important).
We will begin by writing adaptors to standard applications that exist and by defining a common business process reference model which encompasses the key processes, flows and numbers for the different types of businesses (manufacturing, trading, services, etc.). This is again an 'edge' strategy: we are not going and saying we want to replace the application right away. What we are saying is that we can combine data from different applications to provide you reports and views of the business you may not have seen before. In defining these 'APIs', we need to leverage two sets of standards: web services (usage of XML, CORBA, and SOAP) and business processes.
The next step is to start putting together the actual application building blocks evolving from the enterprise models that we have created and tailored to fit the business process standards. The Enterprise Model we need to build must be event-driven. We need to think in terms of a (near) real-time enterprise. We should look at the messaging frameworks and application integration being done currently.
Our enterprise objects would perhaps have 70-80% of the functionality. These are the 'enterprise Lego blocks' which could be assembled by Independent Software Vendors (ISVs) for a customised solution. ISVs will play an important role because they have the 'last-mile' enterprise/industry knowledge and the customer relationship. We should publish the APIs of all our objects so if others build a better object, so be it. This is what Microsoft has done well ' leverage the developer community. Our success to a great extent here will also depend on whether we can get ISVs to extend (and replace) our enterprise objects.
Alternately, the enterprise itself could choose to take an existing set of processes from an online library which exists and then customise it. The latter needs the equivalent of a scripting-base workflow programming environment which does for business processes what Visual Basic has done for software. We need to make this business rules driven.
In parallel, we need to use inter-enterprise standards to streamline document flow based on standards between enterprises.
Our approach here should be like that of a TV serial maker rather than a film-maker: we do not want to wait for 3 years in the hope of a blockbuster. Rather, we want to come out with new objects every week, so we can do course correction any time, if required.
The integrated set of enterprise applications will leverage a common database, wth storage in XML. They will also follow the OHIO principle: only handle information once. Data should not be replicated for storage. It is possible for us to think like this because we are building the base set of objects from scratch. We will use open source to make this happen with EJB/J2EE as the development environment, JBoss as the application server and PostgreSQL as the backend database. We should also see how to integrate OpenOffice Calc ' it is much more than a spreadsheet. We can think of it as a computational engine, with excellent reporting and graphing/charting capabilities.
The Enterprise Applications thus become like the toppings on the pizza-cheese base. Different segments may want different toppings. Either we can provide the toppings (or the recipes) or others can do their own.
What Opensource is Missing
There are many things which we will need to look at over time to complete the picture. I have just outlined some of them here, so we do not ignore them down the line:
- Communications / Connectivity: look at WiFi
- Software for Project Management, Content Management
- Website Publishing with eCommerce support for small businesses
- Content and Community: an Enterprise 'Reader's Digest' meets Slashdot to create clusters of SMEs
- Customisation for different verticals (eg. Schools, Homes) ' what will need to be done differently
- SME Marketplace ' like what eBay has created
- MicroFinancing ' so that SMEs or even individuals can invest in these technologies
- SME 24 hour tech shops in neighbourhoods which can also serve as cybercafes, demo centres, meeting points, payment collection centres and wireless hubs
- Custom Thin Clients, which use cheaper chips to create Thin Clients for USD 50 or so.
- Optimising the Thin Client-Thick Server protocols in two steps: first, to bring down the LAN speed requirement to 10 Mbps so it will work over 802.11 (wireless) networks, and then bring it down to 1 Mbps so it can work on cable, thus enabling us to target the Home segment.
- Collaborative Workspaces across firewalls
Perhaps the biggest thing missing is how we will 'go to market' with Opensource ' how will be brand, price, distribute, sell and support this not just in Pakistan, but in other markets. I don't have many answers right now, but will do so soon. This is where we will need to experiment with different ideas in the next few months.
The Mail Server: A Starting Point
The Linux-based Messaging Server provides support for Instant Messaging (via Jabber), Proxy (via Squid), a basic Firewall, Global Address Book (supporting LDAP), and integrated anti-virus screening.
The Thin Client-Thick Server, and the Enterprise Applications strategies create the foundation for a new IT infrastructure based on Linux and Open Source for SMEs. The solution cost will be a fraction of what the big players charge, but in functionality we must not be far behind. A similar strategy has been used by Huawei in China to take on the telecom majors like Cisco, Lucent and Nortel ' it provides perhaps half the functionality but at a fifth of the price.
There's a lot that we have to do. The vision is quite audacious and large in its scope. But many things in technology are now coming together to dramatically change computing in the next decade. The Internet was one such hurricane which brought the world's computers closer. And yet, a large part of the world is still relatively untouched by technology, largely because it is driven by US companies and dollar-based pricing. This 'other world' is the one that now needs to be impacted by technology. I think of Opensource as the next hurricane which will brings this world's people and enterprises closer.
Currently people are worried about software licencing costs, there is a rapidly growing interest in Linux, and there are plenty of old PCs around which could be usefully employed. At least, there is now an awareness that there is an alternative to the MS-Windows/Office combine.
What needs to be done is :
In corporates we need support for various Windows applications and a mechanism to counter user resistance to change.
In schools, we need support for local languages.
In colleges, we need support for an integrated development environment.
In homes, we need to support lower connectivity speeds and multimedia.
In government, the issues are Linux, the lack of engineers aware of Linux, after-sales support and end-user training.
In remote offices, there is a need for a single desktop.
Basically Linux and Opensource offers a 75% savings in the long run.
However to achieve those savings then resistance to change and fear of the unknown needs to be overcome.
The fastest way to overcome these two fears includes two basic things.
To see the real ease of Opensource you need to get two people who nothing about computers to install and use commercial and opensource software.
This shall prove directly that opensource has reached a point where it is easier for the totally new user.
For the users with more than 1 year or more of commercial software experience, we need to show them a running system and request them to use it for at least three days. This shall make them comfortable with the applications and build their confidence that open source actually does what they get done with commercial software along with reducing their fear of regular crashes and virus worries.
The best areas to start Linux and Opensource immediately are the areas where computers do not exist. Once the new users in the government start with Linux they shall immediately feel the problems and limitations which they need to worry about when they accidentally use someone else's windows based computer.
The value in using opensource are as follows :
Elimination of Virus risk: This came out as a big deal in the small and medium-sized companies. Virus is the biggest threat and pain-point for them.
Reduction in cost of Hardware: This happens through (a) re-use of existing PCs (b) lower cost of new hardware (c) lower maintenance costs for hardware (d) never need to upgrade hardware.
Licencing costs for software come down dramatically: This is true for the ones who are prepared (or forced) to make the necessary investments in legal products.
Greater security: This is because of (a) Linux, which is considered more secure than Windows (b) privacy for each user's data since it can only be accessed via a login and password (c) not having floppies and hard disks on the client side reduces the misuse of sensitive information.
Centralised management and administration: What is appealing is the ease of doing backups, adding new users and being able to even manage remote locations from the central office.
So, the one line summary for opensource is :
"99% Windows-like, 0% Viruses, 25% Cost"
Let us also consider the issues we have faced so far (and there are many), along with solution ideas. First a look at the technology issues:
There is some support for DOS and Windows applications:
We can get DOS applications to run on the DOS emulator. There are still some issues with printing and memory usage we need to get resolved. But in general, to run Windows applications, we will need to use one of two options: Win4Lin, which costs an additional USD 100, or rDesktop with a (separate) Windows Terminal Server. The latter option will run everything, but needs some additional investment along with a separate Windows server. Of course, taken together, the costs will still be much lower than a pure Windows-Office thick desktop alternative. What we have seen is that almost every organisation (especially in the corporate sector) has some proprietary Windows application which needs to be supported. These applications can still be run at a lower cost if the opensource along with emulation is applied.
MS-Office file formats discrepancy:
The impact of this is more likely to be only on power users. Those users who use macros. The ordinary users (99%) do not notice any difference between OpenOffice and MS Office. OpenOffice is able to read and write files of MS Office, StarOffice, and OpenOffice.
Single point of failure on the server: Even though we recommend a dual CPU system with two hard disks (and software RAID enabled on them), the fact of the matter remains that if the motherboard fails, the system is completely down. With PCs, a single PC going down does not make a difference. This is the difference between a centralised approach (ours) and a decentralised one (with PCs). The solution: have a two-server cluster with automatic rollover. Of course, this will increase costs.
For the government and especially public schools in Pakistan, the need to support local languages is very important. The opensource community has done some work in Urdu Local language development, however work needs to be done in at least ten or more local languages so that computing becomes easy to understand. The commercial software industry has done almost nothing in this respect except state that they shall do something in the future.
Migration of existing users and their data: Emails on local disks in Outlook or Outlook Express, Address books, bookmarks and cookies all need to be migrated to the Opensource platform. Plenty of tools are available to automate much of this migration.
Among some of the other issues:
Users need to try out the platform to see that is serves them almost as well as far as windows is concerned and in some cases it is better. To get them to try it we need a special desktop killer application designed only for Linux which provides them some functionality that is not available on the windows platform. There are many technical softwares which lure the extreme power users to the platform already but not for the common person.
There is a mindblock that many people have in using older PCs. Old equals bad in computers. They want to use the latest, even when this is not really necessary. There are alternatives where the new PCs are costing much less. We could even look at diskless terminals. Linux allows PCs costing Rs 2,500 each to be used as a fully functional workstation.
We should try and make the desktop as Windows-like as possible. Maybe the background colour should be made green instead of blue. The perception again is that Windows equals user-friendly. So, to begin with, perhaps we need to just make our desktop look as close to Windows as possible.
It is hard to say if Rs 2,500 per Thin Client is the right price or not. One of the issues we are facing is that there is substantial work that needs to be done for the installation and migration of user data. For the channel, even a 20% margin and a 20-user set-up will only result in a Rs 10,000 revenue, which will probably not be enough to compensate for the work being put in. The second issue is with pricing from the second year onwards: do we continue to charge Rs 2,500 or should it be lower or nothing at all.
Companies and hardware retailers need to be encouraged to provide onsite support just like they do currently with the commercial products. They shall not be worried because they do not sell a lot of windows licences anyways. They would actually be happy because they would not have to worry about the BSA raiding their shop for those installer CDs which they need for installing cracked copies of XP, and Windows 2003.
We need to create training programs on Linux and Opensource.
How do we make people aware that an alternative exists'
The government can set aside a certain budget to advertise on billboards or newspapers especially the business newspapers regarding the opensource initiative.
Once commercial companies start to inquire about the support provided by the hardware retailers, then you shall see the revolution start by seeing Linux and windows both installed as the default on all locally assembled computers. So you shall get the option to boot linux or windows by default even on the computers bought for the home.
There is still a fear of Linux ' it is like the fear of the unknown. Also, there are few people in the market who know or understand Linux well, unlike the availability of Windows knowledge at almost every street corner. This needs to be solved through advertising and training programs.
We need time with the users. An overnight shift is not easy and very wrenching. People must be made to feel comfortable with the solution. It takes a few days to get used to the solution. Perhaps, we need to think of an intermediate path where we put Mozilla and OpenOffice on Windows and then switch Windows to Linux later. This way, the two key applications will not change and it takes the migration process easier and less noticeable.
Lack of an ecosystem:
Microsoft and its partners are entrenched across the value chain. Right from support engineers to Independent Software Vendors (ISVs), from books to training institutes, a person knowledgeable in Microsoft-products is never too far away. The same cannot be said of Linux. Even though we may underplay the Linux part of the solution, the fact remains that we are 'not Microsoft'. Microsoft has made deep inroads into the entire system ' right from being present in school curricula to being pre-installed on desktops to having a huge array of applications and developers continuing to strengthen the chain. It could take a few years for the change to take place.
However the good news is that due to the worldwide demand on IT to reduce spending, increase security, boost reliability and increasing performance simultaneously, the open source revolution shall take place from both inside and outside Pakistan. It is up to us to decide whether we are transforming faster or slower than the international rate of change. A current example is the German city of Munich which decided to replace microsoft completely. We could try the same by starting with a city and then use the lessons learned from the city to do other cities.
We need user and admin guides. In addition, as part of the brochures, we should also explain the philosophy of our approach (server-based computing, open-source) and why it is better than what is out there today.
Microsoft Front Page Extensions for Apache were provided many years ago because most ISPs refused to run NT servers with IIS due to security reasons. By making M$ run their software on Linux today you save at least $100 per machine by not having to pay for the operating system just because you need to run a program on your hardware.
The opportunity is there and the time is right: No doubt about this. The solution touches a chord. It excites. It taps into an anti-incumbency feeling against the existing solution. We somehow need to take that and capitalise on it.
Don't be too radical, try incrementalism:
Let us first try and edge into the organization with a thick server and Linux-boot floppies for Windows machines. This way, they can at least try out the solution. They need a compelling reason to be made to boot under Linux ' perhaps, this could be for virus-free email and faster browsing.
Co-exist with Windows:
This is the point made about being incrementalist. Take a staged approach to shifting people: (a) Use Linux for email, (b) Use OpenOffice and Mozilla for Windows, and then finally replace Windows with Linux. The first two steps ensure that for the key applications they are not dependent on a Windows-OS.
Need for a whole solution:
Maybe we need to take on the responsibility for the whole solution, hardware included. Also, we need something extra for each of the verticals that we are targeting. They need a more complete solution that the base that is on offer today.
Build an ecosystem:
This is very critical. We need to create a syndicate, a consortium of like-minded people and companies. It is not something we can plan and win alone. When dealing with large entities we need to be united.
Identify pro-Linux people and companies:
We need to be the core around which others can coalesce. We need to become the rallying point. There needs to be a directory of such entities, organized by skills and location.
Need for a Linux Desktop:
We also need a single PC (thick client) solution. This means having a Linux Desktop on a standalone machine. We could still use Opensource here, so that in the event of the addition of more computers, they can now just be thin clients. [In fact, this exists even now with the solution - the server and client run on the same system. The good thing about this is that whenever an additional computer needs to be added, it can be a thin client.]
Build an applications base on Linux:
There are a lot of applications available in open-source. We need to get them all together. Look at what is available for schools, engineering, etc. and package them together. Even locally, we need to get various applications running on Linux. Is there an easy way to port VB applications to Linux' If this was needed, who can do it'
Look for niches:
We need footholds to begin the battle. Which are the places we can get success first' Who can be our early adopters' We need to create islands of successes, which later can be joined together.
Even as we make the solution work in Pakistan, we need to aggressively take it to other countries who may be much more receptive.
This is a marathon:
They are no easy pickings. We must be able to sustain the morale and enthusiasm for many years. It is like hammering against a wall. We do not know when it will collapse, but the hammering must continue.
So, there is a 4-pronged approach we need to take:
For organizatons, we need to focus on 3 key decision points wherein we have an opportunity: when they are buying new computers, or planning to new licences for desktop software (OpenOffice and Linux), or planning to dispose off their older computers. A fourth point which is always applicable is the fear of viruses.
We need to focus on building out the channel. This will give us scalability. The channels are what will give us the reach. We have a solution that can dramatically increase margins for channels, with their existing customer base.
We need to look at the education market much more closely - schools, colleges and training institutes. In Pakistan alone, there are thousands of schools. For this, we need a whole solution - with the course material, and support for local languages. What I like about this segment is that we can truly delight our end-users.
Start an Intro to Computers course in regional languages using the platform as a base. Once the course is thru, the people will be ready to use the system and can continue the monthly payments for the use. I strongly suggest regional languages though -- there seems to be a market gap there. Even with schools, you would be better off in the regional language schools, starting early. But make it a commercial course that you would charge for.
The public sector is viewed as a big spender on IT, and as much, will be a big beneficiary. Public sector units, banks, post offices, village centres, railway stations, petrol pumps - all are places where we can set up computer centers with 10-40 computer each.
The chances for success will be greatest where we can deploy clusters of computers in an enclosed area so that we do not have to worry about networking issues. These computers will be used in a shared access manner, and will therefore support a much larger user base. We need to find "chains" of such clusters, so our costs come down.
A 10-computer centre can be set-up for less than Rs 200,000 and can serve 100 people or more, giving a per person cost of less than Rs 2,000. A 40-computer centre cn be set-up for less than Rs 500,000 and can serve 500 people or more, reducing the per person computing cost to less than Rs 1,000.
The greatest strength of Opensource lies in the fact that it can serve as the base for a whole new class of "open 24 hours" - computing and communication centres, which can make technology a utility for the mass market.