Without entering into further details, how this could happen, and why … It’s true that something wrong happened and they assumed the consequences. I’m not talking about the small amount of money they paid to get back the password, but about the scam behind it and the name of the university in question.

I wanted to profit – as usual – and focus again on the strategic decisions of considering Open source as alternative in critical mission projects. Whatever if you are talking about ERP, CRM, BPM, DMS, … the open source alternative is always available and costless compared to proprietary solutions. Best of all, open source is a guarantee to have full control on the business and get lock-free solution.

Imagine if this project was made in an open source ecosystem, the university will never get locks anywhere, and their system might be updated easily and quickly – even if something wrong happened.

Most important things to consider to avoid locks :

• Adopt Open Source technologies
• Document everything : Business, communications, code and databases
• Adopt a backup/restore strategy
• Adopt a disaster recovery strategy
• Share the knowledge in a development team, and avoid using only one developer in a large project to minimize cost. Two or more developers will get work done faster and safer.
• Separate critical mission projects, from new alpha/beta projects. Use SOA as much as possible for integration.
• Do not go e-Business if you are not ready for it and always provide alternative in case of failure. And consider seriously that the project will fail.
• finally -in Saudi Arabia specifically- ask for support, Yesser program is doing great work and their consultants might really help to secure your e-Business.

You still don’t trust open source ?

Related posts:

1. Open Source, Closed Source, We are not enemies
2. Open Source Evangelism

Note: Before beginning I want to mention that master/master configuration is great to avoid single point of failure in your system architecture, however it will not distribute load across the nodes. I use Mysql specifically for users management and permissions and then distribute load via my web application.

Ubuntu Server install ext4 by default, which you will need to change first to ext3, according to Lucid/Lynx release notes, if you have performance-sensitive applications – the case with a Mysql database server.

We will consider two nodes node1 and node2 with respectively 10.10.0.1 and 10.10.0.2

First install MySQL 5.1 in the two nodes :

hatem@node1$ sudo apt-get install mysql-server mysql-client

You will have to set the MySQL root password during installation.

Then we have to configure the first master node1

hatem@node1$ sudo vi /etc/mysql/my.cnf

comment out the line bind-address, then set a unique value to server-id, and the masterdbname you want to replicate.

[mysqld]
# bind-address = 127.0.0.1
server-id = 1
log_bin = /var/log/mysql/mysql-bin.log
expire_logs_days = 10
max_binlog_size = 100M
binlog_do_db = masterdbname
binlog_ignore_db = mysql
binlog_ignore_db = test

Restart database to affect changes :

hatem@node1$ sudo /etc/init.d/mysql restart

If you have a database dump you can import it in node1, however you can just create the database masterdbname manually :

hatem@node1$ mysql -u root -p
mysql> CREATE DATABASE masterdbname;

We will have to grant a username and password replication permission on the node1 :

mysql> GRANT REPLICATION SLAVE ON *.* TO 'user'@'%' IDENTIFIED BY 'password';
mysql> FLUSH PRIVILEGES;
mysql> show master status;
+------------------+----------+--------------+------------------+
| File | Position | Binlog_Do_DB | Binlog_Ignore_DB |
+------------------+----------+--------------+------------------+
| mysql-bin.000034 | 443 | masterdbname | mysql,test |
+------------------+----------+--------------+------------------+
1 row in set (0.00 sec)
mysql> quit

Now we can move to node2 and configure the mysql server :

hatem@node2$ sudo vi /etc/mysql/my.cnf

We will have the similar config here too, but with a different server-id

[mysqld]
# bind-address = 127.0.0.1
server-id = 2
log_bin = /var/log/mysql/mysql-bin.log
expire_logs_days = 10
max_binlog_size = 100M
binlog_do_db = masterdbname
binlog_ignore_db = mysql
binlog_ignore_db = test

Restart mysql

hatem@node2$ sudo /etc/init.d/mysql restart

Finally we can connect to node2 server and setup the first replication :

hatem@node2$ mysql -u root -p
mysql> CREATE DATABASE masterdbname;
mysql> CHANGE MASTER TO MASTER_HOST='10.10.0.1', MASTER_USER='user', MASTER_PASSWORD='password', MASTER_LOG_FILE='mysql-bin.000034', MASTER_LOG_POS=443;

Notice that MASTER_LOG_FILE and MASTER_LOG_POS values are from previous show master status result.

The first replication is done we can start the slave :

mysql> START SLAVE;
mysql> show slave status \G
*************************** 1. row ***************************
Slave_IO_State: Waiting for master to send event
Master_Host: 10.10.0.1
Master_User: user
Master_Port: 3306
Connect_Retry: 60
Master_Log_File: mysql-bin.000034
Read_Master_Log_Pos: 261
Relay_Log_File: Node2-relay-bin.000002
Relay_Log_Pos: 406
Relay_Master_Log_File: mysql-bin.000034
Slave_IO_Running: Yes
Slave_SQL_Running: Yes

If the Slave_IO_Running and Slave_SQL_Running are Yes it mean your first replication is done and you can create a simple table :

At Node1 (actual master) create a table :

mysql> use masterdbname;
mysql> create table testmaster21 (mid int(11) auto_increment, PRIMARY KEY (mid)) Engine=MyISAM;

Check available tables at node2 (actual slave) … Tada :

mysql> show tables;
+------------------------+
| Tables_in_masterdbname |
+------------------------+
| testmaster21 |
+------------------------+
1 rows in set (0.00 sec)

Now the second part is very easy, we need first to check master details in node2 :

mysql> show master status;
+------------------+----------+--------------+------------------+
| File | Position | Binlog_Do_DB | Binlog_Ignore_DB |
+------------------+----------+--------------+------------------+
| mysql-bin.000034 | 261 | masterdbname | mysql,test |
+------------------+----------+--------------+------------------+
1 row in set (0.00 sec)

Then add in node1 :

mysql> CHANGE MASTER TO MASTER_HOST='10.10.0.2', MASTER_USER='user', MASTER_PASSWORD='password', MASTER_LOG_FILE='mysql-bin.000034', MASTER_LOG_POS=261;
mysql> START SLAVE;
mysql> show slave status \G
*************************** 1. row ***************************
Slave_IO_State: Waiting for master to send event
Master_Host: 10.10.0.2
Master_User: user
Master_User: user
Master_Port: 3306
Connect_Retry: 60
Master_Log_File: mysql-bin.000034
Read_Master_Log_Pos: 261
Relay_Log_File: Node1-relay-bin.000002
Relay_Log_Pos: 406
Relay_Master_Log_File: mysql-bin.000034
Slave_IO_Running: Yes
Slave_SQL_Running: Yes

To test it you can drop table from Node2 and make sure the table is dropped on node1 too.

That’s all !

No related posts.

I’m not going to defend open source, because there are enough arguments, technologies, and solutions that are very competitive and no closed technology can compete with them, however my only take is only for education.

Why most education systems “impose” the use of proprietary solutions ? I did not said that MS Office is bad, but do not force me to use this product. Why teaching kids how to use Windows, Word, Excel, PowerPoint… then Access for advanced users !

As a student I’ll have probably to write text, reports, table sheets, presentations … I don’t think this deserve to be taught separately and marked in the final exams ! People behind education programs are sometimes old fashion and love to stick to old stuff they learned, or just prefer the solution of facility if it’s about teaching few basic features documented and sponsored by the software provider.

The only good things is probably teaching Java as main programming language in most universities and colleges, and my apologies for those who are stick to Pascal “forever”. It’s okay to teach .NET but don’t stick to Visual Studio and Windows environment, there is Mono for people who don’t know it which provide an open source alternative.

In business, the choice of open or closed technology is not a matter of evangelism. I had the chance probably to work on a large open source solution using a closed source database for a government institution, and actually another new project which is based mainly on closed technology and integrated with the open source solution.

The core business solution is stable for one year already, and is mainly powered by open source technologies (operating systems, web server, web application), even the closed source database is running on an open source OS.

Getting the new project running with closed technology doesn’t much matter for me – even if I won’t recommend it if I was asked before. Making the business evolute is my only concern, and integration was really made easy between the two technologies.

My only worries today is about the capabilities of the closed technology to support a high number of users, and high traffic during peak times. In the same time I’m not much worried, because the core business is in the safe side, if the new implementation for a reason or another fail, only the “closed source” part of the system will fail and not the entire system.

Businesses and consultancies have the choice, most of the time, but they are mostly interested to work behind a giant company name to make sure that their solution will be accepted. By experience selling a custom made solution (powered by open source technologies) is much more profitable than reselling licenses of any available software, but the lack of experience and confidence doesn’t give much choices.

I think I have answered a key question here : small companies cannot compete with open source software, get behind a large company name and you will be able to sell any open source solution at a very competitive price compared to closed source one. Open and closed source are not enemies, it’s up to you to use one or the other, or make them both coexist.

Related posts:

1. Say YES to Open Source
2. Open Source Evangelism
3. Printing in Intranet Web Applications with IE

A website could be slow for many reasons, and to be able to fix it and make it run faster you have to first find problem. What you will need to know is :

1- Understand the business :

Before looking into PHP or MySQL, have a first look at the website(s) and answer these two questions :

- What is slow
- Why is it slow

Look at the page size, images, video, flash, … your issue might be in the client side and everything else might be good.
Use a tool like Firefox web developer extension and disable everything : CSS, Javascript, Images… and load the page as html to feel the difference.
The business here is NOT the business logic behind the website, it’s the interface that people use to interact with the website. If everything looks okay, you can go for further server-side investigations.

2- Understand the production environment : Operating system, PHP and MySQL versions, PHP extensions running, PHP and MySQL configurations.

Do not start optimizing code or database unless you know the exact problem, otherwise you have to start by understanding and optimizing the working environment. Some tweaks in PHP and MySQL settings might fix your problem and you won’t probably need to dig deeply into any code.

3- Understand the code :

If the websites you are looking to optimize are using a common framework or CMS, let’s say Drupal, WordPress, Zend Framework, Symfony… such websites you can deal with separetely since most optimization issues should be known and you will easily find your way to speed things up.

Otherwise, for custom code and custom application development, start with debugging and focus firstly on the database. That’s where most slow issues come from, especially if the concerned website was running fast in the beginning then become slow over the time due to database size going bigger, query & index issues…

4- Debug : detect and locate bottlenecks

Debuggers are your friends here, whatever the environment you are running. Answer again previous questions (What and Why) but this time with relation to the code itself.

You might consider running a stress test here to simulate real working environment. Such tests could help you understand more networking issues and make it easier to locate slowliness based on real scenarios. Test should also simulate real environment, and you will need to save speed result for comparison later.

5- Solution proposal and action plan

Now you can go ahead and set possible solutions : fixing bugs, caching code/queries, tweak your settings… Make sure your modifications will not affect other working website in the same environment – if any is somehow related or linked.

Finally you can run previous stress test with new configuration and updates to see the difference in speed gained. A report of previously mentioned steps should be written including result.

Related posts:

1. Create a Master Master MySQL replication – Ubuntu Server 10.04 x64
2. Say YES to Open Source

From yesser website, YEFI is defined as “a unified framework to implement e-government. It includes cross – governmental specifications and policies, to enable cross – governmental integration and facilitate G2G transactions and data sharing.”

While reading deeply about YEFI, I noticed that the framework cover almost everything we need – for integration purpose at least – between different government institutions. The common part of the schemas between different ministries is almost the same : Address, ID number, place & date of birth, Last and first name, while they keep different names.

The creation of data catalog is the first step to get these data defined, where every entry should include as fields : Name, Business information, Format, Validation, Verification, XML Schema IDs, Values, Default value, Owner, Version and Acceptance date. The data schema will be in this case the collection of all these elements defined in data catalog.

The schema should include meta data; if we are sending person’s data for example the schema should start with meta data as : Business object created, Business object updated, owner, data schema ID, Global unique object ID; then followed by person data as : Person birth place & date, First & Last name … etc.

A central repository for the standardized schemas is required to handle persistent data of oudated schemas. As every institution data might evolve over the years, and data schema will be updated; a repository should be available to document schemas versions and conversion rules.

The recommended integration technologies standards are : WebServices for middleware, UTF-16 for character sets, XML/XSL and XML schema, and finally RDF as resource description framework.

More details on YEFI program and technical documentation are available at http://www.yesser.gov.sa/english/YEFI.asp

Related posts:

1. Say YES to Open Source

The first session was about Sun SPARC enterprise servers, and I liked the idea of supporting mix of processors versions in the M4000, M500 and M9000 (the M series servers) – and it’s only Sun that offer this compared to IBM and HP. Some benchmarks also in this session and was very informative; also covered different aspect of the M-Series and CMT servers for single-threaded and multi-threaded.

In the second session the Datacenter Architect talked about Datacenter efficiency through virtualization. Andy talked about different level of virtualization Servers, Storage, and Network, in addition to virtualization tool Sun VDI software and Sun Ops Center. The business case in the end was very interesting showing 14% ROI in the first year only. There is something I noticed that virtualization tools are available free with Open Solaris, and coming soon to Solaris; and you don’t have to pay more to use them ! It’s something like when you buy a new car, then you try to use the Radio and they tell you have to pay more for this feature. Well, I totally agree with Andy and find it really ridiculous to do this; especially that I have a similar problem with HP Storage and the way they limit licencing to a fixed number of TB – so to add more capacity we’ll need a new license.

In the last session, Thomas talked about Sun Storage 7000 Unified Storage Systems. Almost all speakers was insisting on the ZFS – the famous Sun File Systems; obviously it was mentioned more frequently with the storage session. I liked especially the Enterprise SSDs, which eliminates storage bottlenecks. The ZFS pool guarantee a better performance for writes via the ZIL pool and reads via L2ARC pool – both connected to SSDs in addition to a third pool with cheap HDs.

The last session was about Sun FlashFire; also by Thomas. There was a case study of the KAUST presented by Mohammed Abdel Aal; Manager KAUST IT Computing Infrastructure. Mr Mohammed talked more about KAUST research strategy, but there is something that I liked most in KAUST IT requirements “No legacy systems, no legacy thinking”.

Overall it was great meeting IT professionals. I have already worked with Sun servers and Solaris, but I don’t have previously good knowledge of their hardware architecture. I was only familiar with ZFS and Open Solaris.

I didn’t stay for the lunch, but I expect it to be great too.

No related posts.

Building e-Government services is not that hard, not that easy also. There are many steps that should be followed to be able – at a certain time – to move totally from an old paper-based system, to a fully electronic system.

The most interesting in e-Government that you are building services for people, and people are the customers, the users, while they usually never interact during the building process – but we have many considerations to keep in mind while building for people.

Some interesting steps in e-Gov creation where we should focus mainly include :

• Business Process Re-engineering
• e-Government system architecture
• Systems interoperability and scalability
• Security : systems, network, and end-users
• Cost consideration : hardware, software, network, development in-house, outsourcing, and management
• Web-enabling requirements : accessibility, usability, technologies standards
• Supporting Multiple Communication channels
• Internal software and hardware for administration (legacy systems) : RH, Finance, DMS, Archiving, CRM, Workflow, Communication (Email, Voice, IM)

It is interesting to turn Governments process into e-Services, but building platforms that support different business process levels G2G, G2B and G2C is much more interesting.

Some experts started talking about Government 2.0, or Government as Platform. A great concept especially since it allows anybody to build government services or integrate it with other already established business process elsewhere.

The problem is that building platforms is totally different from building services, and IT developers/architects used to create web applications that do not scale at the Government level. One of these problems is the execution of the CAP theorem in Governments web applications. The fact that we require “strong consistency” in e-Government is a major limitation for high availability in web platforms.

“eventual consistency is an  eventual solution, but implement it correctly is a major obstacle. The reason why developers/architects have to better understand the best practices for building eventually consistent systems, in highly available environments.

Related posts:

1. Open Source, Closed Source, We are not enemies
2. YEFI e-Government Interoperability Framework
3. Open Source Evangelism

Description :

Though estimates vary, most studies find that about one fifth (20%) of
the population has some kind of disability (WebAIM.org). The
presentation will try to raise awarness about making the web
accessible for people with abilities and disabilities, with practical
techniques and tools to make the web accessible.

Duration : 20-30 minutes
About the speaker : Hatem Ben Yacoub – Senior systems architect and IT
Consultant

العنوان: مقدمة لقابلية الوصول في المواقع
وصف العرض :

على الرغم من تباين التقديرات ، فإن معظم الدراسات وجدت أن نحو خمس (20
٪) السكان في العالم لديه نوع من الاحتياجات الخاصة (WebAIM.org). العرض
سيحاول زيادة إدراك حول جعل الإنترنت في متناول عموم الناس من ذوي
الاحتياجات الخاصة او غيرهم ، من خلال تقنيات وأدوات عملية لبناء مواقع
إنترنت افضل.

طريقة العرض : عرض تقديمي
مدة العرض : 30 دقيقة
مقدم العرض : حاتم بن يعقوب – مهندس أنظمة ومستشار في تقنية المعلومات

Related posts:

1. Introducing CouchDB at the first Jeddah Geeks meetup

Using windows as terminal for a web application is not a problem, but problems start when you have different version of windows and Internet explorer here and there, but especially when we deal with other devices like printers or barcode readers.

One of these issues is completely strange and personally I couldn’t explain the reason why when a user change a printer, the web page margins change ! Some employees have more than one printer available for common A4 printing tasks, barcode printing … etc. So it’s annoying to fix margins – even from registry – and then seeing these values changing for every printer – randomly !

There is always a solution, and for this problem specifically  I wrote an activex script that reset page margins before printing anything. It worked fine, in addition to another script that deploy some IE settings to all clients in the network through active directory.

But seriously, can someone explain why when a user change printer, the web page margins change ?

Related posts:

1. Open Source, Closed Source, We are not enemies
2. Speed up your PHP website, a consultant’s guide

This year I felt the need to talk about Open Source more than any time before, not to invite people to embrace Open Source technologies and ideologies – I’m technologist not an ideologist anyway – but it’s more about showing real-life technical problems and how to solve them using Open Source solutions.

If Evangelist is not an appropriate term, maybe I can use the term Expert instead ? But I don’t consider myself expert in all Open Source technologies the reason why I have preference for the term Evangelism. And you have probably noticed that some evangelists by defending Open Source become totally in opposition to proprietary solutions, this is not my case hopefully.

Both proprietary and Open Source solutions are part of the same Eco-system today and I think everybody is free to decide which is better for them or make them co-exist. But in the same way that there are companies that defend their proprietary solutions, I think Open Source still need more evangelists to defend it the right way.

I just wanted to post this article to clarify some ideas about Open Source Evangelism. You will hear me talking more frequently about “High Availability Open Source in e-Government and Enterprise” in the next days not to say why we should use it, but with the main focus on how we can use it correctly.

Related posts:

1. Open Source, Closed Source, We are not enemies
2. Say YES to Open Source
3. e-Government as Platform