Let’s start with WordPress installation on Heroku. Just recently Heroku support PHP on their platform in addition to other high level language like Ruby/Pyhton etc. Before we go on the step-by-step installation, let’s go through the terms:

• Slugs are compressed and pre-packaged copies of your application optimised for distribution to the dyno manager. Detailed description can be found here.
  
• Dyno is a lightweight container running a single user-specified command, we can see a dyno as our server. Detailed description can be found here.
• Dyno Manager will manage the Dyno under it’s provision, it is responsible for scaling as well as work load balancing. Detailed description can be found here.
• Git the famous Source Control Manager, we will use this tools to deploy our application to Heroku.
• Buildpack contains collection of scripts that will construct our Slug to be run inside the Dyno (or Dynos on distributed environment). Detailed description can be found here.

From conventional shared hosting/full server installation point of view, the installation of WordPress is pretty straight forward (configure the OS, install apache/other web server with PHP support, install MySQL, configure virtual host, install WordPress, configure WordPress) but from Heroku which is a Platform as a Service, the installation is quite different, we don’t have to configure the server since it’s all taken care of and the installation of supporting system is also different. Below is a simple comparison between them:

• OS installation is not needed
• Web Server and WordPress is bundled in the Slug, compared to shared hosting which is usually a shared Apache server and our copy of WordPress in the public_html folder (or whatever personal web folder). The buildpack will be responsible on the construction of the Slug. The Slug itself will also contain script that is responsible to launch our Web Server and PHP subsystem to run WordPress.
• Storage should be external compared to shared hosting, since the nature of Heroku is to enable distributed processing dynamically, the storage of a Dyno is ephemeral, means it won’t survive on reboot, just as an Amazon’s EC2 instance by default. Because of this behaviour, we should always treat our Slug as “Read Only” file system (although write to the file system is possible, but it is not permanent). This discipline is very useful on distributed cluster environment, we don’t have to worry about how to share file storage among the instances and all of our Dynos will always contain an identical Slug. We should design our configuration so that the application will always use shared storage system, in our case we will use Amazon’s S3 (Simple Storage Service, a REST based file storage) for our file storage and SimpleDB for our MySQL server.

With those rules in mind, what we need now is to prepare our Heroku account and our buildpack. There are ready to use custom Heroku buildpack for WordPress out there, but i found this buildpack by mchung is the one which suite my preference. It utilise most of what Heroku ecosystem has to offer via the add-ons. The instruction is very clear at his website, but I did some tweaking on my setup especially for the binaries used.

I. Create a Heroku Account

We need to create an account with Heroku first. You need to have your credit card handy for verification purpose, even if we don’t want to be charged for anything for now. Just go to Heroku website and create your account. Don’t forget to verify your account with the credit card details.

II. Download the Heroku Toolbelt

In order to use Heroku, you need to have the toolset for heroku environment called Heroku Toolbelt. Download it from here.

III. Fork the WordPress Template from Github

First, create an account on github.com, it should be straight forward and free.

We will use two Git repositories hosted in github.com, the reason for this is to separate the main buildpack with our own customization (additional plugins, themes, config). The first one is the buildpack itself, it will be called heroku-buildpack-wordpress.git the second one is the template for our wordpress configuration called wordpress-on-heroku.git. Fork the template repository because most likely you will configure some of the options:

• https://github.com/mchung/wordpress-on-heroku

Or alternatively if you want to tweak some stuff for the buildpack (like your own PHP build, etc), you can also fork the buildpack itself:

• https://github.com/mchung/heroku-buildpack-wordpress

To fork these repositories just log in to github.com, click on above link, and click the “Fork” button at the upper right corner.

IV. Create Heroku Apps and Deploy

First, let’s get a local copy of the wordpress template git repository (wordpress-on-heroku)

$ git clone https://github.com/your_name/wordpress-on-heroku.git ourblog.com

Some explanation:

• The command ‘git’ is bundled with the Heroku Toolbelt, ‘clone’ means that we want to checkout the repository into our local storage
• Please rename the word ‘your_name’ at the  url ‘https://github.com/your_name/wordpress-on-heroku.git’ to your username at github.com
• Last bit, ‘ourblog.com’ is simply the name of the directory of our local copy, you may rename it to your liking

Before we deploy the template and the buildpack to heroku, I would suggest you to tweak some configuration on wp-config.php at this section:

/**#@+
 * Authentication Unique Keys and Salts.
 *
 * Change these to different unique phrases!
 * You can generate these using the {@link https://api.wordpress.org/secret-key/1.1/salt/ WordPress.org secret-key service}
 * You can change these at any point in time to invalidate all existing cookies. This will force all users to have to log in again.
 *
 * @since 2.6.0
 */
define("AUTH_KEY",         "put your unique phrase here");
define("SECURE_AUTH_KEY",  "put your unique phrase here");
define("LOGGED_IN_KEY",    "put your unique phrase here");
define("NONCE_KEY",        "put your unique phrase here");
define("AUTH_SALT",        "put your unique phrase here");
define("SECURE_AUTH_SALT", "put your unique phrase here");
define("LOGGED_IN_SALT",   "put your unique phrase here");
define("NONCE_SALT",       "put your unique phrase here");

Generate your own salts using link suggested. The file will be located at:

 ourblog.com/config/public/wp-admin.php

alternatively you can add your own themes at this location:

 ourblog.com/config/public/wp-content/themes/

and additional plugins at:

 ourblog.com/config/public/wp-content/plugins/

after the changes, we need to tell git that we are done with the changes and commit the changes, first, make sure we are in ourblog.com directory, and then issue these commands:

$ git add .
$ git commit -m "Configure wp-config.php, added new themes and plugins"

Now we can create and deploy our apps in Heroku (still in ourblog.com directory) issue this command:

$ heroku create -s cedar
$ heroku config:add BUILDPACK_URL=https://github.com/mchung/heroku-buildpack-wordpress.git
$ git push heroku master

In my case, I decided to have my own built PHP and Nginx:

$ heroku create -s cedar
$ heroku config:add BUILDPACK_URL=https://github.com/pampie/heroku-buildpack-wordpress.git
$ git push heroku master

Now you can access using this command:

$ heroku apps:open

Configure the domain name

Ok, we are already having our blog up and running, but the blog URL is still pointing at whatever.herokuapp.com, we need it to be ourblog.com.
First we need to tell Heroku that our web will be accessed by using ourblog.com by adding ‘alias':

 $ heroku domains:add ourblog.com

And then configure the DNS for ourblog.com to have a record CNAME to herokuapp.com.

Further customization

Please remember that we cannot use WordPress’ automatic update since it will write on an ephemeral storage. Should you need to change/update something, do it on the local repository, do ‘git add’ or ‘git remove’ and the ‘git commit -m “some comment”‘ and then push the changes to heroku by ‘git push heroku master’.

Here we go:

 Heroku

On that review, Heroku listed as #1 compared to other cloud providers and I could not agree more, except for geographical coverage, it only offers EU and US servers.  However it can be covered by using CDN that will be covered later in this series.

Here is the checklist as listed in my previous post:

1. The server should run Apache + PHP as required by WordPress
   The answer is: Yes. But because I’m using a custom buildpack, it will run on Nginx + PHP stack instead.
2. The system should support custom domain.
   The answer is: Yes
3. There should be a MySQL database
   The answer is: Yes. One of it’s addons will automatically create a database with ClearDB cloud database provider. The custom WordPress buildpack will enable this by default.
4. (Optional) Memcache server to act as reverse proxy for the web server, at the speed of RAM
   The answer is: Yes. Again, one of it’s addons will automatically create an account with Memcachier, one of Memcache cloud provider. The WordPress custom buildpack is also enable this by default.
5. (Optional) Content Deliver Network, users will be served based on their geographical presents.
   The answer is: No. It does not offers CDN by itself, however it can be supported out of the box by other CDN available in the market.

Can all of the subsystem migrated? Yes and no. All of the system can be migrated with a note that Heroku storage is an ephemeral storage, means that the storage will be gone if the server restarted. While some of the WordPress function needed a permanent storage, such as file upload for media library, then we need to store this file to somewhere else. In this setup we will use Amazon’s S3 (Simple Storage Service), an REST based storage by AWS.

What about the cost? As long as you stay in the corridor of free tier, then it will be free. Amazon S3 will charge you a fair amount of disk usage, but if you use Amazon S3 for other use then the cost will be insignificant. There are some restriction on this ‘free’ plan:

• You can only run 1 dyno (this is a Heroku term, it is a virtual container of 512MB or 1 GB of RAM) if you want the server to be up all times. Heroku account enables you to have 750 hours of free dyno usage per month. There is also some restriction too, if you run only 1 dyno and there is no activity served by that dyno then it will ‘sleep’, the subsequent request to the server will need to wait  for the dyno to be awaken.
• ClearDB MySQL database free tier (ignite plan) will allow you to have 5MB of database size, above that it’s a pay per use service.
• Memcachier memcache service free tier (dev) is 25MB of stored objects.

How easy is it? From a shared hosting perspective, the migration to Heroku is somewhat different. The terminology used is unconventional and some of them are exclusive to Heroku. At start the learning curve is steep, be ready.

How fast is it? Google Speed Test reports 91/100 for desktop view and 76/100 for mobile view. I’m using a responsive theme thus the mobile version is not actually a special theme for mobile device, it just ‘responsive’ to the view port of the browser so the performance is not that fast.

This site is currently running on this setup.

 Amazon Elastic Beanstalk

We can say that Amazon Web Service is the mother of all cloud services out there. Many other cloud based services are using Amazon’s PaaS (Platform as a Service) or IaaS (Infrastructure as a Service) solution for the basic of their offering. Amazon Elastic Beanstalk is one of their offering that is basically a bundle of other basic products which are: EC2, Load Balancer, Storage (EBS & S3), and of course bandwidth.

Let’s go to the checklist again:

1. The server should run Apache + PHP as required by WordPress
   The answer is: Yes. It uses EC2 (Elastic Compute Cloud), it is a virtual server in the cloud, so basically you can run anything on it.
2. The system should support custom domain.
   The answer is: Yes. It act as our own server, we can use it freely
3. There should be a MySQL database
   The answer is: Yes. We can install mysql inside EC2, or using Amazon’s RDS (with additional fee, unless you are still under free-tier plan).
4. (Optional) Memcache server to act as reverse proxy for the web server, at the speed of RAM
   The answer is: Yes. Amazon ElasticCache is in memory cache solution from Amazon. It is part of free-tier policy so it will be free for the first year.
5. (Optional) Content Deliver Network, users will be served based on their geographical presents.
   The answer is: Yes. It can use Amazon’s CDN service (CloudFront). In addition, it automatically configured Load Balancing option and auto-scalling capability provided that you choose “Load Balancing, Autoscaling Environment” as the environment option.

Can all of the subsystem migrated? Yes and No. As of Heroku, EC2 instance’s storage is an ephemeral storage, it doesn’t survive on server restart. However there are two options available, first is to use Amazon’s Elastic Block Storage as permanent storage or using Amazon’s S3 storage.

What about the cost? It is free, for one year, for new user only. When you sign-up with Amazon, they will give you a free-tier for some of their services. After that it’s quiet decent compared to shared hosting option. But, if the site you are planning to build will generate some hefty load some day, this option is the best since it will charge you on pay-per-use basis. Pricing example can be found here.

How easy is it? To answer this question it depends on how do you understand AWS’s terminology and services. If you are familiar with AWS then it is very easy compared to configuring individual services by hand (creating EC2 instance, creating Load Balance, etc.). But if you are new to AWS then you should learn AWS’s system and offerings.

How fast is it? With default application, it scores 92/100 for desktop view and 79/100 for mobile view. I plan to write a series of Amazon’s services after this series, and this service should be one of it.

Redhat’s OpenShift

RedHat is a strong player in Linux for Enterprise market, and this initiatives is their move to gain traction into Cloud offering. It has it’s own terminology and platform, it based on AWS’ infrastructure with RedHat’s tools.

Similar to Heroku, OpenShift leverage partners on their offerings including MongoDB, Sendgrid, ZendServer to name a few. I believe if their marketing and strategy matches market demands, there will be more partner to join in the band wagon.

Here is the checklist again:

1. The server should run Apache + PHP as required by WordPress
   The answer is: Yes. OpenShift’s terminology of server is a ‘Gear’, this similar to what Heroku called a ‘Dyno’. Unlike Heroku, we can actually ssh into the box to get the shell, but it also support extensions called ‘Cartridge’. There are not much to choose from, maybe because it relatively new, but for WordPress it has PHP and MySQL cartridge so it’s enough. As a bonus there is also phpMyAdmin cartridge. For PHP there are two option: Apache+PHP and the other is more advanced ZendServer.
2. The system should support custom domain.
   The answer is: Yes. We can configure the custom domain, more than one supported.
3. There should be a MySQL database
   The answer is: Yes. As pointed out on point #1, there is a MySQL cartridge available to install.
4. (Optional) Memcache server to act as reverse proxy for the web server, at the speed of RAM
   The answer is: No. It is not of the box, although integration to Memcachier is possible but requires additional setup.
5. (Optional) Content Deliver Network, users will be served based on their geographical presents.
   The answer is: No. It does not offers CDN by itself, however it can be supported out of the box by other CDN available in the market.

Can all of the subsystem migrated? Yes, although the main part of the directory is in ephemeral storage (remember, it is an EC2 instance anyway) but RedHat generously give the user free 1 GB of EBS storage, it is a permanent storage so we don’t have to use S3 for additional storage.

What about the cost? It is free for the first 3 gears, we only need one so it is more than enough.

How easy is it? It surprisingly easy, the wizard is straight forward and the configuration is also easy.

How fast is it? With ZendServer as webserver and PHP stack, the Google page speed score is 91/100 for desktop view and 78/100 for mobile view.

Other Providers

The other providers either doesn’t have free tier (because when it is paid then the cost will be too much compared to current hosting), or they will not support custom domain for their free-tier offering, so it is out of discussion.

On the subsequent posting I will discuss detail the step-by-step installation based on WordPress 3.6.

Still, my curiosity on the new technologies never stops. I always interested in the new trends that are happened in the IT world, and currently what is on my magnifying lens is two things: cloud and big data, but for this series of blog entries, the discussions will be on how I migrate my traditional web hosting of WordPress and also php websites into cloud with the result of reduced cost and higher performance.

For the past 2 years, a lot of things happened in the Cloud area. The technology is getting mature each and every day, more and more companies are relying on the Cloud for their business. New startups leveraging Cloud are getting momentums and skyrocketed in terms of market value and subscribers. With these mumbo-jumbo surrounding the term, I try to take a subset of it and to map my current setup with Cloud terminology and implementation.

There is no problem with my current web hosting actually. The price is reasonable, it is a shared web hosting so the cost is low, the support is great and fast, no problem so far and as far as I know the service never down. But for the shake of knowledge and experience, I had an idea: what if I migrate the whole system to the cloud, what are the challenges? How far can it be for the subsystems to be migrated to the cloud, partially or full? What are the costs incurred? How easy/hard it is? What about performances? What about security? What about manageability? What is the final verdict?

Previous System

As discussed earlier, the old design is pretty straight forward and traditional. It is a shared hosting setup with cpanel management installed. The multi-tenant configuration is very efficient for this type of website since the traffic is low and the SLA is not too high if not nothing.

The cost of this setup is also low, it charge me about $15/year for the hosting space of 250MB, 2 GB traffic/mo, it just slightly above $1/mo, a bargain really.

The management part is also very easy with cpanel system, although it doesn’t allow me to ssh into the system. It just took me some clicks to install WordPress module along with MySQL database. There is phpMyAdmin installed out of the box, system wide, and the DNS management is also built in the system for me to create subdomains for WordPress MU, it is a no-brainer.

Is there any problem? Barely not.

The Requirement

Before proceeding, there are some simple basic checklist that need to be considered:

• The system should run Apache+PHP, it is a WordPress system and a simple php web
• The system should support custom domain, I will use my own domain with WordPress MU subdomain setup
• There should be MySQL database as a database layer for WordPress
• (Optional) Memcache system to speedup the performance
• (Optional) Content Delivery Network to speedup the performance geographically

On the next post, I will discuss about some alternatives of Cloud provider existed with pro and cons.

It was later I found out that all my documents are stored in a Storage Attached Network (SAN), and every time I logged into an computer, automatically my roaming profiles will be in effect thus mounting my personalized details such as my personal registries and documents to that computer. It was one of it’s kind, not many companies doing that at that time and I must admit that it was very convenient that I can logged in to any computer and do my works without having to carry my documents around. However, this implementation did not take what cloud computing are offers today. The modern implementation is beyond that.

Nowadays, cloud computing is very popular. Many companies are doing it such as Amazon (S3), Google (Google apps, Google docs), Microsoft (Azure) to name a few. What they are doing is to take the applications and documents in the clouds. Take google docs for example, this implementation is for personal or companies who doesn’t need a full blown office applications or their workers are highly mobile. Each of the team member or person can logged in into their google accounts and create/modify regular office documents. These application and documents are virtually available on the ‘clouds’ long as there is internet connection and a browser. Furthermore, the documents can be shared among the team members in an easy way. As for me, I also find it convenient using Google Docs to create and edit some of my personal documents.

For storage purpose, I choose Dropbox with it’s free 2 Gigabytes of storage (it’s a pay service for more than that). I found it easy for me to store documents that will be edited across desktop and portable computers in my household. The clients are also available on Linux, Mac OSX and iDevices. So every time needed, I can logged in to any of the devices and edit the documents as if it’s in my local storage and automatically sync into dropbox’s cloud storage. It also has a handy document revision features, i.e. it will create two copy of the documents versions created in two device at once.

The popularity of cloud model is also applied to web services. Google generously offers their system to be a Content Delivery Network (CDN) for commonly used AJAX libraries. This site of mine is also use Google as the jQuery library CDN. The reason behind this is because jQuery library is a common and famous library and it can speed up things up if it is served from a site as Google servers, it also promote caching too from the user’s browser point of view. The other thing of this site cloud implementation is image storage, it uses google picasa storage as you may notice if you carefully inspecting the images URLs. Some moving further to use dropbox as a full CDN for their wordpress’s theme files, thus offloading precious hosting storage for other use.

I am not surprised if in the future this model will replace locally stored documents and applications. Local harddisk is notoriously easy to fail since it uses moving parts, while the Solid Disk Drive is expensive. Internet speed are picking up and transferring megabytes of data through the internet is nolonger a luxulury. Meanwhile, I slowly change all my computing behavior to the cloud. Cloudify Me.

Opening applications seems like forever, answering a call while an application running is also frustating. The swipe-to-answer mechanism is becoming scratch-to-answer, fells like a monkey finding a bug. But upgrading to iOS 4 was a mandatory thing since more and more appplication can only run on version 4.

Enough said, no more pitty-party, let’s find a fun thing to do on this interesting platform.

DISCLAIMER, this following procedures is for experimental only, I am not responsible for any bad things caused by this experiment.

Let’s see, there are plenty of things to consider when optimizing a system. Defragmenting, processor upgrade, memory upgrade to name a few, none of them are relevant to i device. Well there is one trick that can be done: DNS.

Yes, it might be the only one we can do right now. It’s easy to change DNS settings when using Wifi connection, it only requires changes on the IP setting for that particular connection. But for unknown reason as usual, Apple decided not to include this DNS setting on mobile network.

To complicate the problem, according to Saurik, the guy behind Cydia found here the name-to-IP mechanism on iOS is somewhat different than its *nix brothers. Usually the name server configuration should be placed in /etc/resolv.conf, some variant placed them on /var/resolv.conf but soft-linked to the former. Having proved this myself, created these two files on the iOS filesystem with no luck.

Thanks to this tutorial which tells us that Darwin Kernel in which all of the Apple Operating System derived from, somewhat mysterious on the way they do the name resolving stuff.

Well, if you are having plenty of time to spare you may see those links to study about the behind the scene stories and the mumbo jumbo things surrounding this mystery. But to make things simpler, I’ve created a simple script here that you can try.

#!/bin/sh
state=`echo "list State:/Network/Service/[^/]+/DNS" | scutil | awk '{print $4}'` 

scutil << EOF
get $state
d.add ServerAddresses * 208.67.222.222 208.67.220.220
set $state
show $state
EOF

Make sure that you have mobile terminal handy to run this script as root. Since the Mobile Terminal on the default Cydia repository is not compatible yet with iOS 4, please see here for step by step installing the program on your iPhone.

If you are curious about what this script is doing, it simply find an entry on the Darwin Kernel for the name resolver using ‘list’ command on scutil. The value of the GUID then feeded into scutil again, but for the purpose of changing the name server to Open DNS’ name server.

This script is not persistence, it will replaced by the Operator’s setting if the data connection is restarted. But hey, for the purpose of fun, this one is quite great

Getting mobile device is also a lot of easier nowadays, only by less than $30 to get, and the cost of running it is also cheaper. Operators are like crazy offering packages and offers, luring new subscribers or, hopefully, converts.

This competition will certainly lowering their ARPU (Average Revenue Per User) figure, but the key of getting the profit is the volume of subscriber and calls or texts. The other new stream of revenue is the need of the people demanding Internet connection on their mobile device.

Online everywhere, whenever needed becoming new trends. This kind of life style is supported by new mobile gadgets that are more and more sophisticated, smart and also faster. These gadgets are a mean of tools, communication device, entertainment as well as status pendant to some.

Wether we like it or not, now is the new era where mobile devices are parts of human life. In addition to that, the technology enables these devices to function more or less as personal computer, multimedia platform and also a gateway to the cyber world.

• iPhone OS is basically an *nix variant?
• Every *nix system will have at least 1 administrator called ‘root’ and some regular user?
• Every iPhone shipped is having same ‘root’ password and a regular user

So is it dangerous? Not exactly. On regular iPhone device there is noway for somebody to exploit those userid and password, *but* it will be completely differrent when we are talking about jailbroken device. So if you are a novice user and faithful user of app store then nothing to worry about. You can safely stop reading here and have a nice dream.

But if you are having a jailbroken device then you have to worry, don’t ever install OpenSSH package before you change the default passwords

How? Simply download console package from cydia and do a regular password change procedure (just google on it) don’t forget on root password too.

Life – Dreams = Job

…

Do I still manage to maintain my dreams? Or worse, do I still dreaming? Post me your comments

Now it’s late, I have to ‘dream’ (literally )

This post is made by Qumana

They are inventing the famous method of putting a weight to a website by using their patented algorithm called “PageRank”. As you may notice that the name came from one of the duo. The simple version of this algorithm is that the ‘importance’ of a web page will be determined by other web which having a link to it. It will count the quantity (how many link to that website) and also the quality (who link to that website). Say if you have a link in CNN homepage, then you’re seriously famous and you shall have the honor to have higher pagerank, and thus will have the bigger chance of being appear in a search result.

Now, what about on our work. I notice that this PageRank algorithm is applied to our daily work, let’s call it pseudo-PageRank [forgive me google boys if I steal your term]. The mechanics is still the same: you are what other people ‘linked’ to. Say, if you are known by a high profile people then there will be greater chance of you being ‘in-sight’ and a bigger chance of promotion. Provided that you are more to ‘famous’ than ‘notorious’.

What technical guys do, and how they think, and how they act, is that what matters most is the know-how. They are continuously studying to be the best of their area, more certification, more trainings. But few are having this kind of sight: the importance of being somebody to the management [i.e. the decision makers – no matters how un-techie they are]

Being a technical guy myself, I have very long learning curve on this. Years and years of continuous technical learning, and more people with less technical expertise was already taken off to the moon.

Sounds familiar?

Thanks to Linkedin, I understand that my ex colleague that is now on a higher position is doing their job both technically and strategically. Some of them are being the job-in-a-year mode (I know some of them doing three-in-a-year mode). Some of them are being soooo patience of years of service in a same company. But something I know that PR is needed, even in our workplace.

I am introvert people, even in a MIBT test I score perfect in this area (I am a INTP kind of person, mind you). Thus I must work very hard to achieve this goal. I have to love people, love my boss [genuinely] and to be carefully not being a licker myself [ay… I hate this kind of species, more on this latter on the other post].