DevOps – Dont stop with Agile for developement


5 out of 10 customers that i have interacted in the last 30 days are talking about devops and interestingly everyone seems to go in that direction. In the Banking Organization context, “Run the org” and “Change the org” have been two accepted philosophy from getting things done. The line has blurred over the years. In a ideal world i would like to “run the org” with as much efficiency as possible and redeploy the energy in “change the org”. In short time, this has changed a bit. From budgetary concern , we normally see the squeezing out of “run the org” portion of money and the change the org has taken preference and rightly so. While every school of thought has its own followings and haters, from the business perspective some of the points needs to be treated in the right spirit for success of the organization.

Run the Org: DevOps which normally is a cultural shift from the world of development and operation completely separated, advocates a close collaboration between both and overlap of responsibilities. A typical run the org kind of work would involve keeping the system going on, dirty fixes required to have the day to day operation going on. Some of the characteristic symptoms of “Run the Org” would be as follows

  • Production support
  • Data changes/errors including changes to database or file system.
  • File exchanges or the failure of it.
  • By passing some of programmatic error.
  • Most importantly access to Production instances

If minutely considered there are different kind of activities involved and would mostly be done from ops perspective.

Change the organization advocates the need of more bang for bucks, faster, better. A DEVOPS philosophy would give rise to

  • continuous release
  • incremental features
  • Continuous development.

So in essence, devops could be combined with a lot of competing and complementing philosophy to bring the best practice to the organization.

Social Media Analytics – Have we done it all ?


Enter into a CMO’s office of a Bank with a pitch for sentiment analysis product or service, you probably would be thrown outta the window. The problem with sentiment analysis is,  its too old. Everyone has done it or has seen it working (if at all). Specially if you are in the retail industry (irrespective of what you sell), it makes a lot of sense to know how your products are doing in the market or in the mind of the consumer. People’s opinion and verbiage provides a direct feedback about what they feel while experiencing it. Its absolutely important to listen to your customer’s expression as it might snow ball into a negative larva ball killing your product or service even before the customer actually buys it.

Compare that to a banking parlance, the oddity are too many. But some of the most important attributes are

  1. There is no tangible product.
  2. There is no feeling associated with the product.
  3. There is no innovation in the primary product variance in the banking industry (a deposit account is a deposit account huh!!!)

So it would make a lot of sense to try and use the medium of social media to collect feedback loop around services rendered or in short “customer experience”. Various firms with digital aspirations have already established an arm of customer service through twitter handles. Almost all the banks have a Facebook page to like/dislike. There are dedicated fan/hate pages also for banks of various nature existing in Facebook and twitter without any involvement from the said bank/institutions. The thing about social media or the feedback through it is – You get a feedback only when its negative mostly. No one twits or posts a comment if they like anything without being prompted (paid) mostly. The best cases thus for the banks to use social media aanalytics would probably to

  • Gauge the customer experience – of an incident interfaced with a channel agent
  • Gauge the performance of the channel agent to turn the negative approach to a positive experience.
  • Mix up the blend of customer feedback from formalized channels like (survey) along with unstructured formats to establish like and dislike of customers.
  • Identify conversation threads which potentially could be a competitor intelligence (read it as Topic Modelling).
  • Compare products and services among competitors.

While all these things are technical and mathematical in nature, there is one area of it however is taken for granted. The social media related privacy, framework and rule sets are not formalized yet and is ever changing. So the above use cases are possible only with the assumptions of a free world.

P.S. – Facebook, Twitter or any other social media logos and brands are not associated with me apart from a pure “consumer” perspective. Each of the logos belong to their own registered brands respectively. 

Java 1.7 on CentOS 6.4


JAVAONCENTOSThis one is going to be a little of topic. Normally you would have noticed that my blogs are techno-functional in nature and are more aligned towards business strategy and growth. The origin of this one is no different but the post itself is little more technical. I have been a technology evangelist for quite some time and when I was doing my research in Big Data area, as the technology is open, what I found is that for a noob, its an extremely lucrative option to try our this technology. yet when I did plunge in it, I found that the technology infrastructure ecosystem has changed. Starting from the Linux move to support OpenJava instead of Oracle Java. Some of the installations of Hadoop systems like Cloudera or Apache, when I tried gave me some growth pain without Java. So I did a lot of googling (that’s an accepted word now a days and should be added to dictionary). Unfortunately I did not stumble upon a step by step guide. There were a lot but nothing was full proof. So I decided to write my own. This blog is an after effect of that experience.

Platform : I was trying to install into CentOS 6.4 version (community edition).
Source of Download : (trust me it took me 10 minutes to figure out the download page).
Version : I downloaded the 64 bit RPM version.
Caution : Accent the Agreement before downloading (I am kidding, you don’t have a choice not to accept and download).
I downloaded the version jdk-7u21-linux-x64.rpm

Being a noob (once upon a time, I was a techie), I used the visual package installer which is default with CentOS and had no trouble what so ever. If you are a experienced developer, you can go for YUM installation (you know what I am talking about).

However the steps after this are the most critical. The trouble is that OpenJave comes already installed with Linux and is default version of Java for it. So even though we installed java 1.7.0_21, it does not replaces the default java processing engine (SDK & JRE). Its important for us to replace that systematically.

You can know that by giving java -version command in the terminal. it will show what version of java is running and also other information about it. The next steps are to replace some of the critical utility with the Oracle Java equivalents.

You can use the following commands.

For a little visual person, Please use the following commands and follow the prompt to change the default value

alternatives --config java
alternatives --config javac
alternatives --config jar

Once this is complete, a good idea is to declare the PATH, CLASSPATH and JAVA_HOME to your profile or temporarily.

Now you are all set to install either cloudera or apache version of Hadoop. Most probably I will continue in my next blog with my experience after digging my note which I had taken while I did that. Until then, keep trying.