The new HttpClient provides a fluent interface that is quite intuitive. However, any non-trivial implementation soon starts to miss the flexibility of configuring the connection manager. As a pre-configured HttpClient can be passed into the fluent Executor each time, it is overhead to configure it each time.
posted at http://java.dzone.com/tips/fluency-and-control-httpclient
by Akber Choudhry – http://www.akber.com/
Last year, there was an opportunity to participate in IBM’s Cloud Beta and I wrote about some initial impressions.
IBM have now rolled this out with geographically dispersed data centres. Even with some terms being unique to IBM, the provisioning of instances and management of images should be familiar to those who have worked with Amazon’s EC2 and related cloud services.
This gives ‘true Blue’ customers an option to entrust sensitive systems to a cloud provider with whom there is an existing relationship. It will be interesting to see how this offering from IBM stacks up against managed services from IBM.
by Akber Choudhry – http://www.akber.com/
Networks, storage, and all other information technology infrastructure are fairly complicated things and it takes substantial planning and flawless execution to get it right.
Project managers and analysts do great work to ensure that what runs on these infrastructure systems is properly specified and delivered on time.
And with all this hard work, and whether data is delivered over 4G mobile or 10G ethernet, all that users see is the App. Nothing matters if the app is not there when the user wants it, and how the user wants it.
The App is King.
by Akber Choudhry – http://www.akber.com/
Finally, finally, finally! Some long-needed innovation from IBM, and something other than share buybacks that may boost the long-term viability of IBM software.
by Akber Choudhry – http://www.akber.com/
IBM is beta-testing its own cloud at https://www-180.ibm.com/cloud/enterprise/beta/dashboard, after a successful experiment with Amazon EC2. As a company with hardware and operating system roots, it was smart to get on to the cloud bandwagon, which is essentially the commoditization of hardware and operating systems.
A plain Suse instance was not a problem. However, after some initial trouble launching a WebSphere Portal/WCM instance in the wrong-sized machine, I was able to launch an instance in a ‘small’ machine. I only tried it once, but a WebSphere sMash instance did not launch. My first impression is good — for a beta-testing environment, and quick proofs of concept, it is very handy.
If you know Amazon EC2, the management console should be pretty much intuitive. The only time I had to read the good online help was when I needed to know the exact user to be used for the SSH session into an instance.
Let us wait and see how IBM manages to position this offering once it is out of beta.
by Akber Choudhry
Quick Post: Just tried this today and it worked:
Travel and meeting sites can give you a link that can add an entry to your Google Calendar, among ICAL and other links thare are also provided. This is very handy, as it takes care of time zones etc. and works out quite well. The link is usually of the type:
http://www.google.com/calendar/event?action=TEMPLATE&text=. . . . .
If you are on your company’s Google Apps hosted calendar, this link would not work and it would ask you to log in to a regular google account, not a Google Apps for Business calendar. If you copy the link location and add after ‘calendar/’:
http://www.google.com/calendar/hosted/xxxx.com/event?action=TEMPLATE&text=. . . . .
where ‘xxxx.com’ is your Google Apps domain, it works like a charm. Kudos to Google URL architecture and those that enforce it.
by Akber Choudhry
IBM WebSphere Commerce 7.0 comes with a substantial set of marketing activities: Web activities, e-mail activities and Dialog Activities. In summary, Web and e-mail activities push content along, while dialog activities respond to specific shopper behaviour. Now, we may also want to push content based on some analysis of past shopper behaviour, whether specific to that shopper, or in general. Combine the two with some custom enhancements that put together some logic and data storage and you may have a specialised recommendation engine for your specific business needs.
by Akber Choudhry
If the first generation of eCommerce was the shopping cart, and the second wave was enabling all sales channels in addition to on-line sales, then we are now beginning a new trend: the creation of an ecosystem around offered products and services.
While the term ‘social commerce’ is still being refined, it is becoming apparent that with a wide portfolio of products there is a need to facilitate the interaction of communities — vendors, customers, and even random visitors — with your web presence. It has long been realised that knowledge within a company needs to be unlocked and made accessible to all its employees. The current challenge is that the same knowledge, if marshalled in the right way, can enable internal and external communities to use it, build upon it and add to it, thereby positioning your products in the proper context in the right ecosystem.
Doing so in a cost-effective manner and with tangible results remains the challenge.
by Akber Choudhry
I took IBM WebSphere Commerce 7.0 for a spin over the past couple of days. First impressions:
To be fair to IBM, they have done a good job, and not tinkered much with the market leader, but maybe they should start working on the next major release with a grounds-up re-design:
Well, we can dream, can’t we?
by Akber Choudhry
Caching provides persistent business data and volatile session data from memory instead of from database and/or disk access. This improves performance. Clusters of application servers also provide performance and scalability. The two techniques clash as one application server’s cache is not in sync with others’ caches.
A solution for this clash is through Network-Attached Memory (NAM), which is very similar in role to a SAN (Storage Area Network) that is used for shared disk access. External memory (spread over one or more cache servers) is made available to each application server, who think of it as their own memory. Resources are not wasted and each application server’s view of cached content is identical – as it is in fact the same data in memory that is visible to all.
Read-only, Write-through and Write-behind
Just as a SAN provides continuous service even when individual disks or arrays crash, NAM should provide this shared memory even when one of the NAM servers crashes or is taken out of service. This is distributed caching, or a data grid. Generally, these are read-only, with write-through (writes pass through directly to disk or DB, updating the cache), and rarely, they may be write-behind: accept data commits from applications and ‘write-behind’ so that the application finishes its commit or update, oblivious to whether the data may not have been written to disk or database yet, and the grid will write it out when it deems appropriate.
Short-listing Products
Common products out there are Terracotta, Gigaspaces, JBoss Infinispan and Oracle Coherence
Working on a real-life high-performance portal, I came up with these requirements to create a short-list: