6 Q&As: An Accenture Consultant’s Frontline View on 4 Years with Pivotal Products

December 11, 2013 Adam Bloom

ArjunShahHeadShotAt the recent SpringOne 2GX, we had the opportunity to have a “dialogue from the front line” about several Pivotal products. Our discussion was with Arjun Shah, a Consulting Manager and Technical Architect from Accenture who has worked with the consulting firm for about four years and spent almost the same amount of time working with Pivotal tc Server and Pivotal GemFire. Shah has played leadership roles on projects at an enterprise-level with large scale, global teams and focuses on infrastructure and platform services, largely around Spring.

What is the number one, top of mind advantage you have experienced with tc Server?

This is my favorite application server product and something I regularly recommend on consulting engagements. First off, I have to compare it to traditional, heavy-weight middleware products. Pivotal tc Server is very lightweight and easy to deploy—it can be up and running in minutes. There are so many intricacies with the competitor’s deployment models and operations. Also, tc Server uses very little memory at startup and has a very small disk footprint. Other products take a lot of time to boot up, are much more expensive, and use a lot more infrastructure resources to do the same job. We can all see how cloud providers offer Tomcat, not the traditional, enterprise middleware. I think this speaks for itself in terms of where the world is headed.

What about the financial perspective? How does tc Server provide an advantage there?

Of course, clients are always interested in a financial advantage and enterprise IT organizations want to drive costs down. Since tc Server is based on Apache Tomcat, we are talking about an open source engineering community. This provides financial advantages to companies who deploy Tomcat and is probably the reason it is the most widely used application server in the world. As a drop-in replacement for Tomcat, tc Server doesn’t pose a big risk to our clients if they ever need to replace it, and there is a low cost if they ever need to migrate. It’s also a very competitively priced product in terms of procurement. From an operations perspective, it uses infrastructure and virtual machines very efficiently. The server footprint is small, and tc Server’s additional, enterprise capabilities make it easier to deploy, operate, secure, and maintain.

You’ve worked with GemFire for almost 4 years. Could you summarize some of your key experiences?

Sure. I’ve been working with Pivotal GemFire for almost four years on multiple projects and multiple topologies. I’ve deployed single VMs running as a cache server as well as larger deployments like a 36-node grid spread across a WAN with redundant locators, redundant WAN gateways, and redundant cache servers. One of the implementations I was on had a GemFire network doing about 8000 reads a second with latencies in the milliseconds—that is pretty fast. We’ve used GemFire for traditional storage of key-value pairs as well as sophisticated deployments where GemFire’s native libraries are used to do things like write behind and read through to an enterprise database. GemFire is rock solid. I’ve also found that it requires very little maintenance and overhead.

Where are you seeing traditional architectures change?

The current state of web application architecture is definitely challenging the traditional enterprise mindset. Of course, it depends on the use case. We still see and propose architectures that are 15 to 20 years old like a 3-tier model. If you look at an area like ETL, this architecture is changing. For example, we have the traditional source system. Then a data warehouse or mart was added to run business intelligence and reports. In between, there is an extraction, transform, and load process or perhaps load then transform. While ETL is a proven model, map reduce frameworks are challenging the traditional approach here. Hadoop is great at processing or transforming large batches of data. With it’s distributed architecture, it can do so faster and more cost-effectively than traditional ETL.

Where do you think customers will see an impact with new architectures?

From what I see, there will be a significant impact to non-functional requirements. For example, volumes and latencies are non-functional requirements. Often times it is a struggle to get non-functional requirements nailed down like a 1-second SLA or put rigor around the ability to achieve the SLA because the architecture is assumed to work a certain way from the outset. These new architectures challenge the old way. They are purpose-built, promise, and deliver responsiveness on large data volumes. Now, companies are willing to do a proof of concept or see how their problems can be addressed in a lab. Each time this happens, companies gain confidence in how the new technologies will work in production. This is how new technology waves get adopted and cemented into an industry. We are seeing this happen right now in this space, and it is exciting.

After hearing about some of the new Pivotal products, what has caught your attention?

Many people at Accenture are excited about how VMware’s vFabric product line, Spring, and Greenplum have spun off into this new venture called Pivotal. Pivotal innovations are challenging the traditional mindset with things like Spring XD. Of course, I am also very interested in what Pivotal One and Cloud Foundry have to offer. It looks very promising, and I want to get the products in the lab and get rolling with them already!

Also, I just saw a demo where GemFire was integrated with Pivotal HD, the Hadoop platform. There was a lot of power in this concept especially within certain areas like the traditional data warehouse and ETL model and mindset we just discussed. It takes a little bit of time to wrap your head around using GemFire for transactions and real-time reads, storing data directly on HDFS for the long term, and using SQL for analysis and business intelligence. This is a pretty cutting edge technology innovation, but there are a lot of use cases we’re seeing now with significant amounts of data to persist. Colossal sets of data aren’t feasible with traditional approaches. Accenture wants to bring this to the table with our customers.

Learn more about the Pivotal productions mentioned in this interview:
Pivotal tc Server product page, documentation, and downloads.
• Pivotal GemFire product page, documentation, downloads, and articles.
• Pivotal HD and GemFire XD product page, documentation, downloads, and announcement.
Pivotal One, CloudFoundry.com, and CloudFoundry.org
• Pivotal Greenplum Database product page, documentation, downloads, and articles.

About the Author

Biography

More Content by Adam Bloom
Previous
What SpringOne 2GXers Have to Say on the New Spring Framework 4.0
What SpringOne 2GXers Have to Say on the New Spring Framework 4.0

Spring Framework 4.0 is now generally available for production systems with lots of Groovy, Java 8, HTML5, ...

Next
SAP HANA Service Broker Contributed to the Cloud Foundry Incubator
SAP HANA Service Broker Contributed to the Cloud Foundry Incubator

Service brokers are what make this provisioning and binding happen with ease, and today we are delighted to...

Enter curious. Exit smarter.

Learn More