Open Source Technology in a Down Economy

Untitled Document

One of the hardest hit areas in IT during any slowdown is application development and application stack infrastructure -- application servers, clustering software, caches, databases, etc. During tough times, most application expansions are put on the back burner, and anything that does not have an immediate impact on revenue or cash expenses is placed on hold.

This year's slowdown is no exception. Most of our customers continue to pursue critical projects with a renewed focus on reducing costs in the application stack. Their business needs are not going away, and in fact the end users our customers serve are even more demanding in times like this. Bottom line: everyone needs to be able to do more with less.

Much of the time, the most scrutinized components are understandably the most expensive -- application servers and databases, meaning the software for these functions and the hardware kit needed to run it. On a per CPU or per server basis, the license fees for proprietary database and application servers can be tens of thousands of dollars, with extra thousands added for various upgrades to enterprise functionality. One customer company recently revealed to us the pricing it endures from a certain proprietary database vendor and said, "Please, help us reduce this cost!" Not only are license costs high, but in many cases these proprietary products drive unnecessary hardware purchases, which subsequently increases the concomitant expense of facility and energy costs associated with running servers, as well as the expense of network hardware needed to connect them together.

Fortunately, the industry has made tremendous progress since the last economic recession, both in terms of technology and business model innovation. The enterprise can now turn to products that drive these costs down and help everyone successfully survive the current downturn. For example, today there are excellent open source alternatives to the monolithic JEE application servers, including WebLogic. Indeed, lightweight components like Spring, Jetty and Tomcat not only save upfront costs but can save significant development and ongoing maintenance effort. Similarly, open source databases such as MySQL, when complemented by high performance in-memory data management solutions like distributed caches and network attached memory, can take the handling of certain types of data completely out of proprietary databases. This can save a considerable amount in license fees and ongoing support costs in many organizations, and usually eliminates an expensive performance bottleneck.

Many have long considered open source to be simply a cheaper "me too" software that mimics functionality of proprietary products and brings no original value. In fact, open source vendors and their complementary communities are in the vanguard of innovation these days. For instance, network-attached memory (NAM) provides a service transparently beneath the Java language to thousands of connected clients; objects are manipulated and kept consistent, but can be massively scaled out, saving companies significant time and money. The computer science in our product is definitely not a copy-cat affair, nor is that of another open source product developed by SpringSource, one of our partners. SpringSource is driving some truly meaningful change, often more simply than Java Enterprise Edition (JEE), in the way Java developers approach building applications. These innovations, and many others of equivalent import, are all open source.

While reviewing spending priorities, an activity that has become familiar to all of us, one must consider the following questions (the results could be quite revealing, and just might impress the accountants).

Q: How and where can one take advantage of this movement?

A: Take a look at where you're spending and think about whether you really need Oracle or WebLogic in a particular application.
1. Have you purchased premium features for these products that you initially felt you needed but which you have not been able to deploy effectively?
2. You might be paying for features you don't even need, or features you can't even use. You'd be shocked to hear how often that happens.

A: Look at how you're developing applications:
1. Are vendors pushing development models that are overly complex?
2. Are your engineers are jumping through hoops to handle infrastructure issues when you would rather have these valuable resources writing features that customers want?
3. Would a simpler solution save time and money and still get the job done?

A: Look at the types of data in your application and determine whether you have the proper infrastructure in place to handle each type of data in a cost-effective manner.
1. For example, to replicate temporary data about users' interaction with an application, are you using a database to increase reliability even though the data is short-lived and will never be needed for any business reason in the future?

We had a customer in the media/publishing industry that saved $2M in short order by pulling some types of data out of the database altogether, while simultaneously cutting the load on their database server in half, from 70 percent to 35 percent utilization, leaving them plenty of room for growth without additional database license expense.

Yes, we're in the midst of a tough market; however, for companies willing to adopt new technologies that drive simpler approaches and lower costs while maintaining high quality of service, the economic slowdown may actually create key opportunities to drive badly needed efficiency gains. These opportunities are manifested in the move away from expensive heavyweight application stacks to a more efficient, flexible and open approach to building enterprise applications. The primary beneficiaries of this movement will be the customers and of course the companies behind these lightweight OSS products. In this movement, the customers will be left with a greater share of the benefit than they had with the ancient regime of proprietary vendors.

About the Author

Amit Pandey currently serves as the CEO of Terracotta. Prior to joining Terracotta, Pandey was a vice president and general manager of both the Data Management Business Unit and the Content Delivery Business Unit at Network Appliance. Before Network Appliance, he worked as a senior manager at McKinsey & Company focusing on technology strategy and operations effectiveness with Fortune 500 companies. Pandey has Masters in Management and Engineering from MIT and a BA in Electrical Engineering from Brown University. He grew up in Gabarone, Botswana, and currently lives in Portola Valley, CA.

More by Amit Pandey

About Terracotta

Terracotta is open source infrastructure software that provides affordable and scalable high availability for Java applications.