Now that software vendors' support for the LAMP (Linux, Apache, MySQL and PHP/Python/PERL) stack is becoming the status quo, it's time to extend support to Java platforms and a broader world of open source products, according to Eero Teerikorpi, CEO of Emic Networks. In this interview, Teerikorpi describes the benefits of LAMJ support and MySQL 5.0's stored procedure functions. He also clears up some nagging misconceptions about grid computing and the different types of Linux clusters.
At LinuxWorld Boston Expo & Conference today, Emic rolled out support for Java (JSP and J2EE) application servers in its Emic Application Clustering line, which will now include specialized LAMP, LAMJ, MySQL and Apache clusters. The company also unveiled a bundled, integrated clustering package, designed to help enterprises speed up their adoption of open source, mission critical applications.
What benefits would clustering bring to companies that are running LAMP?
Eero Teerikorpi: Scalability and reliability for mission critical applications have been part of proprietary, expensive solutions, but have been missing to date from the open source stack. Application clustering [can] meet [those] requirements for high availability.
What is the next horizon after LAMP support, in terms of extending the Linux platform in an enterprise?
Teerikorpi: We believe it is essential to extend our reach to open source stacks that include Java. Emic now offers application clustering solutions and support for JSP and J2EE application servers. With this, Emic covers the LAMJ stack, substantially extending the universe of applications that can benefit from greater scalability, manageability and availability.
Also, the 5.0 release of MySQL will add stored procedure functions, meeting a requirement for enterprise deployments. This will propel more MySQL and LAMP/LAMJ applications into the enterprise.
How would you define grid computing? I ask because I've heard users equating it with clustering.
Teerikorpi: There is some amount of confusion over these terms and some vendors use these terms interchangeably, thereby adding to the confusion.
Grid computing is often used in conjunction with utility computing. The metaphor is computing services provisioned just like electricity or water. This term, then, relates to how vendors will charge for the computing services, and relates to how entities will access computing services from out-sourced datacenters.
Grid also means on-demand application usage. For example, at the end of a fiscal year when there is heavy demand for accounting services, these applications will automatically extend their computing resources to service this higher demand. Another use of grid computing is in scientific applications and is most often used for solving very large and very complex mathematical problems.
Is there similar confusion over what makes up a cluster?
Teerikorpi: Clustering is also a term that has many meanings. A Beowulf cluster is a series of low-end machines able to process, in parallel, large complex computations. In essence, it is building a supercomputer from low-cost hardware and open source operating systems.
A storage cluster is normally a large disk farm sitting behind any number of servers that has one copy of the data and all applications/users see this same copy. They are typically very expensive to build and operate.
An application cluster is typically middleware software sitting between the client applications and backend servers in order to perform load balancing and failover, provide scalability and reliability, and logically replicate data to insure that all users see the same data at the same time. Application clusters are focused on providing high availability to enterprise business applications.
Which is better, grid or clustering?
In the end, there isn't one answer as to whether grid or clustering is the better solution. Each entity needs to decide what problem they are trying to solve, and zero in on the right set of potential solutions be it a Beowulf cluster, an application cluster, or grid computing.
Why is Emic taking a bundled approach to its clustering offerings?
Teerikorpi: Our integrated solutions are designed to provide flexibility to our diverse user population. Our solutions are available as integrated offerings to support the entire LAMP or LAMJ stack or as stand alone products to support specific open source components, such as MySQL or Apache.
Our view on this has been shaped by what we have heard in meeting with our customers and by our evaluation of the broader marketplace's requirements. At LinuxWorld Boston, we are announcing new integrated solutions that enable companies to implement our solutions either across the entire stack or for individual open source components.
What are the open source components of Emics cluster bundles? What's proprietary?
Teerikorpi: In addition to Emic lamp/cluster for the LAMP stack and Emic lamj/cluster for the LAMJ stack, we offer Emic m/cluster for MySQL and Emic a/cluster for Apache. Our clustering solutions are not open source themselves, but are designed to sit on and extend the open source stack. We also make contributions to various open source projects.
Do you expect to see more bundled software packages for Linux that include open source software in the coming year?
Teerikorpi: Yes, with the kickoff of SourceLabs and SpikeSource providing certification and certified open source stacks, we would expect more organizations will follow this model. This is a great benefit to open source adoption, as it reduces the risk for entities in adopting open source stacks in more applications.