HP first tried its hand at integrated hardware and software appliances last year with the first iteration of AppSystem appliances. The company is now updating its AppSystem architecture with a new scale-out approach that enables enterprises to easily add capacity as needed.
Up first are AppSystems for Vertica, Autonomy and Hadoop.
Paul Miller, vice president, Converged Application Systems, Enterprise Group at HP explained to EnterpriseAppsToday that HP learned a lot from the initial AppSystem for Vertica that was launched in 2011. One big lesson: Customers prefer a modular approach rather than having a single big box to do everything.
According to Miller, the new AppSystem approach is like a "Lego" system that allows users to just snap in capacity as required. The system leverages the HP DL380 server system as its underlying hardware platform.
"If you think about your data center, you might start out with a Hadoop cluster with 10 servers in a rack and two Vertica servers," Miller said. "Then you can expand that and add another rack, allowing you to grow incrementally as your business needs grow, instead of needing to buy a full rack."
Software Side: Vertica and Autonomy
The new Vertica AppSystem includes the latest Vertica 6.1 release, providing increased performance from the HP Big Data analytics engine. Miller says the new Vertica AppSystem can inject and load data faster than the previous generation and increases the speed of access by up to 90 percent.
An Autonomy eDiscovery AppSystem helps round out the HP Big Data solution set. Miller noted that Hadoop Big Data on its own isn't enough for most organizations. Rather, many customers require a query engine around it to extract data from Hadoop.
"Customers want Hadoop to be the place where they pour all their unstructured data into," Miller said. "Hadoop doesn't really provide robust analytics and search tools, and that's where we can provide the Vertica and Autonomy systems alongside them."
While the Vertica and Autonomy components complement Hadoop, HP is also updating its AppSystem for Hadoop, adding an AppManager dashboard for Big Data management.
HP already partners with commercial Hadoop solutions provider Cloudera and its Cloudera Manager for Hadoop. The HP AppManager goes a step beyond what Cloudera does, looking at both hardware and software.
"Where Cloudera looks at Hadoop from a software layer, it doesn't have all the knowledge of the hardware," Miller said. "Cloudera can see for example that a node is performance constrained, but it can't tell you if that constraint is due to network congestion or hard drive issues."