HP continues its effort to convince IT shops it’s a good long-term bet, and to that end the company disclosed plans this week to more deeply integrate its enterprise product portfolio.
Hewlett-Packard’s new products include an AppSystem bundle for Hadoop, which integrates with Autonomy’s software to manage ‘big data,’ and a new way to handle SAN connectivity within Converged Infrastructure bundles.
But the vendor still has an uphill struggle when it comes to convincing customers who remain shaken by the company’s ongoing upheavals, the most recent being a promise by CEO Meg Whitman to lay off 27,000 employees over the next two years.
“We buy millions of dollars of server and infrastructure equipment from [HP], but that’s a far cry from saying we look for them to be the ones to handle big data for the future,” said one west coast-based IT director who spoke on condition of anonymity.
The IT director’s IT shop was also already heavily invested in Autonomy’s software prior to the HP acquisition, so joint offerings between the two might seem natural – but it isn’t.
“I think [HP knows]that there’s this lucrative space that talks about big data, and cloud storage, and cloud services, and they hope that they can bring the big HP name to bear, [but] that’s not working anymore,” the IT director said
Other IT pros acknowledge the bumps in the road, but say they’re still confident in HP as a strategic vendor.
“I still believe that
Continue Reading This Article
Enjoy this article as well as all of our content, including E-Guides, news, tips and more.
Reynders said he’s not a fan of pre-integrated “pods” such as converged infrastructure, from any vendor. But he said, “I think some parts of their cloud solutions are great, and I’ll have a look at some of them in the future.”
New App System for Hadoop connects with Autonomy, Vertica
HP said this week it would tackle the enterprise ‘big data’ trend with a new AppSystem for Hadoop, expected to ship in the fourth quarter. The infrastructure bundle will consist of a user-selected number of ProLiant DL360p master nodes, a cluster of 18 DL380e servers for data / worker nodes, and two A5830 top-of-rack switches. All of the server nodes will have internal direct-attached storage (DAS). IT shops will have the option of different flavors of Hadoop, including Cloudera, Hortonworks, and MapR.
To provision the system, HP has developed software it calls the Insight Cluster Management Utility, which it says can spin up to 800 nodes in minutes for golden-image deployment via cloning.
Meanwhile, version 10 of Autonomy’s Intelligent Data Operating Layer software can now be embedded in each Hadoop node to provide meta-analysis on data that comes out of Hadoop.*For example, understanding whether “apple” refers to a fruit or a company, or performing sentiment analysis in marketing comparisons. Version 6 of HP’s Vertica software also now contains new connectors to perform parallel analytics and queries on structured, unstructured and semi-structured data within the Hadoop AppSystem.
“This is the kind of offering HP is going to need to keep bringing to market to show [it is] properly leveraging different acquisitions,” said John Madden, an analyst at UK based firm Ovum Ltd.
Converged Infrastructure bundles get new method for connecting servers and storage
Later this year, HP will add a feature to its Virtual Connect FlexFabric module for its BladeSystem servers that could replace Fibre Channel SAN switches in Converged Infrastructure bundles.
A technique akin to Pass-Thru on the Ethernet side, what HP calls its Flat SAN technology, will allow for point-to-point connections between blade servers and storage without requiring a hierarchical network of Fibre Channel switches between them. Instead, a single cable would connect the Virtual Connect system and P10000 SAN. Multiple blade chassis can be connected using this method, or multiple SANs can be connected to the same blade chassis.
The point of this would be to eliminate the cost of Fibre Channel switches and reduce storage latency, HP said.
*Information added after initial publication.