To simplify the roll-out of clusters for big data applications, Skytap is now offering pre-configured Cloudera Hadoop templates that can run in the company's public cloud.
There is a growing interest in big data using Hadoop, but to even start learning, experimenting and developing proof-of-concepts requires a lot of up-front investment in hardware, and it is also fairly complicated to download and configure, according to Brett Goodwin, vice president of marketing and business development. Skytap wants to address that with the introduction of its pre-configured templates, and allow enterprises to get their feet wet.
"Almost every production Hadoop cluster I have seen has been running on the metal itself for performance reasons, but for development, testing, prototyping, experimenting and learning, this virtual Cloudera Hadoop cluster is a great option," said Goodwin.
For example, one user developed new MapReduce components, validated them in Skytap's cloud and then moved them to its own staging and production cluster, according to Goodwin.
Hadoop in the cloud can also be used for temporary workloads.
"If you don't mind that there is a little performance overhead introduced by the virtualization layer it is a heck of a lot better than going out and buying 50 nodes you are not going to need long-term," said Goodwin.
Skytap's big data offering is based on the Enterprise Free edition of Cloudera's Distribution, including Apache Hadoop (CDH) open source distribution, which can be used to deploy and manage physical or virtual clusters of up to 50 nodes. Using Skytap's multi-VPN capability, enterprises can create a hybrid cloud that runs Hadoop cluster nodes both in the data center and in Skytap's cloud. The whole environment is can be managed as a single, unified cluster on-premise or from within the cloud using the Cloudera Manager, according to Goodwin.
The clusters can be shared with anyone, even if they are not a Skytap customer, using a secure URL.
The Hadoop functionality is included in Skytap's monthly subscriptions, which cost from US$500 per month. In addition to that users also pay for the time they use their virtual servers.
Skytap was founded in 2006, and launched its public cloud in 2008. It currently runs in Savvis data centers in Seattle and Virginia. The company has its own hardware, on top of which Skytap runs VMware's ESX 5 hypervisor along with an operating system and orchestration layer that was developed in-house, according to Goodwin.
Send news tips and comments to [email protected]