Spin-up a bunch of servers in the Cloud, orĤ. Buy a preconfigured solution - Oracle’s Big Data Appliance, for example, which has all the software pre-installed along with connectivity to ODI, Oracle Database etcģ.
Download spark hadoop only or cloudera install#
Get hold of a bunch of physical servers (maybe, old PCs or blade servers), install Linux and Hadoop on them, and then do the configuration and setup manually.Ģ. So if we want to set up our own Hadoop cluster, there’s a few options open to us:ġ.
As such, you’re as likely to see Hadoop running on a cluster of Amazon EC2 server as running on physical servers in a datacenter, and in most cases the underlying OS running on those servers is Linux - most usually, Ubuntu 64-bit. Hadoop, as you’re probably aware, was designed from the ground-up to run across multiple nodes, with those nodes typically either being small, low-cost servers, or in many cases servers running in the “cloud”. So it it possible to set up a Hadoop cluster that gets a bit nearer to this multi-node architecture, so we can practice connecting to a cluster and not a single server, and we can see Hadoop process our queries across all of the nodes - as we’d see in real life, given that this low-cost MPP processing is the key benefit of Hadoop as a whole?
Download spark hadoop only or cloudera full#
Whilst the example worked though, I couldn’t help thinking that using Impala against a single node Hadoop install isn’t really how it’d be used in real-life in reality, if you used OBIEE in this way, you’re much more likely to be connecting to a full Hadoop cluster, with multiple server nodes handling the incoming queries and potentially gigabytes, terabytes or petabytes of data being processed. In this example, I connected OBIEE 11.1.1.7 to the Cloudera Quickstart CDH4 VM, which comes with all the Hadoop and Cloudera tools pre-installed and configured, making it easy to get going with the Hadoop platform. The other day I posted an article on the blog about connecting OBIEE 11.1.1.7 to Cloudera Impala, a new “in-memory” SQL engine for Hadoop that’s much faster than Hive for interactive queries.
0 kommentar(er)
