10.07.2015 Views

Expert Oracle Exadata - Parent Directory

Expert Oracle Exadata - Parent Directory

Expert Oracle Exadata - Parent Directory

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

C H A P T E R 15Compute Node LayoutThe term node is a fairly generic one that has many different meanings in the IT industry. For example,network engineers call any addressable device attached to their network a node. Unix administratorscommonly use the term interchangeably with host or server. <strong>Oracle</strong> DBAs often refer to a database serverthat is a member of an RAC cluster as a node. <strong>Oracle</strong>’s documentation uses the term compute node whenreferring to the database server tier of the platform. This chapter is about the various ways in which youcan configure your <strong>Exadata</strong> compute nodes, whether they are members of an RAC cluster (nodes), ornonclustered (database servers).It’s a common misconception that an <strong>Exadata</strong> rack must be configured as a single <strong>Oracle</strong> RACcluster. This couldn’t be further from the truth. In its simplest form, the <strong>Exadata</strong> database tier can bedescribed as a collection of independent database servers hardwired into the same storage and the samemanagement networks. Each of these servers can be configured to run standalone databases completelyindependent of the others. However, this is not commonly done for a couple of reasons—scalability andhigh availability. <strong>Oracle</strong> RAC has historically been used to provide node redundancy in the event of nodeor instance failure, but <strong>Oracle</strong> marketing has made it clear all along that the ability to scale-out has beenan equally important goal. Traditionally, if we needed to increase database performance and capacity,we did so by upgrading server hardware. This method became so commonplace that the industry coinedthe phrase “hardware refresh” to describe it. This term can mean anything from adding CPUs, memory,or I/O bandwidth to a complete replacement of the server itself. Increasing performance and capacity inthis way is referred to as scale-up. With <strong>Exadata</strong>’s ability to provide extreme I/O performance to thedatabase server, bus speed is now the limiting factor for scale-up. So what happens when you reach thelimits of single-server capacity? The obvious answer is to add more servers. To continue to scale yourapplication, you must scale-out, using <strong>Oracle</strong> RAC. Nonetheless, understanding that the databaseservers are not tied together in some proprietary fashion clarifies the highly configurable nature of<strong>Exadata</strong>. In Chapter 14 we discussed various strategies for configuring <strong>Exadata</strong>’s storage subsystems toservice specific database servers. In this chapter we’ll take a look at ways the database tier may beconfigured to create clustered and nonclustered database environments that are well suited to meet theneeds of your business.Provisioning Considerations<strong>Exadata</strong> is an extremely configurable platform. Determining the best configuration for your business willinvolve reviewing the performance and uptime demands of your applications as well as ensuringadequate separation for development, test, and production systems. Here are a few of the keyconsiderations for determining the most suitable compute node layout to support your databaseenvironments:497

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!