|< Day Day Up >|| |
Like it or not, availability is more and more the rule, not the exception. As a database administrator, HA refers to a number of related issues that must be planned and prepared for:
Uptime literally refers to the fact that the database must be assumed to stay open around the clock. Database maintenance and administration has to occur on the fly, with no down windows.
A slow database is an unavailable database. Performance must be accounted for at all times, and performance increases must be accessible without downtime. This means scaling up, scaling out, and reorganizing in-flight.
Whether man-made or of the natural type, disasters happen and must be taken into account when preparing a database for production go-live. This means having a fail-over system in case of complete system loss, but it also means having the technologies in place that help recover from the more minor tragedies: user errors, bad memory, or database corruption.
If a HA solution is overly complex, it can become too top-heavy and collapse in upon itself and the management costs can outweigh the benefits of availability. So manageability must be considered a critical component of HA.
Which brings us to the book you hold in your hands. We have written this book based on what can be called a DBA-centric approach to availability. We concentrate on explaining technologies and practices common to the a database administrator, and that cover database uptime, database performance, and disaster planning.
This is not to say that we ignore hardware or application needs entirely; the web of HA is too interwoven for that. But a database-centered approach reveals many cost-effective approaches to availability that take into account the specific requirements and unique challenges presented by an enterprise-level relational database system. We want to help you in the transformation from a DBA to the HADBA: the High Availability Database Administrator.
Oracle provides a wide spectrum of HA technologies built in to the core RDBMS that you already use. What we offer in this book is a guide to leveraging those technologies that you have in your toolkit already but may not be using yet. While there are a multitude of HA options on the market, if you already paid for your Oracle license, there's a vested interested in exploring how much of the base functionality you can employ before widening your scope of inquiry.
Granted, there is a certain 'six or half dozen' argument that can be made: Oracle provides solutions that cost licensing fees, so money saved on hardware solutions might just be redirected. But we feel that leveraging database-centric
HA technologies provides the most cost-effective approach to HA, as well as making the tech stack manageable by database administrators.
Besides leveraging technologies already waiting at your fingertips, the DBA-centric approach to HA provides more opportunities to focus on the integration of multiple aspects of availability, instead of dealing with them in isolation. This book focuses on explaining the individual HA technologies separately, so that you can pick and choose that that fit your needs.
But we also emphasize the fully integrated package that can be provided by a database-centric HA strategy. We put Real Applications Cluster (RAC) and Data Guard together and discuss the unique challenges provided by this solution. We put a media backup strategy into the mix as well, showing how Recovery Manager (RMAN) and RAC and Data Guard work together to provide a full solution. We put Oracle Flashback technologies together with Data Guard so that you can leverage a database flashback quickly to reinstantiate after failover. The list goes on and on. When you focus on database availability tools, the challenges of integration quickly disappear.
If you haven't been barraged by the publicity yet, you should probably know that the little 'g' in Oracle Database 10g stands for 'grid.' Grid Computing is a philosophy of computing that posits, simply, that computing needs should operate on the same principle as utility grids. You do not know where your electricity comes from, or how it is managed; all you know is that you can plug your appliances in and you get as much electricity as you need, from a single lamp to an entire house of washers, dryers, and water heaters. Your computing needs should be the same: you plug in to the grid and you get your computational needs, your information needs, your application needs. You get as much as you need and do not worry where the computers are or how it is managed.
Such a grid would be based, of course, on the kind of availability that we have come to expect from the utility grid. When the electricity grid goes down, even for a single day, it makes headlines all over the country because it only happens every 30 years or so. Thus, for Grid Computing, one of the foundational pillars must be high availability.
For the Grid to become a reality, there are imminent challenges to uptime, and database administrators need existing solutions. The Grid guides us, but business dictates our actions. And to that end Oracle Database 10g provides real solutions for current availability challenges. In all actuality, these solutions are a natural evolution of concepts and technologies that Oracle has been building toward since the release of Oracle8i. While we have written a book that discusses how to use Oracle Database 10g, many of the options are available in Oracle9i or earlier. Rest assured, we are now officially done writing about 'The Grid.'
|< Day Day Up >|| |