| Pre-RAC | | | | grows. |
| Let's face it, today's IT environments are in constant | | | | While Oracle touts that you can build your system |
| flux. Acquisitions, mergers, global economies, and the | | | | with low-cost commodity parts, I am not too sure |
| general unrest in the marketplace continually put | | | | how many would want their mission critical |
| stress on IT infrastructures. It has become | | | | applications running on low-cost hardware. But what |
| increasingly difficult for IT organizations to predict, | | | | Oracle RAC really provides is for you to purchase |
| plan, and deploy proper computing environments that | | | | hardware that keeps in line with your budget, |
| will satisfy current and future requirements. | | | | business needs, and processing requirements. You |
| Corporations, when planning for database systems, | | | | need only buy today what you really need today and |
| seem to be a constant battle between purchasing | | | | then plug-in or remove hardware when needed. |
| systems that meet today's requirements against | | | | Increasing or decreasing your computing power can |
| systems that might potentially satisfy future | | | | be done "on-the-fly" without disturbing current |
| expectations. The unfortunate reality is that both of | | | | requests to the database. |
| these approaches are inadequately suited when | | | | This is especially important when thinking about |
| compared to the dynamic nature of the systems we | | | | database consolidation. Often times we don't really |
| must supply to the end user community. | | | | know how much processing power will be required. All |
| When systems are built against today's requirements, | | | | we really know is that when we look at our |
| without looking to the future, what's often left after | | | | dispersed systems we are wasting resources in the |
| requirements change, is a set of small discrete | | | | form of CPU, memory, and disk space. RAC allows |
| systems scattered around the organization each | | | | consolidate databases with the flexibility to add |
| supporting one or two applications. These | | | | additional nodes if we miss the mark. |
| independent systems typically are improperly sized, | | | | High Availability |
| too small or too big for the applications they support, | | | | High Availability can be summed up in two words, |
| and are likely to become islands of computing power | | | | "Always On". In a RAC environment, since multiple |
| that are spread thinly around a corporation. To make | | | | nodes share the servicing of requests to a single |
| matters worse these discrete systems are often | | | | database. By having additional nodes, you gain a |
| unable to facilitate communications across application | | | | higher probability that access to a database will |
| layers or even simply share data between | | | | always available if a node failure were to occur. In a |
| themselves. Look at any of these systems across an | | | | RAC configuration, if a node were to fail, the |
| enterprise and you will also see wasted or unused | | | | surviving nodes would perform recovery on behalf of |
| CPU cycles, wasted disk storage subsystems, and | | | | the failed node and then begin to service requests on |
| over-worked DBAs and Administrators trying | | | | behalf of that failed node. |
| desperately to keep some for of control on these | | | | Failover |
| systems. | | | | Closely related to high availability is failover. While high |
| Purchasing large centralized servers or systems larger | | | | availability ensures that access to a database is never |
| then what is currently required is no better. | | | | hindered through multiple nodes, failover is the power |
| Companies going this route often put every | | | | of an application to continue running even when the |
| application on one server or are purchasing more | | | | node that was servicing its' request might have failed. |
| computing power then they currently need. Designing | | | | Oracle RAC seamlessly and transparently provides |
| systems in this manner has proven to be very costly | | | | complete failover for an application. Oracle RAC |
| and wasteful of money and resources. | | | | maintains and hides all system-side failures from the |
| Downtime is always disastrous but having everything | | | | users and provides complete recoverability for an |
| run on one system has the potential of bringing down | | | | application. |
| every aspect of your business at the same time. | | | | Moving to RAC |
| This goes for unplanned downtime as well as planed | | | | Now you may be saying to yourself, "Yes, I know |
| downtime such as maintenance windows. | | | | about RAC and we are dedicated to building our |
| When systems are too large with unending | | | | Oracle RAC environment". Well this is all fine and |
| processing power, these systems are often taken | | | | good but do you really understand what you are |
| for granted and sloppy or ill-tested applications are | | | | getting yourself into? Do not foul yourself, RAC is a |
| deployed without ever going through some form of | | | | complex and difficult beast to handle and should only |
| performance or scalability tests. Down the road when | | | | be architected by experienced professionals. |
| additional CPU, memory, or disk resources are | | | | Someone who has done it, is not just trying to figure |
| needed, they are just not there. | | | | out how it works, and has the experience to pick up |
| Vigilant's prides itself in taking the time to consider | | | | the pieces when something goes wrong is invaluable. |
| your requirements, transforming them into viable | | | | Do you really want to trust your production data |
| solutions, and selecting the proper set of | | | | with someone who is just tinkering with the system? |
| components for your computing environment. | | | | Moving to Oracle RAC is not an overnight task. The |
| Both small discreet and large dedicated systems do | | | | full corporate IT structure needs to be involved and |
| share one thing in common. They both often have | | | | they each have specific tasks. From management |
| multiple single points of failure that limit their ability to | | | | setting the budget constraints, through database, |
| supply the high availability required in today's | | | | system, and network administrators that configure |
| 24X7X365 environments. All it takes is one | | | | these systems, and down to developers, testers, |
| component to fail and then, regardless of how little | | | | and users that must test, validate, and use the |
| or much money you have spent, your system has | | | | system. |
| just become useless and potential revenue is lost | | | | CRS, OCFS, ASM, TAF, FaN, FCF, NIC, and RAID, |
| because of the downed system. | | | | these are but a few of the acronyms and |
| Don't get me wrong. Small discrete and large | | | | technologies that must be learned and understood |
| dedicated systems have their place. They have | | | | before attempting a RAC installation. More |
| served us well for many years. But these | | | | importantly there are literally dozens of decisions to |
| environments just do not supply the required | | | | be made or floundered when configuring a RAC |
| infrastructure for scalability, performance, and | | | | environment. Again, this is not something you want |
| availability. For those environments needing more, | | | | to just inker with. Performance issues will arise. |
| they are turning to Oracle's Real Application Clusters | | | | Oracle RAC is a special case and Vigilant will guide |
| (RAC). | | | | you through the maze, assisting in the selection |
| What is RAC | | | | process of hardware and software. Vigilant will install |
| Before going much further it would be a good idea to | | | | and configure a RAC solution that is going to perform |
| give a brief explanation as to what Oracle RAC is. | | | | with the reliability and high availability you require. |
| Singly, and most importantly, RAC is nothing more | | | | From sizing the complete hardware for Server |
| than a single database that is accessed and shared | | | | Consolidation, Upgrading to Oracle 10g, or installing |
| by many Oracle instances running on separate | | | | and upgrading E-business suites, Vigilant has done it |
| systems. These separate but linked systems in a | | | | and can quote you a fixed up-front cost. |
| RAC environment are coined nodes in the cluster. | | | | Surviving on RAC |
| Each node in the cluster accepts requests from | | | | Putting a RAC system in place is only half of the |
| applications or users to query or update the single | | | | equation. Oracle RAC is a completely different beast |
| shared database. The shared data model employed | | | | that can not be maintained like the typical single |
| by Oracle RAC requires each instance on the nodes | | | | instance Oracle database that most DBAs are familiar |
| to maintain individual components but also | | | | with. |
| communicate through a sophisticated inter-process | | | | Vigilant understands that there is a difference |
| communication channel that synchronizes access to | | | | between single database instances and multi-node |
| the shared components, such as memory cache or | | | | RAC configurations. RAC performance issue should |
| disk arrays. | | | | not be approached in the same manner as a single |
| Vigilant Technologies is proficient in designing, | | | | instance database. Understanding is critical to tuning |
| architecting, installing, and testing all Oracle | | | | applications, meeting performance goals, and |
| environments including RAC systems. | | | | ultimately keeping your RAC installation up and |
| Why RAC | | | | running. |
| It is amazing how RAC solves all the issues | | | | There are unique and different considerations. |
| surrounding small discreet and large dedicated | | | | Networking protocols but be understood, hardware |
| systems. RAC allows for scalability, high availability, | | | | specifications and configurations are more critical, plus |
| and failover, each of which is critical in today's | | | | the monitoring of the different instances on each |
| demanding database environments. | | | | RAC node is different. A complete understanding of |
| Scalability | | | | how all the pieces fit together is essential for |
| While it is true that nothing is free, Oracle RAC allows | | | | troubleshooting performance and reliability issues. |
| you to scale up or down the number of nodes in | | | | If you do not have a qualified staff or are in the |
| your RAC environment such that it sufficiently meets | | | | process of obtaining certification or experience there |
| your immediate, short term, or future computing | | | | is no substitute for having someone around that has |
| requirements. You choose. Purchase what you need | | | | "been there and done that". Where RAC is |
| now and then, through the use of modular | | | | concerned, experience does count. |
| components, upgrade the system as your business | | | | |