What Exactly Is the Grid?


The term grid is very much a part of Oracle Database 10g. In fact, the "g" in "10g" comes from the word "grid," just as the "i" in "8i" was derived from "Internet." And just as Oracle 8i heralded the arrival of Oracle Corporation's flagship RDBMS as a major player in shaping the Internetsomething it accomplished by hosting the massive databases behind the various websitesso too does Oracle hope that Oracle Database 10g will help shape the notion of the computing grid in today's IT world.

So what exactly is the grid? And a deeper question: Why do we need a grid? The answer lies in looking at a familiar example, and considering how we can adapt and use that model in computing.

A Familiar Example: The Electricity Grid

The electricity grid has been around for such a long time that we do not even consider its existence as a grid. However, this grid exhibits most of the characteristics that we would like to have in a computing grid. Some of the characteristics of an electricity grid are noted next.

The Electricity Grid from a Consumer's Point of View

First, we will consider the electricity grid from a user or consumer's point of view:

  • Electricity is available any time you need it. When you come home in the evening and switch on the light, you expect it to worknot to have to wait for the system to "boot."

  • Electricity is available at the right levels of voltage and frequency. Whether it is at 110 volts and 60Hz or 220 volts and 50Hz (as it is outside the United States), you are pretty much guaranteed the right level of voltage, except when the load becomes high and slight variations creep in.

  • Consumers pay for electricity using a "pay-per-use" model. In other words, you pay for what you use, not for the excess capacity that is available in the system. On the other hand, there are some limits on usage, and you may be charged higher rates if you use more than the norm. Rates may also vary by season.

  • Consumers don't have to manage or worry about the availability of electricity in built-up areas. Electric connections are part of every developed location. As a consumer, you just have to purchase the appliance of your choice and plug it in. The choice or brand of appliance does not matter. That said, the appliance does have to conform to certain specifications of voltage and power consumption.

  • Modern life, as we know it, comes to a standstill if this crucial component fails. This became vastly clear on August 14, 2003, when a massive blackout swept across the Eastern United States and parts of Canada. Although individual consumers can protect themselves to some extent, the majority of everyday appliances and systems that are taken for granted simply stopped working. Extended outages, both in terms of time and geographies, will result in massive losses to an individual's, organization's, and even a country's economy.

The Electricity Grid from a Producer's Point of View

From an electricity producer's or distributor's point of view, the following characteristics apply:

  • Individual producers and distributors must agree on and abide by some standards, and know each other's production and distribution capacities and capabilities. The producers and distributors may be entirely different organizations, but have to work together as a cohesive unit.

  • Both producers and distributors do not really care about end-use, but do care about short-term and long-term demand. They are responsible for day-to-day operations as well as forecasting and planning. Most of all, they are responsible for keeping the grid available at all times.

  • To enable this, producers and distributors use sophisticated monitoring tools that can track usage and respond to fluctuations in demand. For example, when demand increases in the evening, additional generators may be started up and placed online; when demand decreases, these same generators may be taken offline. At times, this excess generation capacity might be used to feed growing demand in another time zone.

  • Most of these actions happen automatically, but are overseen and monitored by skilled human operators. Usually they are managed using a central monitoring facility that can look at the entire electricity grid using a single, consistent, up-to-date view.

  • Although glitches and hitches in electricity generation and distribution are day-to-day events, most of these can be smoothed out. Only when a massive power outage occurs, such as the one described previously, does the consumer even become aware of such control centers. Again, when events occur, the operators and planners are ultimately held responsible.

  • Although it is expedient to keep a grid as well connected as possible, the failure of one component could bring down the whole grid as occurred in the massive blackout noted previously. As the grid evolves, disaster-recovery and backup planning as well as forecasting and constant checking and testing needs to keep in sync with these changes.

We have described this familiar grid in detail so that you have an idea of the kind of commodity that computing should become. We are already seeing this in some aspects on the consumer side of computing. For example, specifically in the area of availability and accessibility, anytime-anywhere Web access from mobile devices and always-on wireless PCs at home or work are common in most developed nations.

Similarities in the Computing Grid

A computing grid aims to have similar characteristics to an electricity grid. In short, the following should be possible in a computing grid:

  • Computing power should be available at all times to all users, regardless of geography and time zone. High-availability systems enable such computing power.

  • Applications should be able to run regardless of the platform on which they are located, and be able to access data stored anywhere in the grid. There has to be a high level of interoperability at all levels: hardware, software, middleware, and ultimately data.

  • Applications should respond in the same fashion and level of response, regardless of current load and location of use. Additional computing resources should be brought online as required and also taken off as required.

  • The costing model should be of the "pay-per-use" variety. Users should not have to pay for excess capacity that is held in reserve.

  • From an administrative and operational point of view, the computing grid should automatically adjust and heal itself, and be available for both monitoring and control from a single, consistent tool.

  • When a computer system is down, depending on the use of that system, it might spell doom for the organization. For example, when a development server goes down, your internal programmers are unproductive. But if the server that is your online store and your only source of revenue goes down during a peak sales period, your organization may not be able to recover from lost customers who moved their accounts and allegiance to a competitor. Connected computing systems in the grid are so complex that they will become extremely difficult to manage. Complexities also multiply exposure to mismanaged changes as well as extend recovery periods due to complexities. The requirement, however, is that recovery should be quick, with as little impact as possible, and as simple to perform as possible.

Virtualization and Provisioning

Just as with any new technology, buzzwords abound in grid computing. We will define two of them here in plain English: virtualization and provisioning.

Virtualization refers to the methods and tools used to expose the services of every physical and logical entity in a grid. For example, a four-node Oracle Database 10g RAC is virtualized using a service name. A user connects to the virtual service, but any one of the four instances virtualized by the service name could be serving the user.

Provisioning is the controlled allocation of resources or privileges when required. For example, in disk provisioning, ASM disk groups can be used to provide storage space for specific types of files with specific protection. For computing-resource provisioning, policies can be used to provide controlled resources at the right time using the Oracle Scheduler and the Database Resource Manager.


In short, the basic idea of grid computing is the notion of computing as a utility, analogous to the electricity grid described previously. A user in the grid does not care where the data is present or where computation is done, as long as it is done and information is delivered whenever required. From the provider's side, the grid is about virtualization and provisioning. All your resources are pooled together and provisioned dynamically based on the needs of the business, thus simultaneously achieving better resource utilization.

A number of recent industry trends helped shape the growing presence of the computing grid. Hardware trends include low-cost, high-volume processors that power server blades with low power and size footprints that reduce cost and increase the density of servers. This in turn ultimately leads to lower costs in running IT services and data-center operations. Network attached storage (NAS) and storage area networks (SANs) provide terabyte-sized storage, while gigabit ethernet and infiniband networks connect them all. Software trends include the availability of open-source software such as the Linux operating system and the Apache Web server, both of which can run on these inexpensive boxes. As a result, an increase in the number of servers does not necessarily mean an increase in software license costs. Both these downward trends, combined with the rapidly changing and increasing computing requirements, have necessitated and enabled the computing grid.

Simply put, grid computing can be defined as applying resources from many computers in a network to a single problem, usually one that requires a large number of processing cycles or access to large amounts of data. As you saw in previous sections, the computational power grid is analogous to the electric power grid. Basically, grid computing allows the coupling of geographically distributed resources to offer consistent and inexpensive access to resources irrespective of their physical location or access point. A wide variety of distributed computational resources varying from supercomputers, computer clusters, storage systems, and data sources can be connected via the Internet or dedicated networks and presented to the user community as a single, unified resource.

In other words, grid computing enables devices, regardless of their operating characteristics, to be virtually shared, managed, and accessed across an enterprise, industry, or workgroup. The keyword here is virtualization, which means that these resources will appear as a single, virtual piece of computing equipment. This virtualization of resources places all the necessary access, data, and processing power at the fingertips of those who need to rapidly solve complex business problems, conduct computation-intensive research and data analysis, and operate in real time. For example, through grid computing a company with about 2,000 desktop computers can use those computers to harvest nearly one teraflop (one trillion floating-point operations per second) of computing capacity. Thus, grid computing promises to consolidate, simplify, and fully utilize the available computing resources.

Now that we have defined what a computing grid should have, let us see how Oracle Database 10g fits into the picture. Before that, however, let's look at a little history of grid computing.

Grid Computing: A Little History

Grid computing is not a new concept. Universities and research institutions have been using grid-computing technology for decades. This concept has recently started making inroads into the business market, however. When applied to an enterprise, this kind of computing is specifically known as enterprise grid computing, in order to clarify that it supports the IT requirements of an enterprise.

According to a recent survey conducted by Insight Research, awareness of grid computing in the business domain is rapidly growing, with 37% of respondents saying they are now evaluating the case for grid computing. Insight Research expects global enterprises to spend $4.9 billion (U.S. dollars) on grid computing in 2008, which is nearly 20 times higher than the 2003 spending level of $250 million (U.S. dollars). By 2007, market researcher International Data Corp expects the world to spend $3 billion (U.S. dollars) on the software needed to virtualize grid components such as storage, security, and applications. That's a doubling of the $1.5 billion (U.S. dollars) spent globally in 2002. This translates to sales on the horizon for providers and early implementers of this technology, including Oracle Corporation. Hence the suffix change from "i" to "g" in the RDBMS versioning.

Oracle, along with a few other organizations in the computing world, have formed a consortium named the Enterprise Grid Alliance (EGA). The EGA is an open, nonprofit, vendor-neutral organization formed to develop enterprise grid solutions and accelerate the deployment of grid computing in enterprises. The other organizations in the consortium include EMC, Fujitsu-Siemens, HP, Intel, Sun Microsystems, NetApp, and NEC. Hopefully, the EGA will develop and implement standards that are beneficial to the user community rather than to themselves.



    Oracle Database 10g Insider Solutions
    SUSE LINUX Enterprise Server 9 Administrators Handbook
    ISBN: 672327910
    EAN: 2147483647
    Year: 2006
    Pages: 214

    flylib.com © 2008-2017.
    If you may any questions please contact us: flylib@qtcs.net