CRN Interview: Thomas Kurian, Senior Vice President, Oracle

Thomas Kurian, senior vice president of Oracle server technology, disclosed details of Oracle's 10g strategy to provide software to enable low-cost, grid computing in an interview with

CRN

Senior Editor Elizabeth Montalbano on the eve of the OracleWorld, being held in San Francisco. In this excerpt from the interview, Kurian discusses which technologies Oracle's 10g database, application server and Enterprise Manager software will contain, and how Oracle's strategy differs from those of its competitors.

CRN: What are Oracle's plans for its 10g software strategy for grid computing?

Kurian: What we're talking about with grid computing is twofold. The first is that today if you look at organizations and the biggest element they're looking at from an IT infrastructure point of view is that the cost of their IT infrastructure is very, very high. Additionally, they have lots of excess capacity in their IT infrastructure just sitting there, whether that's hardware or storage or memory. ... The notion of grid computing and Oracle's strategy is essentially to allow you to use pools of cheaper, lower-cost computing resources, both storage and servers, pull them together so you can run Oracle software, both our application server and database, on these low-cost resources so you get the same quality of service that you had before, but now using drastically lower-cost computing resources.

The second element that we focus on actively is that when you go to lower-cost, more modular computing resources, because you pulled them together, you have more efficient capacity utilization, less wasted capacity; second [you have] the ability to add capacity in incremental, modular units; and third, something we focus a lot on, the constant challenge of while the hardware is cheaper, the cost of managing the hardware is higher because you have lots of small systems now to manage. The third element of our strategy is to allow you to get drastically lower cost of management, both when you manage a single system and when you manage a large number of systems ,[to provide] more intelligent software that reduces your need to actually manage it and [also] by automating tasks across many, many systems, to our centralized enterprise manager facility, also called grid control.

CRN: When will the pricing for the products and the software itself be available?

Kurian: We will come out with pricing sometime after Oracle World ... the timing has not been finalized yet. Availability of all products will be by the end of the calendar year.

CRN: Could you talk about what technologies will reside the products that will create these grid-computing capabilities?

Kurian: In both the application server and database, there are a set of services that allow you to use the grid-computing infrastructure. What those services do is to both understand what resources are being consumed on the grid, a resource manager that has an idea what is being consumed on the infrastructure, and second, a policy manager that says, based on priorities that are set by an administrator, here are the most important jobs and the less important jobs. Then [there is] a job scheduler as well as a workload manager that directs the infrastructure to say, this job gets more priority than other jobs. It's a set of technologies, an intelligent software layer, that allows you to run both our database and application server.

On the database side, there is also a couple of different capabilities that we're announcing around automated storage management. [These] allow you to automate management of pools of storage and eliminate I/O tuning and a lot of those kinds of things so when you run an Oracle database on storage you get automatic backup and recovery and storage management. [You also get] continued evolution of the clustered capabilities in the database, the real application clusters, now extended to run a larger, fan-out environment. So [you can] run [the software] on many more nodes as opposed to fewer nodes in a clustered environment. On the application server side we're also offering additional capabilities over and above the grid infrastructure, which is around some new technologies for scalability and clustering, new facilities for high availability and how you centralize, how you get all these services for enterprise portal, packaged applications, how you manage application lifecycle on this infrastructure.

Then the third product, Enterprise Manager, we're really announcing two or three different new capabilities. The first is much broader capability for software provisioning and lifecycle management on this infrastructure. Provisioning and lifecycle management means installing, configuring, upgrading, patching ,automating all those tasks. The second is centralized administration of both the application server and the database across the grid, so you can manage all of your application servers and databases across the grid environment. And third is some advanced capabilities on performance monitoring, tuning and optimization in a grid environment.

CRN: Are you talking about software management only, or is there a hardware element as well?

Kurian: There are specific areas where we manage hardware as well. So for example in the storage management area, we do manage some of the storage components, such as the disks and so on. In the Enterprise Manager environment, in the context of an application server, there are some other components you need when you run applications, such as load balancers, routers and firewalls and things like that. We do have some capability to manage some of those components as well within Enterprise Manager. But primarily [we manage] software.

CRN: What's the reasoning behind Oracle going in this direction, since there is a lot of competition out there already, such as IBM, Hewlett-Packard, Computer Associates, even Sun Microsystems. A lot of vendors are trying to do the same thing and even have a head start. Why is Oracle targeting this area?

Kurian: Let me clarify what we're doing on the platform side. We're taking a different approach than IBM or HP or Sun, and that is that first and foremost, we view grid computing as requiring software intelligence in order to have a facility to pool hardware resources together, the intelligence that allows you to decide how to pool these and how to use these efficiently is really a software capability. Secondly, that intelligence has to be closely aligned with the system that is requesting those resources. For example, whether it's a database or an application running on an application server, those are the systems that ask for these resources in a grid environment. Fundamentally we don't believe that this intelligence we abstract into our software can be part of just hardware, in the sense of either a chipset or an operating system or a processor. I think that's one big difference.

[In addition,] I think there's a big difference in the sense that our focus on grid computing is on using much lower-cost hardware resources to run your databases and application servers with very high quality of service. Some of the hardware vendors, such as IBM, talk about how grid computing means to rent time on a mainframe so that you can get capacity when you need it. We actually think it's about using much lower cost resources rather than taking the most expensive resource available and then renting time on it.

From a management point of view, I think our goals are different in two ways. When we talk about managing software, we talk about managing Oracle software, and the only components we manage in an environment such as storage are those pieces that are required to manage Oracle software more intelligently. We have no plans to manage a huge variety of third party product the way IBM has with Tivoli, or HP has with Openview. Most importantly, we don't try to manage the CPUs themselves, the hardware. The second thing I think is that to lower the total cost of management what the other vendors do is they give you a centralized monitoring tool to find information in one place from all these different products, but they don't automate the day-to-day management and what we do in contrast is twofold. First, we automate how the systems can manage themselves, we've made them more self-managing so you don't need an expensive management tool. Second, we also automate task management, meaning your day-to-day administration jobs, across many systems. Those are focused on drastically lowering the cost of managing Oracle software.

CRN: A lot of companies have already made investments in, for instance, which application server they use. Do you expect customers to give up investments in existing technologies to implement Oracle's 10g stack?

Kurian: No. The beauty of the grid-computing architecture is you can start small and add in pieces as you go, so it's designed to be modular. You can use just what you need to get started. You can interoperate these with existing environments. It's also open so it's based on standards, and that allows you to use any third-party infrastructure in combination with Oracle technology.