Industry Leader Q&A with Oracle's Tim Shetler - The Oracle Exadata Database Machine Explained

<< back Page 2 of 4 next >>

There are two models as you mentioned and the high-end Exadata X2-8 can be expanded to an eight-rack grid with over 2,300 CPU cores and 2.6 petabytes of raw storage. Has anyone ever maxed out the capacity?
Shetler: You were talking about eight full racks cabled together with that core count and that storage capacity. Nobody is running with eight full racks cabled together today. We certainly have many customers that have bought more than eight full racks but they tend not to cable that many together in one spot.  I think the largest Database Machine that is in production today maybe has three full racks, maybe four full racks now. 

We have customers with over a dozen full racks that have been purchased but they will put one in this location, two in this location, maybe cable a couple together if they have a really big data warehouse for example, but given our really good compression on the data and how fast these compute cores are, it is pretty unlikely that any particular application will need to have anything approaching eight full racks cabled together all operating on the same database cluster. That would be huge, probably bigger than any application that exists on the planet today.

It comes pre-installed with Linux and Solaris and then you actually select the operating system that you want when it is deployed. Is there one or the other that customers are tending toward?
Shetler: Well, Solaris is just now becoming available so we don't really have any history on that other than that we know of a handful of customers that are eagerly waiting to work with Solaris and these of course are all folks that are coming from Solaris. Everybody today that is installed is running Oracle's Linux because that was the only choice until this month.

The Exadata has been described as a 'cloud in a box.' Do you present this to customers as a way to go, taking the Exadata approach over a public cloud direction?
Shetler: Oracle's database and several of our other products all are available today in a public cloud model. Amazon now in fact includes the Oracle Database in their service in the cloud. We kind of leave it up to the customer as to what is appropriate for them. Exadata doesn't run today in any public cloud model. It only runs in a private cloud, or private model today.

Customers generally today are obviously choosing to control the assets more tightly and if performance is a concern or a consideration, that tends to move everything into the private cloud model more than the public cloud and, of course, security may be another issue as well. I can tell you that even though there are hundreds of Oracle Databases running in the Amazon cloud today, I am not aware of any of them that are running in what I would consider to be a mission-critical context.

How are customers using cloud technology in combination with Exadata?
Shetler: A lot of them are running as sort of standby servers. We have an interesting Exadata customer today that has deployed E-Business Suite in their data center and they use an Amazon cloud system as their disaster recovery system, so they will ship updates to the cloud and apply them there and then if they were to lose their primary system they would convert over to the cloud.

They just consider that to be a more economical model for them than to buy a second Exadata Database Machine in their data center. We expect that will happen more and more — or that people will ship data into the cloud that maybe isn't as sensitive and they will use the cloud to store that data for archiving or what have you.

And then I think we fully expect over time that the cloud will be in a hybrid model of some form where most of the critical applications or maybe all the critical applications, are running in a data center that you control, a private cloud if you will, and there will be some connection into a public cloud aspect either for shipping data or for peak capacity periods maybe or for other kinds of applications that aren't quite as sensitive to performance or security that run more economically in a public cloud model.

When we talk about private cloud with Exadata, we most often refer to that as our consolidation play. We position Exadata as great for OLTP, great for data warehousing and great for database consolidation. We think that when customers talk about private cloud you could replace that phrase with "database consolidation" and they would not disagree with that. They are looking at Exadata because it is such as beefy system as a place where they can move multiple Oracle Databases and then decommission a bunch of older platforms and, in some cases, actually decommission some data centers, and gain economies in so doing.

<< back Page 2 of 4 next >>