Battle of the Machines: Oracle Cranks Up the Heat
For CFOs, it might sound a bit like deploying an elephant gun to shoot a mosquito.
Oracle’s new US$175,000 Exalytics X3-4 claims a scan speed of up to 100 million rows of data per second, allowing super speeds in data crunching and modeling, dashboard updates and delivery of other reports. That’s similar to claims by in-memory rival HANA, developed by SAP, which reports 100,000 times improved query performance.
Does finance really need such blinding fast computing power, especially the smaller enterprises? Yes it does, says Shridar Jayakumar, Program Director, Business Analytics, Asia Pacific, at Oracle. “It’s not about the size of the company; it’s about how competitive they want to be,” he says.
Jayakumar spoke with CFO Innovation’s Cesar Bacani about business analytics in Asia, the new Exalytics machine and competition with SAP’s HANA. Excerpts:
Tell us about Oracle’s new Exalytics machine, which was released in July.
Exalytics X3-4 is the industry’s first business intelligence machine. What this means is that the BI software is actually in-memory and engineered to work off the same hardware. That concept of hardware and software engineered to work together by one vendor is unique in the marketplace for analytics.
In the first release [in 2011], we’re talking about best-of-breed BI software. In that case it was Oracle BI Foundation, which includes mobility, scorecards, all traditional reporting and ad hoc analysis tools that you would expect.
There was an update to Exalytics in the second half of 2012, which brought in two other key analytic components. One was Hyperion – this is for the office of finance – where [database management system] Hyperion Essbase and Hyperion Planning were certified on Exalytics. The second was [data discovery platform] Endeca Information Discovery, which Oracle acquired eight months before.
Now with X3-4, we’ve doubled the amount of RAM [random access memory]. It’s gone from 1 terabyte to 2 terabytes. We’ve brought in 2.4 terabytes of flash memory as well, which helps speed up a lot of the analytical applications, such as those running on Essbase. We still have 5.4 terabytes of raw hard disk space.
On the software side, we have now certified for all the latest releases, including the latest release of Endeca 3.0, the Hyperion EPM suite, as well one additional module, which is Hyperion Profitability and Cost Management.
X3-4 is a piece of hardware. I imagine it’s about the size of a mini-fridge?
I would say it’s more like a coffee table book. It’s basically a small rack.
In theory, it can replace the servers I currently have in my data center?
Absolutely. Data center consolidation or service consolidation is one of the key drivers for the whole Exa- series, not just Exalytics but also Exadata and Exalogic.
Can I run my ERP system on Exalytics?
Exalytics is specifically for analytical applications.
When you’re talking about ERP, there are two ways you can run that on [in-memory] analytics. One is in Exadata, which is primarily a data warehouse machine or database appliance.
The other way you can run your ERP is on SPARC SuperCluster, where the SPARC SuperCluster has both Exadata and Exalogic components. Exalogic is where you have your web service in an ERP environment.
If my ERP is by SAP, can my Exalytics machine still extract the data it needs from it?
Absolutely. We have successfully deployed Exalytics on SAP ERP and SAP DW [data warehouse] customers. Exadata, Exalogic and SPARC infrastructures are all certified to run SAP applications.
What if my ERP is on the cloud? Can Exalytics work with that data as well?
That’s a good question and something we’re coming up against a lot because customers are looking for flexibility. They want the best technology and performance, but they also want to keep their options open. They want to have the flexibility to put some or all of their applications and databases on the cloud.
Engineered systems can form part of your cloud strategy. We have global customers where we host their entire ERP, BI and performance management. So they buy Exalytics that’s hosted by us, but dedicated to them.
Another way that we do this is we have cloud adaptors for bringing data from the cloud, for example, if they have Oracle Fusion Financial or Oracle Fusion CRM or any of the other Fusion cloud offerings, to bring that data into the on-premise environment.
And those analytics cloud adaptors can also be adapted for non-Oracle cloud offerings?
The first priority we’re looking at is on the large volume of Oracle cloud customers. But we’re absolutely going to cover the rest of the spectrum in due course.
Given its processing power, storage and in-memory speed, is the X3-4 appropriate only for very large enterprises? It’s probably not for a company with sales of just US$500 million?
Compared to our revenue share of business analytics to the other regions in the world, we sell proportionally more Exalytics in Asia Pacific, despite [company revenues] here being smaller. I think that goes to show that this is a way companies are differentiating, even if they are smaller companies.
It’s not about the size of the company; it’s about how competitive they want to be. How they want to scale out – do they want to give people mobile access, for example?
A lot of countries, especially the more developing countries in South Asia, have leapfrogged a lot of the fixed-line elements and moved straight into wireless. Mobile devices, iPads, consumerization of IT have proliferated.
What this is means is that they bring to the workplace the consumer mentality of ‘I’m not going to wait more than two or three seconds to access data.’ If you don’t have the analytical server that can handle that demand, then people are going to move off BI.
We’re actually seeing a lot more demanding customers when it comes to BI. The West has data warehouses and BI for coming to a couple of decades already and people there have become accustomed to waiting for reports for 10 to 12 minutes.
Except for the more mature larger customers that had invested in BI many years earlier, the paradigm is different in Asia. It’s the consumer mind-set [of getting rapid responses], and so we need things like Exalytics to meet that performance requirement.
I see Oracle’s price list as of September 1 has Exalytics X3-4 at US$175,000, with premier support for systems at US$21,000 annually and sundry other costs. Is that more expensive than an ordinary server?
I would actually say it’s much more economical than the equivalent commodity hardware. Exalytics X3-4 is a 20-CPU, 40-core, 2-terabyte-RAM machine. How much would a single 40-core machine cost from a commodity hardware vendor?
If you look at it just from the sheer hardware perspective, although that is not the way we justify ROI to our customers, Exalytics is way cheaper than the alternatives. And the fact is, you don’t need to re-architect your applications, because whether it is Hyperion or Endeca or Oracle BI Foundation, you just bring them to Exalytics, load and then use and expand.
SAP also has an in-memory machine called HANA. Is Exalytics different, or are these two devices essentially the same except for the brand name?
My understanding of what HANA stands for is High Performance Analytical Appliance. My understanding of an appliance is something like a toaster – it’s a hardware, it’s good for a certain process, you just plug it in and it works.
Exalytics is the only engineered system and also the only in-memory appliance software and hardware that cover analytics. By analytics, I don’t mean ERP processing applications. I don’t mean data warehouse. What I mean is business intelligence, data discovery, planning, budgeting, consolidation.
No other vendor has those software on an in-memory machine.