View stunning SlideShares in full-screen with the new iOS app!Introducing SlideShare for AndroidExplore all your favorite topics in the SlideShare appGet the SlideShare app to Save for Later — even offline
View stunning SlideShares in full-screen with the new Android app!View stunning SlideShares in full-screen with the new iOS app!
Software Quick Take Oracle — Neutral (2) ORCL: $29.51 Quick Take: Big Data Appliance: Good News or Bad October 6, 2011 News? Analysts Peter Goldmacher Joe del Callar (415) 646-7206 (415) 646-7228 peter.goldmacher joe.delcallar @cowen.com @cowen.com We maintain our Neutral opinion on ORCL. On Day 3 of OpenWorld we joined about 20 customers in the session on the Oracle Big Data Appliance. Although the product isn’t available yet and the folks giving the presentation weren’t able to say when it would be available, there is a clear spec on the box. It’s an appliance in the sense that it is one box with chips, storage (disk, not flash) a fabric switch and some pre-loaded, pre-configured software, but it clearly lacks the sophistication and componentry of an Exadata box. For the more technical folks, it’s essentially an 18-node Hadoop cluster. The box can include open source or Enterprise versions of Hadoop and NoSQL. Our guess on the price tag would be about $750K per box versus the $8M-ish list price tag for a fully loaded Exadata box. We show component costs below.Big Data Appliance Component Costs Component List Price Quantity Price Comment Sun Fire X4270 M2 Server, Large Configuration, 1.8 Tb Disk $16,592 18 $298,656 Sun InfiniBand QDR Host Channel Adapter $2,295 18x1 $41,310 One Infiniband port per machine 2 TB 7200 rpm 3.5-inch SAS HDD with bracket $949 18x11 $187,902 22 Tb to add to 1.8 Tb already on machine Sun Datacenter InfiniBand Switch 36 $20,995 1 $20,995 Estimated cost of Rack Cabinetry, Power Supplies, Cabling $10,000 About $5K for comparable cabinets from 3rd parties Total Component Cost $558,863 Source: Company Website, Cowen and Company If the price tag is much more than $750K, the price/performance math against commodity boxes starts to break down even with the time/cost savings of pre configured Hadoop, and the embedded optimization to work with customers’ existing Oracle environments. Oracle laid out its Big Data Appliance as a first step in a three step process that assumes your Hadoop and/or NoSQL output then goes into an Exadata box for organization and then into an Exalytics box for analysis. Oracle Loader and Oracle Data Integrator will facilitate moving data from the BDA to Exadata. Please see addendum of this report for important disclosures. www.cowen.com
Oracle Parenthetically, all the Big Data vendors we spoke to are wildly enthusiastic about Oracle’s validation of the Big Data movement and are extremely grateful for the opportunity to tuck in behind Oracle’s big marketing budget. Oracle’s Role in Big Data is Unclear Oracle’s role in Big Data is unclear. Big Data technology arose because it was becoming impossible to cost effectively handle the volume, velocity and variety of data being created by the massive use of the Internet, mobile phones, etc in a traditional relational database. Over the past decade, a number of new database technologies have been created to handle workloads and data types that either A) were handled in an RDBMS at a hefty price tag or B) were ignored because it was impossible to cost effectively work with the data. Fast forward 10 years and we are now at a cross roads where the number of new, non-traditional database oriented projects that aren’t using an RDBMS are exploding because of price/performance improvements. The issue investors have to grapple with is: is Big Data incremental or decremental to traditional RDBMS work loads? On the positive side, more non-traditional workloads will create data sets that have to be analyzed against traditional operational data thereby creating more demand for a traditional RDBMS. For example, if I tweet that I hate American Airlines, American Airlines is using Big Data to capture those tweets and analyze content for trends. However, in order to make real use of that data, they need to know who the tweeter is and whether or not I matter. If I am a garden variety frequent flier (which I am not), American doesn’t care. If I am a top of the heap frequent flier (which I am), they care a lot and want to make me happy. The only way they can figure out if I matter is running their Hadoop output against operational data from a CRM system. Given that most RDBMS are priced per processor, adding that Hadoop output will drive incremental RDBMS revenue. On the negative side, we see two potential issues. 1. Many (if not most) new Hadoop and/or NoSQL workloads never need to be run against operational data in a traditional RDBMS. The most successful Big Data companies like Google and Facebook lead successful traditional RDBMS-free lives. There is no need to tie their vision of a web index or a social graph to operational data. Other common use cases for Hadoop are 1) Web analytics to analyze web traffic, abandonment, conversion etc., 2) Security: who is trying to access your servers and what it the likelihood it represents a threat, and 3) Fraud detection: find fraudulent behavior through analysis of purchasing patterns. 2. There are many workloads that were traditionally handled in RDBMS that can now be handled much more cost effectively in Hadoop or other, newer, lower cost database technologies. For example, back in the old days of Web 1.0, if a customer wanted to create a web presence, IT would have written an online catalog and a shopping cart leveraging an RDBMS. Today, web-based catalogs can be coded faster and cheaper using NoSQL programs, and increasingly, shopping cart technologies can be written in other non-traditional RDBMS technology as well. This is clearly dilutive to the traditional RDBMS market. So the biggest question remains, is the Big Data opportunity incremental or decremental of the traditional RDBMS vendors?2 October 6, 2011
OracleOracle’s Big Data Distribution DilemmaThen there is another big question for Oracle specifically, which is how does it sellBig Data technologies without cannibalizing its own traditional RDBMS business. Wehave no doubt in Oracle’s technical abilities to sell open source and proprietarydistributions for Hadoop or NoSQL. What we worry about is how does Oracle bringdatabase technologies that cost a fraction of its flagship offerings to market? Willan Oracle sales representative really suggest a virtually free open source solution toa client at the expense of a multi-million dollar RDBMS/Exadata deal? We think not.It is our expectation that Oracle is making claims on Big Data early and often inorder to have something to discuss with customers that are interested in Big Data. Itis also our expectation that Oracle will put a very anemic sales effort behind theseofferings.October 6, 2011 3
OracleCowen and Company, LLC. New York (646) 562-1000 Boston (617) 946-3700 San Francisco (415) 646-7200Chicago (312) 577-2240 Cleveland (440) 331-3531 Atlanta (866) 544-7009 Dallas (214) 978-0107 London(affiliate) 44-207-071-7500 Geneva (affiliate) 41-22-707-6900 COWEN AND COMPANY RATING DEFINITIONS (a)Rating DefinitionOutperform (1) Stock expected to outperform the S&P 500Neutral (2) Stock expected to perform in line with the S&P 500Underperform (3) Stock expected to underperform the S&P 500(a) Assumptions: Time horizon is 12 months; S&P 500 is flat over forecast period. COWEN AND COMPANY RATING ALLOCATION (a) Pct of companies under Pct for which Investment Banking servicesRating coverage with this rating have been provided within the past 12 monthsBuy (b) 51.3% 7.7%Hold (c) 46.4% 2.3%Sell (d) 2.3% 0.0%(a) As of 09/30/2011. (b) Corresponds to "Outperform" rated stocks as defined in Cowen and Company, LLCs rating definitions (see above). (c)Corresponds to "Neutral" as defined in Cowen and Company, LLCs ratings definitions (see above). (d) Corresponds to "Underperform" as defined inCowen and Company, LLCs ratings definitions (see above). Note: "Buy," "Hold" and "Sell" are not terms that Cowen and Company, LLC uses in itsratings system and should not be construed as investment options. Rather, these ratings terms are used illustratively to comply with NASD and NYSEregulations. October 6, 2011 5