Archive for September 2009
Oracle list the following costs for a full-rack Exadata system:
- $1,150,000 for the hardware
- $1,680,000 for the storage server software
However, there was an asterisk next to the storage server software price. In small print, it says that you must “Use your existing database licenses“. That of course assumes that you have available licenses of Oracle Database and all required add-on features. Of course, I probably don’t need to mention Oracle’s reputation for auditing customer environments and later discovering very large compliance issues. Business Week covered this very topic last week in a very interesting article titled Oracle Has Customers Over a Barrel. Here is a short excerpt from that article:
One sore point with customers is the company’s audits. Most software companies audit their customers occasionally, checking through their offices and tech systems to make sure they’re paying for all the software they use. But Oracle has a reputation for being unusually aggressive, says Jane Disbrow, an analyst at Gartner Research (IT). She says Oracle’s policies can be confusing, and contracts with customers often don’t clearly spell out their rights. As a result, some get presented with bills ranging from $200,000 to $4 million after they get audited. “It’s easy to be out of compliance with Oracle licensing. They do nothing to help people stay in compliance,” says Disbrow. “Then they audit you and hand you a big bill.”
I thought it would be interesting to to to figure out what a full-rack Exadata system would cost when you include all the required and recommended software. And, of course, when you include the first year of maintenance. Note that I am making some educated guesses here. For instance, it appears that Oracle require you to buy Advanced Compression because the compression now occurs in the storage. In time, I’m sure that details like this will be flushed out. In the mean time, here’s my current best guess at the math:
$1,150,000 for the hardware
+ $1,680,000 for the software on the storage servers
+ $369,600 for Support & Maintenance on the Exadata storage software (calculated at 22%)
+ $1,520,000 for Oracle Enterprise Edition ($47.5k * 8 servers * 8 cores * 0.5 Intel core factor)
+ $736,000 for Oracle RAC ($23k * 8 servers * 8 cores * 0.5 Intel core factor)
+ $368,000 for Oracle Partitioning Option ($11.5k * 8 servers * 8 cores * 0.5 Intel core factor)
+ $368,000 for Oracle Advanced Compression ($11.5k * 8 servers * 8 cores * 0.5 Intel core factor)
+ $160,000 Enterprise Manager Diagnostic Pack (recommended)
+ $160,000 Enterprise Manager Tuning Pack (recommended)
+ $728,640 for Support & Maintenance on the above non-storage database software (calculated at 22%)
That all comes out to a whopping $7,240,240 for license and first year support/maintenance, which is a little more startling that the $2.7M indicated in the Oracle materials. Of course, this is the list price, and does not reflect any discounts you may be able to negotiate. A couple of additional things to consider:
- The list price for the annual support costs of the above software components is $1,098,240.
- Don’t forget that these costs do not include installation, which is a custom quote.
It seems like its raining Oracle news just now. Oracle had their Q1 earnings call after the close of trading today. Many of the headlines covering this story mention their 5% drop in revenues. But if you examine the numbers a little deeper, it gets interesting for people in the database business (and Oracle Database is what I’m interested in, so I’ll focus on that).
For new software licenses, their “Applications” revenue declined by 4%, but their “Database and Middleware” revenue declined by 22%. A 22% drop is quite a large number (especially in light of some of IBM’s recent earnings calls where they revealed a 30% growth in IBM’s distributed database revenues). But let’s have a closer look at this 22% decline in revenues…
On the earnings call, Oracle mentioned taking advantage of cross-selling opportunities with recent acquisitions like BEA to fuel growth. This makes sense. When you acquire a company, you get the opportunity to sell the acquired products to your existing customer base, and vice-versa. Many large companies use this strategy to realize very nice returns on their investments. Well, if you follow this line of thinking, and revenue from these acquired product lines are growing, then what is happening to the database revenues they are lumped with? If “Middleware” revenues are growing (and this is speculation on my part), and if “Database and Middleware” revenue is down 22%, then what is the true rate of decline in Oracle Database revenues?
Next consider the up-sell of add-on features like Advanced Compression and Real Application Clusters. We know that Oracle derives a lot of revenues from up-selling existing customers on add-on features like these. I wonder how much of their new license revenues were for the core database, and how much were for add-on features. Of course, you could argue that these add-on features were also part of last year’s revenues, and so this is a moot point. However, you must keep in mind that—for a company like Oracle–revenues from these add-on features typically grow at a strong rate. Oracle has been crowing lately about the strong adoption rates for Real Application Clusters. If they are indeed experiencing such strong adoption for add-on features like these, then it is not unreasonable to expect that revenues from add-on features are growing faster than revenues from new core database sales. Or, I should say, that they are not declining as fast as revenues for the core database product.
When you take these factors into consideration, you have to wonder what is the real rate of decline for Oracle Database sales? You start with a 22% decline. Then you factor in that middleware revenues are probably declining at a lower rate than Oracle Database revenues. And then you also factor in that add-on feature revenues are also probably declining at a lower rate that the core database product revenues. What are you then left with? I think its fair to speculate that rate of decline for the Oracle Database product is considerably worse than the 22% decline reported for “Database and Middleware”. I suspect that Oracle executives will have a tough time arguing that they are not losing market share under these circumstances.
If you are interested in IBM’s take on yesterday’s Exadata v2 announcement, check out the article published by eWeek titled Oracle, Sun Show Off Super-Fast, Flash-Based OLTP Server. In this article, Bernie Spang offers IBM’s views on the announcement. Highlights include:
- Pointing out that Oracle neglected to include the cost of the software when indicating system price
- Pointing out that this is an x86-based system and not a SPARC system, and discussing the implications
- And highlighting IBM’s momentum in winning over both Sun customers and Oracle Database customers.
I’ll provide some additional personal analysis on this announcement soon…
IBM just issued a press release about its momentum in helping organizations move off Oracle Database and on to DB2. The press release concentrates on two trends:
- Organizations switching their SAP applications to use DB2.
- Organizations taking advantage of new features like PL/SQL in DB2.
The press release mentions that more than 100 organizations have made the move in the past 6 months and includes quotes from Coca Cola Bottling Company, Rossi Residencial, and Neusoft. It includes some great benefits that these organizations are realizing like being able to “reduce their IT costs by 35 percent and improve efficiency by 65 percent”.
You can read the press release at IBM DB2 Gains New Enthusiasts in Search of Better Performance at Lower Cost
According to his profile on it.toolbox.com, Lewis Cunningham is an “Oracle ACE, Database Architect and self-professed database geek”. Many people know Lewis as an influential influential voice in the database world. A couple of days ago, Lewis blogged about PL/SQL Anonymous Blocks (in DB2 9.7). For those of you who are not aware, DB2 9.7—which was released a few months ago—supports the most commonly used PL/SQL syntax. In fact, PL/SQL support in DB2 is just the tip of the iceberg. The following table describes some of the new features in DB2 9.7.
|Oracle Database Feature||DB2 9.7 Support|
|Concurrency control||Native support|
|Data types||Native support|
|SQL dialect||Native support|
|PL/SQL packages||Native support|
|Built-in packages||Native support|
|JDBC client with extensions||Native support|
|SQL*Plus scripts||Native support|
You can get more details about these new features at DB2 9.7: Run Oracle applications on DB2 9.7 for Linux, UNIX, and Windows on the IBM developerWorks Web site.
Several hundred organizations were part of the Beta testing for these new features. We got some great feedback from them. Here is a small selection of that feedback:
DB2′s PL/SQL compatibility is excellent. We’re looking forward to integrating the current dual source code base into a single one. This will increase our development and testing productivity. In addition, SQL compatibility is significantly improved. We ran an Oracle-version program ‘as is’ on DB2, and the test result was better than we expected. The compatibility level that DB2 9.7 achieved is amazing.
—Masahiko Kudo, Works Applications (Japan)
The IBM DB2 9.7 compatibility is amazing — and there are no queries or DB2-specific code in our applications! Everything is compatible with Oracle and DB2 9.7.
—Gene Ostrovsky, VP Research and Development, ExactCost
As a migration specialist, the new DB2 9.7 PL/SQL compatibility features help Oracle users migrate to DB2 seamlessly. These features drastically reduce the time required for migration efforts and significantly lower overall costs. The custom Perl scripts and defined manual workarounds, which I previously relied on for database porting services, are now obsolete. DB2 9.7 allows native support of PL/SQL, weak typing, enhanced locking methods and a vast array of other features designed to make database migration virtually effortless.
—Axel Puerner, Puerner Unternehmensberatung
Ryan Tate claims that “Larry Ellison will always be a shameless truth-bender” in an article titled Larry Ellison Can’t Be Bothered With the Facts. The article talks about an advertisement that appeared recently in the Wall Street Journal and goes on to share some interesting anecdotes about Oracle advertising.
The advertisement in question suggests that Oracle plan to announce a new TPC-C result using Sun hardware on October 14. One thing this article does not mention is that, as far as I can tell, the advertisement does not follow the TPC rules. If you read Section 8.2.2 Unfair Use, you will see that Oracle could be interpreted to have violated several of these rules and unfairly used the TPC benchmark. You see, the TPC does not allow organizations to “make TPC-related claims or lead the reader to TPC-related conclusions which are untrue or cannot be substantiated by the entire body of results.” TPC does this to prevent misleading advertising. You may wonder how this could be misleading. Well, it could be misleading because it does not tell the full story. For instance, what if their upcoming benchmark result requires an inordinately expensive hardware configuration. Without certain primary metrics, TPC results cannot reasonably be compared.
Aside from Oracle’s apparent flaunting of the TPC rules, this advertisement is also notable because Oracle indicates that the upcoming TPC result will be on Sun hardware. The reason that this is interesting is that Sun has publicly derided the TPC-C benchmark in the past. In fact, they have not participated in the TPC-C benchmarks for several years, claiming that the benchmark does not reflect real world transactional workloads, and that the benchmark is easily gamed by vendors. It is an interesting about-face for Sun. I wonder what prompted this change in what was a very firm stance on their part.