InformationWeek: The Business Value of Technology

InformationWeek: The Business Value of Technology
e2 Conference & Expo - Boston 2013
= Member Content
Facebook Twitter Share

E-mail | Print | Permalink | LinkedIn | RSS

Oracle Spotlights Its Database Memory Gap


Oracle takes Microsoft to task for its SQL Server Project Hekaton announcement, but in doing so it underscores what's missing from Oracle DB and Exadata.




If you attended this year's Oracle Open World event (and perhaps even if you didn't), you may have formed the impression that the combination of Oracle Database and the latest Exadata database machine, the X3, offers a formidable combo for companies looking to exploit the speed of in-memory database processing. But in a blog post published on a Forbes satellite site on Wednesday, Oracle made it clear that it has yet to deliver in-memory performance with Oracle Database.

The post by Oracle senior communications VP Bob Evans, "Oracle Calls Out Microsoft over Vaporware Fantasy," challenges Microsoft's Nov. 7 announcement of Project Hekaton. Evans, who was formerly editorial director here at InformationWeek, singled out my coverage in the article "Microsoft In-Memory Move Challenges SAP, Oracle", calling my headline "dubious" given that Hekaton won't arrive, as analysts guess, until 2014 or 2015. (Before joining Oracle, Evans held a similar corporate communications role at SAP.)

Evans' blog goes after Microsoft's claims about Hekaton and comparisons that Doug Leland, Microsoft's general manager of SQL Server marketing, made to the Oracle in-memory strategy. In doing so, however, Evans underscores the fact that Oracle, too, has yet to deliver an in-memory strategy tied to its flagship database.

Quoting Oracle database executive Andy Mendelsohn, the piece points out that Oracle's in-memory database is TimesTen, which runs inside the Exalytics appliance (along with an in-memory version of the Essbase database). That box is a sort of analytics turbocharger, and it's plugged into yet another box, Exadata, which runs on Oracle Database. But here was Evans, protesting that Microsoft was disingenuous in suggesting that Oracle is "forcing customers to buy, learn, and manage a separate solution."

[ Here's the article that sparked the discussion. Read Microsoft In-Memory Move Challenges SAP, Oracle. ]

If Oracle Database had in-memory capabilities, you wouldn't need a separate Exalytics box. But Mendelsohn acknowledges that Oracle Database is limited to "some in-memory database features." He doesn't say how or when, but Mendelsohn promises that Oracle "plans to continue to enhance Oracle Database in-memory capabilities."

So on Microsoft's side it's true we have only a beta product and fuzzy delivery date, but on Oracle's side we have a "yeah, we're going to do that, too" promise by one executive. Will the in-memory capabilities planned for Oracle Database handle transactional apps (as Hekaton promises) while Exalytics continues to handle the analytic needs? Mendelsohn didn't say. If the Oracle DB ends up handling both tasks (as SAP Hana can), will the time Exalytics buyers have invested in learning TimesTen and the in-memory version of Essbase have been unnecessary? That's unknown at this point.

The thing is, all these particulars about Oracle's in-memory plans only undermine the gauzy impression Larry Ellison tried to create at Oracle Open World around Exadata X3 as a memory powerhouse. Having finally admitted that the world is turning to memory and away from disks -- something SAP has been saying for three years -- Ellison had all the zeal of a reformed smoker, talking up the "26 terabytes of memory" available in X3. But don't mistake memory for "in-memory." In fact, the vast majority of that capacity (22 TB) is flash, not RAM, and there is a big difference.

"Flash is not memory, and I disagree with Larry Ellison on that totally," Gartner analyst Don Feinberg told InformationWeek last week. "Exadata doesn't give you an in-memory database, it gives you a disk block cache."

The practical difference is that Exadata cache is not as fast as Hana or what Hekaton promises. Until Oracle delivers full in-memory capability as part of Oracle Database, you give up the potential for real-time performance. Nevertheless, as I quoted Feinberg saying last week, "cache is still fast, and for most clients of Exadata, it's going to be fast enough."

So where does that leave Oracle customers? If you want in-memory analytical performance, you'll have to buy the separate Exalytics appliance. If you want in-memory transactional performance, you're out of luck, but Exadata might do.

As for the competition, SAP just announced SAP CRM running on Hana, and Microsoft just announced Hekaton, its plan for in-memory transactional processing. Microsoft previously delivered in-memory data exploration with Microsoft PowerPivot, introduced in 2010, and in-memory data visualization with Power View, introduced in April.

Hekaton will run transactional applications, so you will be able to take a Microsoft SQL Server application and run it in memory without changes. "That's not something you can do with an Oracle application today," said Feinberg.

Oracle has confirmed, by way of Evans' blog, that the company plans to do more with in-memory capabilities within Oracle Database. But we don't know what it will do or when it will do it. With Microsoft we don't have a release date, but at least we have a roadmap.

Federal agencies must eliminate 800 data centers over the next five years. Find how they plan to do it in the new all-digital issue of InformationWeek Government. Download it now (registration required).

Upgrading isn't the easy decision that Win 7 was. We take a close look at Server 2012, changes to mobility and security, and more in the new Here Comes Windows 8 issue of InformationWeek. Also in this issue: Why you should have the difficult conversations about the value of OS and PC upgrades before discussing Windows 8. (Free registration required.)




InformationWeek encourages readers to engage in spirited, healthy debate, including taking us to task. However, InformationWeek moderates all comments posted to our site, and reserves the right to modify or remove any content that it determines to be derogatory, offensive, inflammatory, vulgar, irrelevant/off-topic, racist or obvious marketing/SPAM. InformationWeek further reserves the right to disable the profile of any commenter participating in said activities.

Disqus Tips To upload an avatar photo, first complete your Disqus profile. | View the list of supported HTML tags you can use to style comments. | Please read our commenting policy.
Subscribe to RSS


Advertisement


InformationWeek Reports

report Cloud Implementer's Checklist
Once your agency has completed the business case for a private cloud, how do you actually move ahead with your data center transformation? Our report provides a practical set of steps to get you there, including a "to do" list that will be helpful to anyone on your IT team who's involved in the project. By the time you're done, your data center should be home to a more flexible, on-demand IT services.

report Cloud Compliance in Government
Compute clouds created for government data centers must adhere to a range of specifications designed to support data and system security, privacy, and governance. FISMA, HIPAA, SOX, and SAS 70 are just some of the requirements that have to be taken into account as federal IT pros deploy a shared-services cloud model. In this report, we identify the key specs that need to be factored into any federal cloud architecture.

report Government Cloud Platform Strategy
This report analyzes the key IT infrastructure considerations that must be taken into account for implementing cloud services in federal data centers: software/hardware environment, multi-tenancy, security, virtualization, and management tools. We also explain the key important role that APIs play in supporting hybrid scenarios that tap into public cloud services.

report The Business Case for Government Clouds
This report assesses usage scenarios, barriers, and other variables that factor into the decision of whether and how to implement cloud computing in federal environments.