Oracle puts applications in the Fast Data lane
With the X3 generation of the Oracle Exadata and Exalogic engineered-systems platforms significantly upping the amount of Flash (SSD) and memory (DRAM), Oracle is now cranking up the speed on its applications portfolio. With Flash and DRAM prices plummeting, it is now cost-effective to persist, rather than temporarily cache, data. Oracle claims that much of its existing application portfolio can run up to 16x faster, unmodified, on the X3-2 Oracle Exadata and Exalogic models. Oracle is now taking the next step in optimizing portions of its application portfolio to take full advantage of the new X3-2 line. Oracleâ€™s Fast Data approach to enterprise applications follows SAPâ€™s release of SAP Business Suite and CRM on the in-memory HANA platform. Ovum believes that Oracleâ€™s approach is a logical first step, and would like to see this eventually yield a new generation of application-specific, single-appliance engineered systems.
Fast processes dictate a Fast Data approach to applications
Back in the heyday of MRP II, we wrote an article discussing whether enterprise applications should run in realtime. Back then, realtime meant interactive, rather than batch processing. The conventional wisdom 20â€“25 years ago was that running planning processes in realtime would be destabilizing; for instance, just-in-time operation meant delivery on a specific day.
Today, the speed, immediacy, and ubiquity of connectivity has enabled (and forced) business to operate faster. Globalization, the increasing fragility of supply chains, and an increasingly interconnected world are spawning scenarios where isolated events can quickly disrupt markets. For instance, false reports of natural or man-made disasters quickly ricochet around the world. It can drive the trajectory of stock or commodity prices, and in turn, force all players in a global or regional supply chain ecosystem to change their sourcing and logistics decisions. Thought leadersâ€™ opinions on social networks can grow viral, changing consumer behavior, sometimes within hours.
The Ovum report What Is Fast Data? outlined business scenarios and technology enablers that are making Fast Data applications a necessity. Running multiple what-if scenarios in response to disruptive events is becoming the norm for many applications.
The unsung hero of Fast Data: Silicon storage pricing
Flash and DRAM are getting cheap. Oracle doubled the DRAM and quadrupled the Flash in the current X3 generation of the Oracle Exadata and Exalogic engineered systems, without raising prices. Oracleâ€™s rivals have also been taking advantage of cheaper Flash and DRAM.
Traditionally, DRAM cached transactions prior to writing them to disk, cushioning the lag that comes with heavy disk I/O. Today, both DRAM and Flash are cheap enough that application logic and transactions can run without having to touch disk (or relegate disk only for â€ścold,â€ť rarely used data). Over the next five years, Ovum expects Flash to become the primary storage medium for OLTP (online transaction processing) applications.
Introducing Oracle in-memory applications â€“ A Fast Data approach to applications
Oracle has announced a new series of â€śIn-Memoryâ€ť applications that aggressively leverage the age-old approach of tiering data. These applications pair a varying mix of its Exadata, Exalogic, and/or Exalytics engineered systems that are optimized for those offerings.
As a Fast Data approach to optimizing applications, this is a welcome step for Oracle. It responds to the need for applications that support critical realtime or near-realtime processes, and enables embedded analytics that help enterprises couch their decisions in an increasingly volatile world.
Oracle â€śIn-Memoryâ€ť applications put their logic in, and write transactions straight to, Flash storage; disk is only used for cold data. In-memory applications include selected offerings from the PeopleSoft, JD Edwards, Oracle E-Business Suite, Siebel, and Hyperion portfolio. Oracle also claims that existing applications not part of this release will run up to 20x faster on the X3-2 systems and promises more â€śIn-Memoryâ€ť applications in the pipeline.
Oracle wasnâ€™t first here; SAP recently released in-memory versions of Business Suite and CRM that are optimized for the HANA platform. Oracle, like SAP, has followed similar scripts â€“ starting off with analytics, and then releasing applications to run on these high-speed platforms.
A logical first step for Oracle
A key Oracle message for engineered systems is that they simplify IT by eliminating the need for customers to deploy software. With In-Memory applications, Ovum believes that Oracle should take the next step and consolidate them onto a single applications appliance. Oracle has already accomplished that with several analytic-oriented offerings from PeopleSoft, Siebel, and Hyperion, but hasnâ€™t indicated any plans to do so with the rest of the In-Memory portfolio. The fact that Oracle requires two appliances for many In-Memory applications allows SAP to claim that its approach is simpler for Enterprise Business Suite and CRM; Ovum believes that Oracle should put such consolidation on its product roadmap.
Admittedly, while such a move is clearly feasible (a rack can be configured with Oracle Exadata and Exalogic modules inside), the size of Oracleâ€™s business applications portfolio makes this a logistical challenge: of its portfolio, which applications should get a specialized SKU? The answer is those applications that clearly benefit from a Fast Data approach and that are most in demand.
Fast Data Applications would be better branding
Oracle has heavily promoted the â€śFast Dataâ€ť branding for a broad cross-section of its event processing, transactional data grid, and analytics tooling portfolio. So weâ€™re surprised that it did not brand this new set of releases as â€śFast Data Applications.â€ť Ovum believes this is more consistent with Oracleâ€™s branding, and is less ambiguous.
The term â€śIn-Memoryâ€ť may draw attention, but in this case, it is confusing. Today, common usage is that memory refers to DRAM, whereas Flash is often considered storage. Flash has different performance and write characteristics than DRAM. It may perform slower. But more importantly, it is less volatile than DRAM; when the power goes down, the data remains on Flash but disappears from DRAM. That is actually an important benefit that Oracle could trumpet.
These are minor distractions from the main messages: Oracleâ€™s business applications are riding trends toward Flash storage. They are leveraging the more abundant silicon-based storage capabilities in the Oracle Exadata and Oracle Exalogic engineered systems to address a growing range of Fast Data business use cases.
Tony Baer, Principal Analyst, IT â€“ Enterprise Solutions
â€śOracle Engineered Systems provides consistency and reliability in service delivery,â€ť IT017-004079 (March 2012)
SAP Sharpens Analytics Strategy on HANA, IT014-002604 (June 2012)
SAP Well Positioned to Support the Customer-Adaptive Enterprise, IT015-001818 (November 2012)
All Rights Reserved.
No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior permission of the publisher, Ovum (an Informa business).
The facts of this report are believed to be correct at the time of publication but cannot be guaranteed. Please note that the findings, conclusions, and recommendations that Ovum delivers will be based on information gathered in good faith from both primary and secondary sources, whose accuracy we are not always in a position to guarantee. As such Ovum can accept no liability whatever for actions taken based on any information that may subsequently prove to be incorrect.