Oracle Coherence 3.7.1 extends beyond Java, .NET, and C++

Discussions

News: Oracle Coherence 3.7.1 extends beyond Java, .NET, and C++

  1. Announcing Oracle Coherence 3.7.1 - introducing a REST API, exalogic infiniband integration, improved data access performance due to more efficient in-memory and disk-based storage, and query explain plan support - which analyzes the efficiency of queries against data in Oracle Coherence by generating an “explain plan” to describe the calculated effectiveness of each element of an Oracle Coherence query.

    Oracle Coherence is an in-memory data grid solution. Organizations can predictably scale mission-critical applications by using Oracle Coherence to provide fast and reliable access to frequently used data. Oracle Coherence enables customers to push data closer to the application for faster access and greater resource utilization. By automatically and dynamically partitioning data in memory across multiple servers, Oracle Coherence enables continuous data availability and transactional integrity, even in the event of a server failure. Oracle Coherence is a shared infrastructure that combines data locality with local processing power to perform real-time data analysis, in-memory grid computations, and parallel transaction and event processing.


    Download Oracle Coherence 3.7.1
    http://www.oracle.com/technetwork/middleware/coherence/downloads/index.html

    Oracle Coherence Release 3.7.1 Documentation:
    http://download.oracle.com/docs/cd/E24290_01/index.htm

    - Release Notes:
    http://download.oracle.com/docs/cd/E24290_01/coh.371/e22623/toc.htm

    - Read the Press Release
    http://www.oracle.com/us/corporate/press/511975

    Threaded Messages (7)

  2. The reading was a bit too much in-your-face with word "Oracle" used so many times that it was almost a put down.  But that aside, good to see new features like "explain plan".  Is there a plan to expand CohSQL for joins, triggers and some sort of PL/SQL adaptation? Base product already has those features, just needs some compiler enhancements with say antlr (joins will be a killer feature). From .Net integration perspective - entity framework support and plinq would be other features that would really put some more distance with other products.  

    Some of the bug-fixes listed look a bit scary, is complexity taking it's toll (too many missed messages/chances for deadlocks/etc)?

    Thanks for the good work overall on the product.

  3. Coherence[ Go to top ]

    Some of the bug-fixes listed look a bit scary, is complexity taking it's toll (too many missed messages/chances for deadlocks/etc)?

    Fortunately, many of the improvements showing up in Coherence are related to issues that have not shown up in real world use, but instead are the result of internal stress testing.

    Indeed, correcting some of these last remaining issues takes quite a bit of effort, partially because of complexity and partially because of the fact that fixing some of these issues requires actual architectural changes -- deep in the heart of a large product.

    The net result over time though is that the underlying architecture is becoming simpler, and the resulting behavior is more and more predictable, and the systems built with Coherence are becoming more and more reliable.

    Our goal is continuous availability, and to never lose a single transaction or piece of data, ever. All while providing the highest levels of performance and the most insane amount of scalability possible.

    Needless to say, it's both a giant challenge, and an immense amount of fun to work on.

    Peace,

    Cameron Purdy | Oracle

    http://coherence.oracle.com/

  4. Coherence[ Go to top ]

    I know the serverside is no more the place for lively discussions, but thanks for the response.  If the kernel is becoming simpler, then its definitely a good thing.  I have exprienced first hand the bloating of a product and then pairing down to simpler/reliable kernel and yet add many new feature overall to the product as time passes by.  Counter-intuitive but very understandable.

    How about the other quick-wins - the features I mention above.  They can be easily added to the periphery by users like myself; however a branded and properly stress tested feature set coming from the product itself will be definitely attractive.  Perhaps starting out in incubator.  But I do think it's time for some of the "incubator" projects to come out of there and be part of the proper product release it self (e.g. push-replictor).  Any chance of that happening?

  5. More coming :-)[ Go to top ]

    I'll definitely pass on your suggestions to the development group.

    But I do think it's time for some of the "incubator" projects to come out of there and be part of the proper product release it self (e.g. push-replictor).  Any chance of that happening?

    We've been working on this. One of the prerequisites for this was to replace the internal configuration model with a much more fluid & modern (e.g. mock-able, injectable) approach. This should allow us to plug in capabilities such as push replication (that previously have increased the complexity of configuration significantly) without any additional complication. We've also simplified and unified the internal event model, to make things like push replication and the messaging pattern much simpler and easier to understand and maintain.

    Lastly, we're investing significantly in easing the management burden of Coherence, by supporting the same life-cycle, management and deployment features as found in WebLogic (e.g. 2-phase transactional clustered deployment).

    While our policies prevent me from tying any of this work to a release number or date, it should give you a better idea of how we continue to invest in the software to ease its adoption and use.

    Peace,

    Cameron Purdy | Oracle Coherence

    http://coherence.oracle.com/

  6. More coming :-)[ Go to top ]

    Thanks for the info Cameron.

    It would be good to also know how much influence you have on marketing of "your" product, now that is it part of Oracle.  I think in ~4 yrs after acquisition, the sticker price has gone up by an order.  Whereas as most companies don't quite pay the listed price for Oracle products, and I sure wish you commercial success and wish you  continue R&D investment in the product; there is still something to be said about moving from ~2.5K/CPU to 25K/CPU license cost.  Moreover quite a few organizations have had, let's say, "interesting" experience of dealing with Oracle sales reps.  Also, at that sticker price, for some, it is not even considered an option.

    With 10GigE/Infiniband, it is quite attractive to move compute nearer to data given; and on very large scale (1000s of multi-6-CPU Xeon nodes (24 cores+72GB RAM quite common on server infra)).  Have you done any testing/deployment with such a target architecture?  I would like to retain transactionality (so at least EE version on compute+data nodes) and perhaps have Grid version on periphery for .Net clients.  But you see, why should I even have to consider working around the license cost in architecture (or why should I need to hassle with sales folks).  There was a sweet-spot price point, that I sorely miss now adays (not unique to Coherence, I used to recommend blind-folded certain other products, but I get sticker-price shoks for these in just 5 years).

    Has someone in Oracle done a price-elasticity of demand type analysis for esp Coherence vs your investment needs?

     

  7. More coming :-)[ Go to top ]

    Hi K S -

    I think in ~4 yrs after acquisition, the sticker price has gone up by an order.

    The pricing is largely unchanged, except that the Oracle "processor definition" increases the processor count by 2x on multi-core x86 CPUs, so the base prices have effectively doubled. OTOH, the Oracle discounts tend to be larger. For big corporate buyers, it's probably a lot easier (and cheaper?) with Oracle. For smaller companies, I'm guessing that it was easier dealing with Tangosol.

    With 10GigE/Infiniband, it is quite attractive to move compute nearer to data given; and on very large scale (1000s of multi-6-CPU Xeon nodes (24 cores+72GB RAM quite common on server infra)).  Have you done any testing/deployment with such a target architecture?

    Oh yeah! As a matter of fact, I have a few racks with dual 40Gbps Infiniband fabrics, full of 1u servers with dual Xeon CPUs, 96GB of RAM and flash storage. The problem is finding a problem hard enough to use all that mojo ;-)

    Peace,

    Cameron Purdy | Oracle

    http://coherence.oracle.com/

     

  8. Java[ Go to top ]

    Thanks for this information. But i think is that Java most effactive then Oracle.

    Destoner, redler conveyor, screw conveyor