Friday Nov 13, 2015

Questions You Asked: When using interval partitioning, will new partitions be placed In-Memory?

This is the second in our series of "Questions You Asked" and this time the question has to do with interval partitioning and whether newly created partitions will be populated into the IM column store.

We'll begin our experiment by creating an interval partitioned table based on the SUPPLIER table from our SSB schema. I've added a key_no column to make the interval partitioning easy. We then insert data into three partitions and list the results. Note that the first partition is named p1 because we had to create at least one partition with our CREATE TABLE statement. The other two are system generated names and those partitions were created automatically as part of the interval partitioning feature when we ran the second and third insert statements as you can see below.

[Read More]

Friday Nov 06, 2015

Oracle OpenWorld 2015 Wrap up!

I can't believe another Oracle OpenWorld has been and gone.

Thanks to all those that attended the In-Memory sessions and stopped by the In-Memory demopod during this years Oracle OpenWorld. It was an amazing conference this year with lots of great technical sessions and interesting discussion at the demo grounds.

I've been inundated with requests to share the presentation we gave on In-Memory at the conference, so I thought it would be a good idea to share the links to the presentations here.

A number of Oracle Database In-Memory customers present on their experiences of using In-Memory. You can find details from their sessions below;

We also gave folks the opportunity to try out Oracle Database In-Memory for themselves in our hands-on lab, Oracle Database In-Memory Option Boot Camp: Everything You Need to Know, so I thought I would share the lab handbook in case anyone want to try some of the labs at home.

Finally there was a lot of talk about how Oracle Database In-Memory takes advantage of Software in Silicon ,so I thought it would be good to share this link to a short video of Juan Loaiza describing exactly what we mean by Software in Silicon.

We hope you enjoyed the conference as much as we did and we look forward to see you all again next year at OOW 2016 September 18–22, 2016 in San Francisco.

Monday Oct 26, 2015

Day 2 of Oracle OpenWorld 2015 October 26th

Oracle OpenWorld started yesterday and San Francisco is just buzzing with Oracle folks.

If you are attending the conference don't miss the opportunity to chat with the Oracle Database In-Memory team at one of our sessions or at the Oracle demogrounds.

Today kicks off with one of the best general sessions of the conference, General Session: Next-Generation Data Management Architecture Session GEN6765 at 11am in Marriott Marquis—Salon 7/8/9 (Yerba Buena Ballroom) in which Andy Mendelsohn, EVP of Database development, will share how Oracle Database 12c and its next-generation architecture are helping customers on their journey to the cloud.

Then in the afternoon the first 2 of the 8 customers speaking about their experiences with Oracle Database In-Memory will have their sessions. First up is Facebook’s Use of Oracle Database In-Memory from Inception to Rollout Session CON1685 at 2:45pm in Moscone South - room 301. Sudhi Vijayakumar will describe how Facebook evaluated Oracle Database In-Memory and then rolled it out in production to reap the benefits of faster response times for analytical queries.

Then at 4:00pm Five Ways Swiss Mobiliar Improved Business Value with Database In-Memory Session CON2715 in Moscone South - room 310. In this session folks from Swiss Mobiliar explain how they extended their traditional business intelligence (BI) systems to a more real-time solution using Oracle Database In-Memory. They will also compare this new solution to their original IBM DB2 solution.

Today is also your first chance to catch our hands-on-lab, Oracle Database In-Memory Bootcamp (Session HOL10483) in the Golden Gate room (25th Floor) at the Hotel Nikko. In this hands-on lab we'll provide a step-by-step guide on how to get started with Oracle Database In-Memory, how to identify which of the optimizations are being used, and how your SQL statements benefit from them. You will also get to experience firsthand just how easy it is to start taking advantage of this technology and the incredible performance improvements it has to offer.

Remember it's about a 15 minute walk from Moscone to the Hotel Nikko, so make sure you give yourself plenty of time to get there.

Members of the In-Memory team will also be available at booth SLD 28 from 10:00am to 6pm at the Oracle Database demogrounds in Moscone South, to answer any questions you might have. Remember the In-Memory demo booth is the best place to snag a limited edition In-Memory bumper sticker!

Sunday Oct 25, 2015

Day 1 of Oracle OpenWorld 2015 October 25th

The biggest week of the Oracle calendar has arrived. Oracle Open World 2015 officially starts today and it's going to be a fantastic conference for anyone interested in Oracle Database In-Memory. We have tons of great sessions planned for you, as well as hands on labs and of course plenty of opportunities to talk directly with the In-Memory development team.

The show kicks off with tons of great technical sessions today, selected by the Oracle User Groups, including the following sessions, which feature Oracle Database In-Memory:

  • Eighteen Months of Production Experience with Oracle Database 12c at CERN Session UGF 6827 at 9:00 am in Moscone South - room 306. Eric Grancher and Prasanth Kothuri from CERN share the reasons why they decided to become early adopters of Oracle Database 12c  and their experiences both during the migration and also now that they have been in production for over a year.

  • Oracle Database In-Memory Option: Challenges and Possibilities Session UGF10319 at 11:00 am in Moscone South - room 262. Christian Antognini explains the key concepts of Oracle Database In-Memory option, reviews what to expect from this new technology, and looks at situations in which the user can take advantage of it.

  • More Than Another 12 on Oracle Database 12c Session UGF 3190 at 1:30pm in Moscone South - room 306. In this session twelve of the best speakers in EMEA—all ACE directors— and little old me will share our knowledge on Oracle Database 12c. Quick-fire, seven-minute topics provide nuggets of wisdom to consider and explore during and after Oracle OpenWorld. A must attend session if you love different accents!

And of course, Larry's keynote is this evening from 5:00pm – 7:00pm, Moscone North. A must see, as he is bound to make some exciting announcements to get the show started!

Tuesday Oct 13, 2015

Star Schema Challenge - Part 5

In my previous post I established a baseline for 100 concurrent users running a variety of business intelligence queries against a 500 million row star schema on a commodity Intel-based server with 60 cores and 1 TB of DRAM.  All tables were loaded in the In-Memory column store (IM column store). 

At 100 users, the star schema ran at a median of 21.5 seconds per query and the de-normalized fact table ran at a median of 10.0 seconds per query. Given that this test runs 100 users on a 60-core server, it should be obvious that it is asking a lot of the server.  With 25 users the star ran at a median of 3.85 seconds and the de-normalized table ran at median of 1.57 seconds per query, an excellent result for this more reasonable workload. 

[Read More]

Friday Oct 02, 2015

Questions you asked: What happens if a column is not In-Memory?

When we're talking to customers or giving presentations questions often get asked that seem simple, but could have an answer that is worth showing with an example rather than just saying, "yes, it works this way". So I'm going to start a series of posts that will address those questions. I will continue the series as I come across what I think are interesting questions.

This first post in the series of "Questions You Asked" will be "What happens if a column is not In-Memory?"

[Read More]

Friday Aug 28, 2015

Popular Statistics with Database In-Memory

Throughout our previous posts we have mentioned various session level statistics that are available to help identify what is actually going on with Database In-Memory. Since these statistic definitions didn't make it into the Reference manual I was asked recently if there is a list anywhere that defines what these statistics are. Since there isn't I thought I'd post a summary of the popular statistics that we've mentioned in the blog and that you might want to know about.

[Read More]

Wednesday Aug 05, 2015

New White Paper on using Database In-Memory with the E-Business Suite Released!

We would like to draw your attention to a new white paper has just been released titled "Using Oracle Database In-Memory with Oracle E-Business Suite".

The white paper details how Database In-Memory works, what the requirements are to use it with the E-Business Suite and works through some use cases and examples.  More details can be found in the My Oracle Support (MOS) Note 2025309.1.

The EBS team worked on hard on this paper and we think it's a great source of information for E-Business customers who want to take advantage of Oracle Database In-Memory. Enjoy the read!

Tuesday Aug 04, 2015

Oracle Database In-Memory on RAC - Part 4

Setting Up Independent In-Memory Column Stores

In previous posts we've talked about how to use RAC services to enable the IM column store to be run on a subset of nodes in a RAC environment. We also mentioned that it is possible, using RAC services and the DUPLICATE sub-clause on engineered systems, to enable rolling patches and upgrades when running the IM column store on a subset of nodes.

In this article we're going to talk about how to set up independent IM column stores on a RAC cluster using services and some database initialization parameters.

But first let's ask and answer the question, why would you want to do this? This might be a good idea if you are trying to enforce application affinity at the node level and don't want to allow inter-instance parallelism.

[Read More]

Friday Jul 24, 2015

What to do with optimizer statistics when upgrading to 12c to take advantage of Database In-Memory

Before most customers can take advantage of Database In-Memory they will need to navigate the tricky terrain of a database upgrade. One of the most challenging aspects of an upgrade is figuring out how to minimize performance regressions due to execution plan changes.

And if that wasn’t enough to handle, the introduction of Database In-Memory into a 12c environment has the potential to change even more of the execution plans. So, what should you do?

[Read More]

Tuesday Jul 21, 2015

Oracle Database In-Memory Bundle Patch 10 Released

The latest Bundle Patch for Database In-Memory has been released. The Bundle Patch is 21188742 or Bundle Patch 10 for Engineered Systems and DB In-Memory (July2015)). This Bundle Patch improves the performance of mixed workload environments (OLTP & DW workloads), as well as enhancing the performance of analytic queries with aggregation. More information on the latest Bundle Patch can be found in the MOS note 21188742.8 or in the Mos note: Bundle Patches for Engineered Systems and DB In-Memory (Doc ID 1937782.1).

[Read More]

Friday Jul 03, 2015

Oracle Database In-Memory Bundle Patch 9 Released

The latest Bundle Patch for Database In-Memory has been released. The Bundle Patch is 21053000 or Bundle Patch 9 for Engineered Systems and DB In-Memory (June2015)). More information on the latest Bundle Patch can be found in the MOS note: Bundle Patches for Engineered Systems and DB In-Memory (Doc ID 1937782.1), and for for specific details on Bundle Patch 9 see MOS note 21053000.8.

[Read More]

Friday Jun 26, 2015

Do I really have to drop all of my reporting indexes?

I'm back on the road this month, meeting with customers to discuss their initial impressions and experiences with Oracle Database In-Memory. During one such discussion, I got asked a very peculiar question. The question was, "Do I really have to drop all of my reporting indexes if I use Database In-Memory?"

I have to admit I was a little taken aback by this question. After all, I thought most folks would be delighted to have an opportunity to give up the majority of their indexes, not just because of the space savings and DML performance benefits but also the maintenance nightmare that indexes can sometimes become.

Assuming this was a trick question, I deployed the standard stalling technique of answering a question with a question, “Can you tell me a little more about your situation?”

To which the system architect explained that they were in production with Oracle Database In-Memory on a 2 node RAC cluster running on commodity servers and a crap IO subsystem (his words, not mine). They had a snowflake schema, and had enough memory to accommodate all of their dimension tables but only the last 3 months of data in their two fact tables. Following my guidelines, they had kept their primary key indexes but dropped the rest of their indexes. He assured me that the performance of most of their queries had improved 100X and their ETL jobs were finishing 2X faster without the indexes but there were some queries that accessed more than just the last 3 months worth of data in the fact table and their performance had gotten worse, a lot worse.

It was in that moment that I realized that our guidance on dropping all reporting indexes with Database In-Memory had been based on an assumption that was not always true. The assumption I had been working under was; all of your performance critical data resides in memory or you have a good IO sub-system (engineered system etc.)

[Read More]

Friday May 22, 2015

Memory Usage with Oracle Database In-Memory

We often get asked the same two questions at conferences and customer presentations. That is  can't I just cache my tables in the buffer cache or put them in the keep pool, and how much memory do I need to allocate to the IM column store? While not directly related, I'm going to try and answer both questions in this post. First let's tackle the issue of what "in-memory" mean.

[Read More]

Monday May 18, 2015

Star Schema Performance Challenge - Part 1

Over the last 20 years or so I’ve had the opportunity to help build a variety business intelligence solutions for large organizations. It will be no surprise that one of the core requirements is fast query response times. This requirement for fast response times seems pretty obvious – faster response timeslead to greater end user satisfaction. The performance of a database can also determine what types of applications are practical. Breakthroughs in performance often lead to innovating new applications that can change how people work. A more efficient application is usually less expense to operate since it might not require as large of a server.

Oracle Database In-Memory and the availability of more powerful servers with many cores and plenty of memory really is a game changer, when it comes to improving query response times. With Database In-Memory far fewer CPU cycles are used to scan, filter, join and aggregate data. With modern servers, there are plenty of CPU cycles available for processing and  DRAM is relatively inexpensive. So, Database In-Memory can do more with less and, since Database In-Memory is a feature of the Oracle Database that works with all of the others performance enhancing features of the Oracle Database, it has become yet another tool in my big tool box of performance improving techniques.

[Read More]

The Oracle Database In-Memory blog is written by the Oracle product management team and sheds light on all things In-Memory.


  • Oracle
  • Oracle Database In-Memory
« November 2015