Pythian Group

The end is nigh for Adobe Flash at My Oracle Support. The first step will happen during this weekend’s planned maintenance of My Oracle support, when all of MOS will be down for 5 hours starting at midnight eastern, Saturday January 28. Once it comes back up, the unadvertised non-Flash supporthtml.oracle.com will come up as a quasi-user acceptance test of the new application, while the flash-based support.oracle.com stays as-is. And once the bugs get worked out (“in the coming months”), the flash-based support.oracle.com will cease to exist and point to the new HTML-based site.
Read the rest of this entry . . .

The 2012 edition of RMOUG Training Days in Denver less than a month away, running February 15 and 16 at the Colorado Convention Center in Denver. Although it’s only two days, there’s a lot of technical content there, and a refreshing reduction in marketing-oriented presentations from “product managers”. It’s not too late to register, and it’s a pretty nice excuse to get to the Rockies in ski season. I’ll be doing two presentations, and am polishing up whitepapers and presentations for the submission deadline tomorrow:

• Exadata I/O Resource Manager: Ideal Technology for Database Consolidation, Wednesday 1:15pm in room 4B
• Peering Through the Cloud: The Straight Goods on Oracle Exalogic, directly after, Wednesday 4pm in room 4C

They’re right after each other in the grid, but I do get a short break for the dedicated exhibit hall time and paid vendor presentations (yes there are still a few; they have to pay the bills somehow).
Read the rest of this entry . . .

Hot on the heels of 11.2.0.3 coming out for Exadata, there’s yet another Exadata patch schedule out: the Quarterly Database Patch for Exadata (QDPE). They’re designed to being some of the predictability of Oracle’s quarterly critical patch updates (CPU) to the Exadata world. Behind the new naming, it looks like these are ordinary Exadata bundle patches, and even have BP numbers, but will have the predictable quarterly release schedule, synchronized with the CPU schedule (quarterly Tuesday nearest to 17th of the month it appears). Ordinary bundle patches aren’t going away quite yet though: there’s still a need to get patches out more frequently, and will still come out monthly or bimonthly on top of the quarterly patches. Oracle’s patching recommendations have changed too: QDPE patches are recommended, but other bundle patches are recommended only if experiencing issues resolved by them. From My Oracle Support note 888828.1, the following patches for Oracle 11.2.0.3 are planned:
Read the rest of this entry . . .

Since Oracle 11.2.0.3 came out last September, there have been questions about Exadata availability. As of today, the patchset is now available.

Reviewing the upgrade document (MOS note 1373255.1) a few things that jumped out at me:

• There is already a bundle patch (11.2.0.3 BP1, patch 13343057) that must be installed directly after the DB upgrade
• If running 11.2.0.2, a bugfix for unpublished bug 12539000 Synchronization problem in the IPC state affects ASM rolling upgrade and is required. BP12/13 have it (though the installer will still complain and must be ignored on install), and there are backports for BP7 through BP11.
• A recent storage server version (11.2.2.4.0+) is required, though with the critical issues fixed in 11.2.2.4.2, An upgrade there is probably in order.
• Install happens in a new ORACLE_HOME that should not be under /opt/oracle (presumably due to storage space limitations)
• Automatic memory management must be permanently disabled in the ASM instance, in favor of fixed SGA and PGA targets. Keep in mind that AMM would have prevented ASM from using hugepages in the past, and should be explicitly disabled with use_large_pages as part of the change.
• As for other database version upgrades, the data dictionary update requires system-wide downtime, though this can be minimized using a logical standby or GoldenGate.

Hat tip to R. Kundersma’s blog for the notification.

For the final day of UKOUG there are quite a few Pythian presentations. Unfortunately a lot of them happen at the same time, so it won’t be possible to attend them all live.

Using Oracle GoldenGate to Minimize Database Upgrade Risk

10:10, Media Suite
Marc Fielding

Even the best-planned database upgrades can leave nagging questions: what happens if my upgraded system performs unexpectedly? Is there a way I can go back to the previous version without downtime and data loss? Oracle GoldenGate allows DBAs to give affirmative answers to these questions. Drawing on upgrade experiences ranging from mid-sized databases to a large 10TB 90-CPU OLTP system, this session will show how to optimally configure Oracle GoldenGate, minimize downtime, maximize replication performance, and integrate Oracle GoldenGate into existing infrastructure such as Oracle ASM and storage-based replication.

Learn the various deployment scenarios where GoldenGate can minimize upgrade risk and delivery business value. Find out how GoldenGate can help not only during the upgrade, but after the upgrade as well. Learn how to use GoldenGate in medium- and high-volume environments. See real-world, tested GoldenGate configurations. Find out lessons learned and pitfalls to avoid.

The Answer to Free Memory Swap and Everything

10:10, Hall 8A
Christo Kutrovsky

Do I have enough memory? Why is my free memory so low? Am I swapping to disk? Can I increase my SGA (db cache) size? Can I add another instance to this server? Are my system resources used optimally? These are all questions that often haunt DBAs. This presentation is The Answer. It covers in detail the different types of memory, how to monitor memory, and how to optimally use it with Oracle. Multiple examples in the presentation demonstrate how certain actions on the database side cause different memory areas to be allocated and used on the OS side. Key underlying differences in operating systems approaches to managing memory will be highlighted, with special attention given to Linux and Solaris. Using Linux as an example throughout, this presentation explains how to effectively use tools such as “top”, “vmstat” and “/proc/meminfo” to look into into a system’s allocation and use of memory.

Amazon RDS, EC2 and S3 for Oracle Databases

11:20am, Hall 10B
Alex Gorbachev

This technical session focuses on specific recommendations and guidelines for leveraging the Amazon Web Services platform to host Oracle databases. We will looks into traditional database hosting using EC2 platform as well as recently introduced Amazon RDS on Oracle. We will look into how to configure, provision, backup, restore, monitor, and secure your databases in AWS. We will also look on how you can leverage S3 cloud storage for hybrid cloud deployments, particularly for backup and archival storage.

Backup and Recovery Roundtable

11:20am, Roundtable Area
Michael Abbey

A discussion of backup and recovery technology, problems and solutions. We will poll the attendees for an agenda on the day and proceed with an informal discussion not limited to:

RMAN, OSB, Sans, Data Guard backups, RAC backups and other topics of interest.

Concurrent Processing Performance Analysis for Apps DBAs

2:25pm, Hall 10B
Maris Elsins

Concurrent processing is one of the key elements of Oracle E-Business Suite, that’s used by most of modules for scheduling and processing background jobs. Keeping this functionality healthy is important to get maximum performance out of it. The paper describes the key metrics to estimate the performance of the concurrent managers, discusses approaches and techniques that can be used to understand how well the concurrent processing is set up, what are the bottlenecks and delays in processing of concurrent requests and provides tips on how to deal with each of the identified problem. This paper is targeted for Oracle Applications DBAs and technical consultants.

On tap for Tuesday is a 2-hour master class from Michael Abbey, along with an all-day drop-in RAC attack workshop with Alex Gorbachev and the RAC SIG.

Read the rest of this entry . . .

For those of you attending UKOUG today, there is a healthy dose of Pythian presentations on tap this afternoon. Actually, you can do it wall to wall 2:30pm to 6:30pm if you like.

To note:
Read the rest of this entry . . .

I was just looking at an Exadata X2-2 ordering document and noticed that it included 144GB of RAM. The sales rep pointed at the Exadata X2-2 datasheet and showed the 96GB to 144GB memory expansion option. Based on my reading of Intel Xeon (Nehalam) memory configurations, as long as each channel has a single dual-ranked module, all the memory can run at full 1333MHz speed. (Update: as noted in the comments, this is unfortunately not the case in Exadata; with the expansion unit memory runs at 800MHz). It populates the normally-empty third socket for each memory bank with an additional memory module.

It isn’t particularly cheap: $6250 per database node at US list price, but is a performance booster that doesn’t have ongoing support costs either. For OLTP environments, I like to say cache is still king, and even for those of you with pure data warehouses, 50% more PGA space can help out your sorts too.

And yes, I realize this isn’t particularly new; according to Kerry Osborne’s blog it came out (but wasn’t officially announced per se) at the same time as the storage expansion racks in the summer

The website for Oracle Database Cloud Services at cloud.oracle.com is now online, in conjunction with Larry Ellison’s announcement during the Oracle OpenWorld keynote going on now. It’s a hosted database service running Oracle 11gR2. The database can be accessed using a hosted Oracle application server, via JDBC across the Internet, or their own RESTful API a la Amazon. Notably lacking is Oracle’s own TNS network protocol.
Read the rest of this entry . . .

Update 7-October-2011: the log write caching capability has been officially announced as “Exadata Smart Flash Log”. I saw a few Oracle product management slides at OpenWorld presentations; one slide deck online is here on slide 30. A sample graph is provided, showing how the peak response times drop significantly with the additional cache.

These peaks would correspond to the times when the controller RAM cache is full. Another feature of the cache is that it returns write success status to the database when either flash or disk controller acknowledge the write, meaning tat the flash memory functions as a type of upper bound to redo write latency.

Exadata storage server software version 11.2.2.4.0 (patch link) has just been released. The readme file (My Oracle Support login required) lists 218 different changes, but one in particular sticks out:


11781936 NEED SMART FLASH LOGGING OF RDBMS REDO

Read the rest of this entry . . .