ODC Appreciation Day : Oracle Exadata Database Machine

Those that know me well, will know about my “appreciation” of the “Oracle Exadata Database Machine“, more commonly known as “Exadata” 🙂

So this will be my contribution to ODC Appreciation Day formally known as OTN Appreciation Day, a great initiative by Tim Hall aka Oracle-Base.com.

You can see a summary of last year’s blog post here:
OTN Appreciation Day : Summary

The very first Exadata, was the V1 model, the hardware by HP and the software by Oracle.  I still remember being very excited by this in my previous employment at Auto Trader and trying very hard to convince them to get one 🙂

I, of course, became an instant fan of the brawn hardware with smart software, Oracle labelling as “Hardware and Software optimised together“.

Oracle’s partnership with HP only lasted a year with Oracle switching to Sun on the V2 model, when shortly after Oracle then brought Sun in 2010.  This is when Oracle switched from the V models to X models, with the initial models being the X2-2 (2 socket) and X2-8 (8 sockets).

I still remember this old video “Oracle Exadata. Are You Ready?” that I played at an internal Auto Trader conference which was about sharing knowledge, interesting new things, etc:

Exadata has come a long way since the initial release that was aimed at being a Data Warehouse to a Full On OLTP, Data Warehouse, mixed load, consolidation platform, etc with “record-breaking” IOPS and scan rate!

My favourite feature is the smart scan, the ability to off load data intensive SQL operations from the database servers directly into the storage servers, mitigating the need to pull lots of data from storage to database server.  Yes you can have very fast All Flash Storage, but the network to ship all this to the database server becomes the bottleneck and the compute to filter the data on the database server.  Exadata does this at the storage server meaning only the rows and columns that are directly relevant to a query are sent to the database servers.

Another one is storage indexes where the min and max values are stored of a column in 1Mb chunk in memory to allow for unnecessary I/O to be avoided when it’s known that block of data doesn’t meet the predicate condition.

I didn’t manage to convince Auto Trader, however I have since been very fortunate in my current employment at Version 1 to have worked on Exadata since 2014 from the Exadata X2-2 through to X5-2.  I do really appreciate these “Engineered Systems” for the Extreme Performance, Reliability and Availability.  The whole concept of being “Engineered” and the whole stack optimised, really works and the fact that all Exadatas are the same hardware makes you appreciate their supportability.  Even patching them with patchmgr is pretty much a doddle these days! 🙂

For more info, visit the following site:
www.Oracle.com/Exadata
https://en.wikipedia.org/wiki/Oracle_Exadata

Tuesday 10th October 2017

Happy ODC Appreciation Day! #ThanksODC #ThanksOTN 🙂

If you found this blog post useful, please like as well as follow me through my various Social Media avenues available on the sidebar and/or subscribe to this oracle blog via WordPress/e-mail.

Thanks

Zed DBA (Zahid Anwar)

Advertisements

VMware Expert Database Workshop Program Oracle Edition – Day 2

Day 2 kicked off again with another early start at 7am, coupled with the late night, I was a bit tired to say the least, but was all worth it 🙂

Again, it was a very intense day, with lots of presentations and technical deep dives, ending with a lab session:

  • My OCM buddy Yvonne Murphy, kicked the day off by talking to us about “The Best Oracle Support Team on Earth – Global Support Services Oracle Team”
  • Dave Welch, House of Brick, talked to us about “Oracle on vSphere Licensing”
    • Some interesting Oracle license cases of which the main one discussed is available here
  • Jad El-Zein, talked to us about “vRealizeAutomation and Oracle”
  • Andreas Scherr, talked to us about “vSphere Core Storage Fundamentals” & “Modern Converged Storage, & vSAN & vVols”
    • I was impressed with the vSAN concept of using spare drive bays in an ESXi host to put a Hard Disks or SSD Drives in and/or using spare PCIe slots to put a Flash Cards in (including NVMe 🙂 )
    • Then these server attached storage devices are pooled together to provide a shared Datastore that has resilience built in using software rather then hardware 🙂
    • Can be all flash or hybird but each ESXi needs a cache device i.e. SSD Drive or Flash Card
    • More info can be found here
  • Valentin Bondzio, gave us an very enjoyable deep dive on CPU usage in Virtual Environment and how to troubleshoot 🙂
  • We then finished the day with a lab session with Sudhir Bala and Mohan Potheri, where we got to for example:
    • Create a Virtual Hard Drive
    • Attach an Existing Virtual Hard Drive (useful for RAC clusters 🙂 )
    • Then we played a game to stress test a Pure Storage, however I could only get 2Gb a second using 10 sessions running a parallel query of 10 on the largest object in the database.  This is not because the Pure Storage but because of I/O queue in the VMware stack, which we didn’t get time to change but highlight the point 🙂

Another great day of the 3 days workshop, in which I got to gain even more new knowledge in regards to VMware, in particular CPU usage and vSAN 🙂

The day ended with a meal and drinks at The Oliver Plunkett, more socialising with Johannes AhrendsRon EkinsFrits HooglandTim Hall and Mohan Potheri.  Then a nice walk back to the hotel with Mohan PotheriJohannes Ahrends and Martin Klier.  It’s a small world as I found Martin Klier and I had overlap with some customers and people 🙂  The socialising then continued in the reception lounge with Carl Dudley, Ron EkinsFrits Hoogland and Tim Hall.

Many thanks to VMware and Pure Storage, I’m looking forward to the rest of the workshop 🙂

My tweets for the day can be seen here.

The VMware Expert Database Workshop Program hashtag is #VMWORA

My related Blog Posts

VMware Expert Database Workshop Program Oracle Edition
VMware Expert Database Workshop Program Oracle Edition – Day 1

Other related Blog Posts

Tim Hall (Oracle Base) – VMware Expert Database Workshop Program Oracle Edition
Tim Hall (Oracle Base) – VMware Workshop – The Journey Begins
Tim Hall (Oracle Base) – VMware Workshop – Day 1
Tim Hall (Oracle Base) – VMware Workshop – Day 2
Michael Corey (Columnist) – VMware Experts Program Oracle Edition
Michael Corey (Columnist) – Day 1 VMware Experts Program Oracle Edition
Michael Corey (Columnist) – Day 2 VMware Experts Program Oracle Edition

If you found this blog post useful, please like as well as follow me through my various Social Media avenues available on the sidebar and/or subscribe to this oracle blog via WordPress/e-mail.

Thanks

Zed DBA (Zahid Anwar)

Assess Performance using Calibrate on Exadata

For those who are fortunate to have an Oracle Exadata Database Machine, may wonder if their Exadata meets the IOPS/MBPS as per the technical specifications.  Well with the command CALIBRATE in CellCLI, you can run raw performance tests on the cells’ hard disks and flash drives, enabling you to verify the disk/drive performance:

[root@v1ex1celadm01 ~]# cellcli
CellCLI: Release 12.1.2.3.4 - Production on Tue Jun 13 19:02:05 IST 2017

Copyright (c) 2007, 2016, Oracle. All rights reserved.

CellCLI> calibrate force;
Calibration will take a few minutes...
Aggregate random read throughput across all hard disk LUNs: 1823 MBPS
Aggregate random read throughput across all flash disk LUNs: 9973 MBPS
Aggregate random read IOs per second (IOPS) across all hard disk LUNs: 3002
Calibrating hard disks (read only) ...
LUN 0_0 on drive [8:0 ] random read throughput: 152.00 MBPS, and 243 IOPS
LUN 0_1 on drive [8:1 ] random read throughput: 157.00 MBPS, and 246 IOPS
LUN 0_10 on drive [8:10 ] random read throughput: 161.00 MBPS, and 253 IOPS
LUN 0_11 on drive [8:11 ] random read throughput: 157.00 MBPS, and 251 IOPS
LUN 0_2 on drive [8:2 ] random read throughput: 157.00 MBPS, and 244 IOPS
LUN 0_3 on drive [8:3 ] random read throughput: 158.00 MBPS, and 245 IOPS
LUN 0_4 on drive [8:4 ] random read throughput: 156.00 MBPS, and 248 IOPS
LUN 0_5 on drive [8:5 ] random read throughput: 161.00 MBPS, and 250 IOPS
LUN 0_6 on drive [8:6 ] random read throughput: 159.00 MBPS, and 252 IOPS
LUN 0_7 on drive [8:7 ] random read throughput: 158.00 MBPS, and 251 IOPS
LUN 0_8 on drive [8:8 ] random read throughput: 157.00 MBPS, and 251 IOPS
LUN 0_9 on drive [8:9 ] random read throughput: 159.00 MBPS, and 254 IOPS
Calibrating flash disks (read only, note that writes will be significantly slower) ...
LUN 1_1 on drive [FLASH_1_1] random read throughput: 2,157.00 MBPS, and 280525 IOPS
LUN 2_1 on drive [FLASH_2_1] random read throughput: 2,156.00 MBPS, and 274304 IOPS
LUN 4_1 on drive [FLASH_4_1] random read throughput: 2,158.00 MBPS, and 282083 IOPS
LUN 5_1 on drive [FLASH_5_1] random read throughput: 2,160.00 MBPS, and 287786 IOPS
CALIBRATE results are within an acceptable range.
Calibration has finished.

CellCLI>

 

The CALIBRATE FORCE, allows the test to run when CELLSRV is still up, this is acceptable if there is no user workload.  It is therefore recommended to not run during normal operations.  Without the FORCE, the CELLSRV must be shut down.

PLEASE NOTE: This is a single run on a single storage cell, you will need to run on all storage cells in the Exadata Machine to get the total IOPS/MBPS.  You can use dcli to run this across all the cells 🙂

If you found this blog post useful, please like as well as follow me through my various Social Media avenues available on the sidebar and/or subscribe to this oracle blog via WordPress/e-mail.

Thanks

Zed DBA (Zahid Anwar)