Oracle: Overhaul or Overkill?

SQLServerCentral is supported by Red Gate Software Ltd.

Popular Topics

Home

Members

Calendar

Who's On

Home » SQLServerCentral.com » Editorials » Oracle: Overhaul or Overkill?

Oracle: Overhaul or Overkill?

Rate Topic

Display Mode

Topic Options

Author

Message

camassey

Posted Saturday, February 20, 2010 1:00 PM

Grasshopper

Group: Administrators
Last Login: Tuesday, May 21, 2013 4:50 AM
Points: 14, Visits: 384

Comments posted to this topic are about the item Oracle: Overhaul or Overkill?

Veni Vidi Velcro
I came, I saw, I stuck around

Post #869706

SQLRNNR

Posted Sunday, February 21, 2010 6:44 PM

SSCoach

Group: General Forum Members
Last Login: 2 days ago @ 10:25 AM
Points: 18,754, Visits: 12,337

Interesting read. I hope MS does not go the same path as Oracle. I think R2 is far from that path. Customized hardware setups is not necessary in many setups and would alienate a lot of potential clients.

Jason AKA CirqueDeSQLeil
I have given a name to my pain...
MCM SQL Server 2008

SQL RNNR

Posting Performance Based Questions - Gail Shaw
Posting Data Etiquette - Jeff Moden
Hidden RBAR - Jeff Moden
VLFs and the Tran Log - Kimberly Tripp

Post #869987

Matt Miller (#4)

Posted Sunday, February 21, 2010 9:26 PM

SSCertifiable

Group: General Forum Members
Last Login: Yesterday @ 8:43 PM
Points: 6,998, Visits: 13,952

While Microsoft hasn't gone into "owning" specialized hardware, my understanding of the R2 "parallel data warehouse" setup is that as of right now you will need on-site help from a dedicated MS professional to install it. Some outlanding amount of hardware: four high-end servicers with a LOT of CPU's, with SQL Server datacenter, being licensed at $57K per CPU retail. Then again - for that kind of pricetag, I'd expect some help on site, thrown in.

Even without bring ing R2 - MS has had a host of high-end techs that can be engaged to deal with high-end setups for quite a while. These would be in addition to the Gold/platinum partners (It used to be nicnamed the "firm" by MS employees as I recall). Still not necessarily a LOT of customers need this level, but when they do it's for a high level of complexity and a LARGE investment, so the extra on-hand support should again be expected.

----------------------------------------------------------------------------------
Your lack of planning does not constitute an emergency on my part...unless you're my manager...or a director and above...or a really loud-spoken end-user..All right - what was my emergency again?

Post #870035

Irish Flyer

Posted Monday, February 22, 2010 8:44 AM

Ten Centuries

Group: General Forum Members
Last Login: Friday, June 08, 2012 7:51 AM
Points: 1,219, Visits: 225

It seems to me that while SQL Server is perfectly adequate tool for most users, who just do not place a lot of demand on their instances, it really doesn't scale very well for other more demanding users. I'll qualify that further. SQL Server scales quite well in an OLTP environment, provided you throw enough horsepower at it, and its architecture seems geared to this type of environment. It can handle really high volumes of tiny transactions without even blinking. Unfortunately, it does not scale very well at all for OLAP, with competing giant, long-running transactions.
Microsoft's answer to this is the "scale OUT" instead of "scale UP" (see their white papers "Introduction to New Data Warehouse Scalability Features in SQL Server 2008" and "Best Practices for Data Warehousing with SQL Server 2008" on MSDN). If you look at their recommendations critically, you see that their "scale out" simply removes all cost advantages they initially seemed to have over Oracle. As an example, for the main warehouse platform, they recommend you configure enough main memory to equal 20% of your total data warehouse size. Try that with a warehouse that is 30 TB and up. I am not sure there are any platforms out there that will allow configuring 6 TB of main memory or more. The other part of their recommended "scale out" is to place every OLAP cube you build on its own server. If you have a large warehouse that is actively used by your organization, you may be looking at 30 or more cubes to satisfy the needs of your user audience.
In addition, SQL Server does not cluster at all. All clustering is done at the operating system level, while SQL Server is only cluster "aware". As someone who manages a large SQL Server installation that has been using a big Itanium-based clustered server as one of our main platforms for the last year and a half, I have first-hand experience with what it can and can't do. We have operated in both SQL Server 2005 and SQL Server 2008. We have a very large data repository (not an architected warehouse) containing around 3500 tables in 120 databases (around 40 TB). Our largest table partitions have approximately a billion rows, and we are using date-based partitions. We have 22 analysts running giant OLAP mining queries that can take as much as 2 days to complete, plus all of the normal ETL activity to keep us supplied with current data, several hundred daily, weekly and monthly production jobs, and the standard array of backups and stats rebuilds any DBA needs to ensure continuity of operation. We frequently put this big Itanium-based server on its knees with this load.
On the other hand, in a previous job, I ran an Oracle ODS carrying even more load than this, with more users and more data, on smaller, less powerful server hardware. It did use the UNIX operating system instead of Windows, but even so, I never saw it even close to being overloaded.
It is also my experience, that in terms of DBA effort, SQL Server DBAs have to stay constantly on top of their instances, because they seem to be much more fragile than any Oracle instance I've worked on. Oracle is, to a large extent, self repairing, and it is very rare for there to be any corruption of data. SQL Server does not have any true data streaming for loads or exports of data. In fact, its ETL capabilities are incredibly crude and nasty. SSIS is not well integrated, has innumerable bugs and just doesn't work very well compared to most of the third party ETL tools like Informatica, etc.
I am tired of people pushing SQL Server as the database answer to all of ITs problems. It is very good at some things, but is a total dog at others.

Post #870415

Gift Peddie

Posted Monday, February 22, 2010 9:31 AM

Hall of Fame

Group: General Forum Members
Last Login: Friday, March 29, 2013 11:17 AM
Points: 3,432, Visits: 14,332

I am not sure Oracle will do this because Oracle implementation on Windows was 30%, then there is Oracle on IBM ISeries, HP UNIX, Linux and Solaris which I think was about 10% or less. So Oracle will replace Linux and to a less extent IBM ISeries because banks that runs Oracle on IBM ISeries may not be allowed to swap out ISeries for Solaris overnight.

And I find it very funny that Informatica which cost more than SQL Server license is compared to SSIS which is just a free subsystem of SQL Server.

I have also had the opportunity to watch both SQL Server and Oracle with monitoring tools and the most important point is the numbers to watch in SQL Server and Oracle are very different. The reason Embacadero remains a very profitable business.

Kind regards,
Gift Peddie

Post #870487

Steve Jones - SSC Editor

Posted Monday, February 22, 2010 4:26 PM

SSC-Dedicated

Group: Administrators
Last Login: Yesterday @ 11:20 AM
Points: 31,437, Visits: 13,752

There's probably value in places for having a complete SQL Server solution setup. I think the Fast Track setup was an attempt here. However for most users, I think we don't use all of SQL Server, so it's overkill.

Follow me on Twitter: @way0utwest

Forum Etiquette: How to post data/code on a forum to get the best help

Post #870815

« Prev Topic | Next Topic »

Permissions