THE SQL Server Blog Spot on the Web

Welcome to SQLblog.com - The SQL Server blog spot on the web Sign in | |
in Search

Linchi Shea

Checking out SQL Server via empirical data points

Creating features for 1% of the users

While at PASS Summit 2011 in Seattle, I had a conversation with an attendee from one of the largest US companies. As expected and like any of the largest US companies, this company has multiple DBMS platforms.

The conversation then turned to how SQL Server was being used, and the attendee told me that they target SQL Server primarily for their second tier applications and Oracle is the platform of choice for the more critical tier 1 applications. The key reason, according to this person, is that SQL Server does not support active-active clustering and therefore when a server fails, there can be significant down time to fail over all the connections, a situation not acceptable to some critical applications.

Although it is arguable (1) how much a down time a SQL Server failover may incur, (2) to what extent that may not be acceptable, and (3) to what extent an active-active cluster may reduce the down time, the attendee was not the first one expressing this view (or bias if you will) and this did remind me of the comments that I constantly hear from the SQL Server community that Microsoft should not focus too much on creating features just for the largest enterprise customers that probably account for only 1% of its customer base (although revenue wise the percentage is probably much much higher).

I have always thought the comments on creating SQL Server features for only 1% of the customers to be very misleading and not helpful.

I don’t see Microsoft focusing too much on the largest enterprise customers, not to the extent that is detrimental to its smaller customers. In fact, I think Microsoft needs to focus more on the large enterprise customers in terms creating features they want. These features may not be needed by smaller customers today. But they lead the way, and the same customers very often will find their requirements change and start to use more higher-end enterprise features.

More importantly, these features are a must to narrow the gap between SQL Server and more established DBMS platforms, neutralizing any argument against its adoption on critical applications in large enterprise data centers.

Most importantly, these high-end and more challenging enterprise features (such as the aforementioned active-active clustering for a single instance) help make SQL Server a better product over all. After all, if it can meet the more stringent challenges of largest enterprises, it should have no problem serving easily as a platform for smaller customers.

I hope that in not too distant future when this same attendee reviews his DBMS platform strategy, he’ll still use not-having-active-active-clustering as the key differentiating factor between a tier 1 DBMS platform and a tier 2 DBMS platform except that he can’t no longer classify SQL Server as a tier 2 DBMS platform.

Published Monday, October 17, 2011 6:50 PM by Linchi Shea
Filed under: ,

Comment Notification

If you would like to receive an email when updates are made to this post, please register here

Subscribe to this post's comments using RSS

Comments

 

Stephen Colbert said:

I applaud you, sir, in speaking up for the 1%!

October 17, 2011 7:57 PM
 

Linchi Shea said:

Not in the true Stephen Colbert way though!

October 17, 2011 8:30 PM
 

Stephen Colbert said:

Very truthy.

October 17, 2011 10:32 PM
 

István Sáfár said:

I don't get it. Sql server does support Active/Active clustering!

We use it!

October 18, 2011 10:35 AM
 

Linchi Shea said:

Istvan;

Active/active is a loaded term. For a specific SQL Server instance, there is no support for active/active clustering. There is only failover clustering, meaning that the instance can only be running on a node at a time, although it can run on different nodes at different times. The activ/active clustering you are talking about necessarily involve multiple instances with different instances running on different nodes at the same time.

When you have a single-instance active/active cluster, you can have a node failure without being forced to restart the database instance, therefore without any outage.

October 18, 2011 11:14 AM
 

István Sáfár said:

ok, I see now.

October 18, 2011 4:07 PM
 

Mario said:

I wonder if it's really about the Enterprise features. As you rightly question the persons arguments for having an active-active cluster, it's the power and magic of Oracle marketing mostly, that brainwashed  customers believing they need things like RAC (and Oracle of course needs the extra licence income). RAC once developed as OPS, a necessity to make the first SMP (VAX) servers scale beyond 2 cpus, was kind of obsolete when we went beyond the 4 CPU's. RAC was developed for scalability not availability: the latter is a marketing invention.

November 13, 2011 4:47 AM
 

Truth About Abs said:

As  much as i know SQL server do not support active or active clustering.By the way thanks for the informative post.

December 9, 2011 5:45 PM
 

Windows Auditing said:

Hi,

I genuinely loved this brilliant article. Please continue this awesome work.

March 30, 2012 12:30 AM

Leave a Comment

(required) 
(required) 
Submit

About Linchi Shea

Checking out SQL Server via empirical data points

This Blog

Syndication

Powered by Community Server (Commercial Edition), by Telligent Systems
  Privacy Statement