I have a database that is roughly 1.5gb total size.  One of the tables has approx. 5m rows in it.  Just executing a count:

select count(id) from dispensing

Takes almost 3 minutes to complete, no other queries being executed on the server.  Running this on a local sql instance on windows 7, through parallels, on a 2 year old macbook, takes < 1 second.  The id column is the primary key w/ clustered index. 

Here is a chart of utilization just from running the count query a few times: http://i.imgur.com/Rn4qz2e.png

Am I expecting too much from S0?  Would hate to see what would happen if I was using more than 0.5% of the size quota.

Can you also share CPU and physical disk read percentage as well , while running this query?

Like this?

http://i.imgur.com/HWCwgJx.png

 When it comes to S0 , it's not only about storage you need to look but also on these factors. If you see the graph , it seems like you are hitting 100% on data IO / DTU  percentage.

You probably need to try this query with higher performance tier. Check this post for more information - http://dbcouncil.net/2015/07/24/for-a-sql-server-dba-how-to-start-azure-learning-part-5-sql-azure-database-monitoring/

Just for fun, I spun up an RDS instance on AWS to compare.  I did it with sql express on a micro instance, that is 1vCPU and a comically low 1gb of ram, such a weak server Amazon gives them away for free.  End result: query executes in ~5 seconds.  Something doesn't seem quite right.

I understand it's not only about storage but when I am using a minuscule amount of space and having trouble just counting rows in a table, I can't imagine how it would be usable for anybody with a more realistic database workload.  If I had a very complicated query I would understand, and there would probably be opportunities to optimize it... So if its not something I am doing wrong and I am hitting 100% of the data io quota with the simplest of queries, and that appears to be expected, I guess the real question is: Why is MS providing such lame performance tiers?

I upped the db to S2 and now the query returns in 1-3 seconds on average (although the first execution still took over a minute).  A definite improvement, but for $75? I'll stick with free RDS if that is the only solution.

• Edited by svickers 11 hours 20 minutes ago

I understand it's not only about storage but when I am using a minuscule amount of space and having trouble just counting rows in a table, I can't imagine how it would be usable for anybody with a more realistic database workload.  If I had a very complicated query I would understand, and there would probably be opportunities to optimize it... So if its not something I am doing wrong and I am hitting 100% of the data io quota with the simplest of queries, and that appears to be expected, I guess the real question is: Why is MS providing such lame performance tiers?

I upped the db to S2 and now the query returns in 1-3 seconds on average (although the first execution still took over a minute).  A definite improvement, but for $75? I'll stick with free RDS if that is the only solution.

• Edited by svickers 11 hours 30 minutes ago

I understand it's not only about storage but when I am using a minuscule amount of space and having trouble just counting rows in a table, I can't imagine how it would be usable for anybody with a more realistic database workload.  If I had a very complicated query I would understand, and there would probably be opportunities to optimize it... So if its not something I am doing wrong and I am hitting 100% of the data io quota with the simplest of queries, and that appears to be expected, I guess the real question is: Why is MS providing such lame performance tiers?

I upped the db to S2 and now the query returns in 1-3 seconds on average (although the first execution still took over a minute).  A definite improvement, but for $75? I'll stick with free RDS if that is the only solution.

• Edited by svickers Thursday, August 27, 2015 7:47 PM

I understand it's not only about storage but when I am using a minuscule amount of space and having trouble just counting rows in a table, I can't imagine how it would be usable for anybody with a more realistic database workload.  If I had a very complicated query I would understand, and there would probably be opportunities to optimize it... So if its not something I am doing wrong and I am hitting 100% of the data io quota with the simplest of queries, and that appears to be expected, I guess the real question is: Why is MS providing such lame performance tiers?

I upped the db to S2 and now the query returns in 1-3 seconds on average (although the first execution still took over a minute).  A definite improvement, but for $75? I'll stick with free RDS if that is the only solution.

• Edited by svickers Thursday, August 27, 2015 7:47 PM

I understand it's not only about storage but when I am using a minuscule amount of space and having trouble just counting rows in a table, I can't imagine how it would be usable for anybody with a more realistic database workload.  If I had a very complicated query I would understand, and there would probably be opportunities to optimize it... So if its not something I am doing wrong and I am hitting 100% of the data io quota with the simplest of queries, and that appears to be expected, I guess the real question is: Why is MS providing such lame performance tiers?

I upped the db to S2 and now the query returns in 1-3 seconds on average (although the first execution still took over a minute).  A definite improvement, but for $75? I'll stick with free RDS if that is the only solution.

• Edited by svickers Thursday, August 27, 2015 7:47 PM

I understand it's not only about storage but when I am using a minuscule amount of space and having trouble just counting rows in a table, I can't imagine how it would be usable for anybody with a more realistic database workload.  If I had a very complicated query I would understand, and there would probably be opportunities to optimize it... So if its not something I am doing wrong and I am hitting 100% of the data io quota with the simplest of queries, and that appears to be expected, I guess the real question is: Why is MS providing such lame performance tiers?

I upped the db to S2 and now the query returns in 1-3 seconds on average (although the first execution still took over a minute).  A definite improvement, but for $75? I'll stick with free RDS if that is the only solution.

• Edited by svickers Thursday, August 27, 2015 7:47 PM

Thanks for the feedback.  We know that you have choices including AWS.  A more realistic comparison point to compare Azure SQL DB and AWS RDS would be a business related benchmark that's a more rounded mix of IOPS, CPU, and throughput.

In Azure SQL DB, we use this benchmark to measure our own performance and to understand where we stand compared to others. https://msdn.microsoft.com/en-us/library/azure/dn741327.aspx?f=255&MSPPError=-2147217396 .

Let me know if I can help further.

Guy

Thanks for the feedback.  We know that you have choices including AWS.  A more realistic comparison point to compare Azure SQL DB and AWS RDS would be a business related benchmark that's a more rounded mix of IOPS, CPU, and throughput.

In Azure SQL DB, we use this benchmark to measure our own performance and to understand where we stand compared to others. https://msdn.microsoft.com/en-us/library/azure/dn741327.aspx?f=255&MSPPError=-2147217396 .

Let me know if I can help further.

Guy

• Proposed as answer by Casey KarstMicrosoft employee, Moderator 12 hours 9 minutes ago

Thanks for the feedback.  We know that you have choices including AWS.  A more realistic comparison point to compare Azure SQL DB and AWS RDS would be a business related benchmark that's a more rounded mix of IOPS, CPU, and throughput.

In Azure SQL DB, we use this benchmark to measure our own performance and to understand where we stand compared to others. https://msdn.microsoft.com/en-us/library/azure/dn741327.aspx?f=255&MSPPError=-2147217396 .

Let me know if I can help further.

Guy

• Proposed as answer by Casey KarstMicrosoft employee, Moderator Friday, September 04, 2015 7:01 PM

Thanks for the feedback.  We know that you have choices including AWS.  A more realistic comparison point to compare Azure SQL DB and AWS RDS would be a business related benchmark that's a more rounded mix of IOPS, CPU, and throughput.

In Azure SQL DB, we use this benchmark to measure our own performance and to understand where we stand compared to others. https://msdn.microsoft.com/en-us/library/azure/dn741327.aspx?f=255&MSPPError=-2147217396 .

Let me know if I can help further.

Guy

• Proposed as answer by Casey KarstMicrosoft employee, Moderator Friday, September 04, 2015 7:01 PM

Thanks for the feedback.  We know that you have choices including AWS.  A more realistic comparison point to compare Azure SQL DB and AWS RDS would be a business related benchmark that's a more rounded mix of IOPS, CPU, and throughput.

In Azure SQL DB, we use this benchmark to measure our own performance and to understand where we stand compared to others. https://msdn.microsoft.com/en-us/library/azure/dn741327.aspx?f=255&MSPPError=-2147217396 .

Let me know if I can help further.

Guy

• Proposed as answer by Casey KarstMicrosoft employee, Moderator Friday, September 04, 2015 7:01 PM
• Marked as answer by Casey KarstMicrosoft employee, Moderator 15 hours 38 minutes ago