I have a table that holds aggregated data to support a PowerBI report. The table is only 8 columns, and uses INT, VARCHAR and DATETIMEOFFSET types. There's a regular PK int identity column. There is a clustered index on the PK.

There are 300k rows

I'm on db.t3.small instance on AWS RDS (2 vCPU, 2GB RAM)

My CPU is not spiking, and the DB is not being used for anything else at all -- this server is a clone for testing.


id: PK, int, not null
column1: varchar(16), not null
column2: varchar(50), not null
column3: varchar(50), not null
column4: int, not null
column5: int, not null
column6: datetimeoffset(7), not null
column7: int, null


Just a simple select takes a full minute, and that seems slow since there's no joining or even calculations. It's literally just selecting those rows.

SELECT * FROM report_MyTableName


  1. Is that a reasonable amount of time to wait? It just seems like such a long time to wait for a relatively few/moderate number of rows.
  2. Why does it show that there are 6 selects? there's no view or joining involved so I assumed 1 select.
  3. Why are there 7 round trips to the server?
  4. Is there a practical way to improve performance for this query?
  5. I ran a query on a similar table (with 214k rows) and it took < 2s on my local SQL Server instance on my laptop. I also had a coworker run a similar query on a different DB server (which admittedly has more horsepower) and it took < 10 seconds.


enter image description here enter image description here

If I've left out anything important let me know. I'm happy to update.

Requested Info

Here's the output from COUNT(*) enter image description here

enter image description here

And here's the output when I ran with STATISTICS IO, TIME

SQL Server parse and compile time: 
   CPU time = 0 ms, elapsed time = 0 ms.

(306214 rows affected)
Table 'report_MyTableName'. Scan count 1, logical reads 3098, physical reads 0, read-ahead reads 0, lob logical reads 0, lob physical reads 0, lob read-ahead reads 0.

(2 rows affected)

(1 row affected)

 SQL Server Execution Times:
   CPU time = 218 ms,  elapsed time = 42203 ms.
SQL Server parse and compile time: 
   CPU time = 0 ms, elapsed time = 0 ms.

 SQL Server Execution Times:
   CPU time = 0 ms,  elapsed time = 0 ms.

Completion time: 2021-01-28T17:06:36.4403170-05:00

Actual execution plan


3 Answers 3


The 6 Selects are due to live statistics. Never mind them.

Based on the fact that Client procesing time takes about 99% of the total time, issue is definitely network/client.

As proposed, try the

select count(*) from report_MyTableName


select * into #tmpTable from report_MyTableName

to check performance without network.

In statistics, there is the total amoun received from server. too bad the full info is missing so we can only guess about amount of data vs network speed. Either provide full info on bytes send or info about columns (types and size) so we can estimate the data size.

EDIT: So based on the count(*), it is obvious the total amount DB server works on it is about < 0.1s. The CPU time matches but the elapsed time is not normal.

How long did the count(*) take? Live query show ~0.25s, which seems right, elapsed time shows 42 seconds.

My guess is still that the issue is network/client, not server.

What is your ping against the server?


If you can get the actual execution plan, you should be able to confirm that the network is to blame for most or all of the runtime of this query.

I ran a very similar query on my laptop just now with the "Actual Execution Plan" setting turned on, then right-clicked the "Select" operator and chose "Properties."

Screenshot of QueryTimeStats and WaitStats in SSMS execution plan

The screenshot shows an elapsed time of 3263 ms, but only 536 ms of CPU usage. What explains the difference (2727 ms)?

Looking down in the wait stats, we have 2734 ms of ASYNC_NETWORK_IO waits, which accounts for the difference.

That wait type doesn't necessarily mean network issues - in my case, there is no network involved really (the database is also on my laptop). The wait is tracking how long it takes SSMS to render (consume) the rows coming from the server.


SELECT * FROM report_MyTableName
Rows returned by SELECT statements | 306220
Bytes received from server | 2.25036E+...

I'd really like to see all of that last line ...
The value is clearly big enough that SSMS decided to use an engineering format for it, so I'm guessing at least 2.5MB is coming back across the network here, but it could be considerably more.

Lifting and dropping large amounts of data takes time.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.