Performance Testing Series
This post is part of a series of blog posts about my performance tests in Azure SQL Database. For the first post in this series (and links to all of the other posts) please see here. For a summary of all the results please see here.
For a general overview of the test architecture, test components and test types, please see here.
Inserts Test Overview
The Inserts test involves calling a stored procedure to insert rows into a test table. The Cloud Service fires up a number of worker threads as specified in the test definition. The worker threads generate random data to be inserted. Row Id values (the primary key field) are generated sequentially. The worker threads then call a simple stored procedure to insert rows one-by-one (see stored procedure definition at the bottom of the post here).
The worker threads in the UT tests are not limited to a specified request rate. As soon as an insert stored procedure call has completed another is generated. The worker threads in the LT tests are constrained to generate requests at a limited rate specified in the test definition. The LT tests gradually increase the rate over the course of a number of tests.
UT Test Results
Results from the 30 UT Inserts tests are shown in the two charts below. In these charts, the “T” values are the number of worker threads generating requests against the database, e.g. “Business 4T” = a test against Business edition, running with 4 threads continuously generating requests.
As you may have noticed, the two charts above are more or less the same chart – because on average the row size was around 400 bytes throughout. The data volume figures here are based on the data content (i.e. summing the sizes of the data values, according to the number of bytes each data value requires in the SQL Server Data Type scheme but ignoring internal SQL Server overheads).
Several different tests of the current Web/Business edition were carried out. The Web 1T test was carried out on a A1 sized cloud instance, the Business 1T test on an A2 sized instance. This explains the higher throughput on the Business 1T test. As the number of threads was increased in the Business edition tests, the throughput increases though is variable.
During two of the three Business 8T tests a short outage occurred (one during the second test and one during the third). This is likely a failover of the test database as Azure balances workload. Interestingly, one of those outages occurred during the second of the Business 8T tests (the highest performer of the three) indicating the presence or absence of short outages had no significant effect on the results here. Nonetheless, the fact the failovers occurred in two out of three of the Business 8T tests (but none in any other test) indicates we were very likely approaching the maximum tolerable load.
In contrast to the Business Edition tests, performance of the new service tiers is very consistent. From these results S2 generally equals or outperforms Business Edition. Interestingly, for this workload, the gap between S2 and P1 is not all that significant, but P2 is substantially better than P1 and towers above Business Edition.
I will demonstrate below that all of the test results for the new service tiers shown above are limited by the SQL Server Log Write Rate limit, i.e. this is the determining factor in the performance.
The charts below show the performance profiles over the lifetime of the tests in the charts above. Since 30 lines is rather a lot to fit onto one chart, data for the different editions / tiers has been split across several charts.
It is also worth noting that due to the way the Azure SQL Database logging into sys.resource_stats works, data is only available at five minute intervals and the last data point in each test (which would have been plotted as minute 27 in these charts) is not useful and so is omitted.
Basic to P1 Tiers
Performance of the new tiers is very consistent – other than a blip towards the end of one of the S2 tests. The charts show that these tests weren’t constrained on CPU but were clearly constrained on log write quota.
Std2 to P2 Tiers
Again, these profiles illustrate good consistency, with the exception of the same S2 blip in one of the three S2 tests.
These profiles show the much more variable nature of the Business Edition. Those tests where multiple threads are executing show considerably more variable performance. Crudely speaking, this is the Wild Wild West of performance.
In addition, the Log Write profile looks suspiciously odd, being at 100% for all data points in all tests. More on this below.
LT Test Results
Three series of LT tests were carried out, against the Web, Basic and S1 editions / tiers.
The chart below illustrates how the performance of the Basic Tier varies as the workload (number of requests per second) is increased. A 30 minute test was conducted at each request rate.
As the request rate is increased, the CPU utilisation and log write utilisation increase. At around 70 rows per second, we have hit the log write limit and the database is not able to keep up with the request rate.
Performance is very consistent throughout.
Again, a similar type of profile to the Basic tier database – in this case we hit the log write rate limit at 140 rows per second.
The web edition is slightly less consistent – i.e. we start to miss requests in a more rough and ready manner.
It is interesting to note that this chart also proves the well known fact that there is no clearly enforced log write limit per database in the Web Edition. I.e. we apparently hit the reported log write limit at 30 requests per second, but we can actually continue increasing our request rate to four times this “maximum” – i.e. 120 requests per second – before we actually see a significant number of requests failing to be processed.
Inserts Test Conclusions
The Insert Tests have shown that, for a workload of purely stored procedure based row-by-row inserts (of average row size 400 bytes), performance of the new S2 / P1 tiers generally equals or exceeds the current Business Edition. P2 significantly outperforms Business for this workload. More specifically:
|Service Tier||Insert Limit||Insert Limit Comparison|
|Web / Business =||4.8 MB per minute||–|
|Basic =||1.4 MB per minute||29% of Web / Business|
|Standard S1 =||2.7 MB per minute||58% of Web / Business|
|Standard S2 =||5.5 MB per minute||120% of Web / Business|
|Premium P1 =||7.0 MB per minute||150% of Web / Business|
|Premium P2 =||13.5 MB per minute||280% of Web / Business|
The new service tiers also show an impressive level of consistency compared to the Business edition.
Appendix – UT Test Configuration
|Cloud Svc Inst. Size||A1||A2||A2||A2||A3||A1||A1||A2||A2||A2|
|Req. Gen. Thread Count||1||1||2||4||8||1||1||2||4||8|
|Initial Test Data (MB)||0||0||0||0||0||0||0||0||0||0|
Appendix – UT Test Results
|Configuration||Avg Rows Per Second||Avg MB Per Minute|
|Std S1 1T||124||132||131||2.64||2.80||2.79|
|Std S2 2T||254||269||248||5.42||5.74||5.29|
|Prem P1 4T||316||333||332||6.75||7.11||7.08|
|Prem P2 8T||619||646||635||13.21||13.79||13.57|
|Configuration||SQL Avg Log Write %||SQL Avg CPU %|
|Std S1 1T||96.0||96.3||96.0||35.3||31.6||34.1|
|Std S2 2T||98.7||99.3||88.4||34.8||35.9||29.9|
|Prem P1 4T||99.7||99.7||99.8||14.4||13.8||13.5|
|Prem P2 8T||97.1||97.5||95.4||18.9||19.5||20.6|
|Configuration||Cloud Svc Avg CPU||Error Count|
|Std S1 1T||3.5||3.5||3.5||0||0||0|
|Std S2 2T||3.8||4.3||3.6||0||0||0|
|Prem P1 4T||4.2||3.8||4.3||0||0||0|
|Prem P2 8T||7.0||7.4||7.3||0||0||0|
Appendix – LT Test Configuration
|Cloud Svc Inst. Size||A1||A1||A1|
|Req. Gen. Thread Count||1||1||1 – 2|
|Min Req. Gen. per Sec.||1||1||1|
|Max Req. Gen. per Sec.||200||100||200|
|Initial Test Data (MB)||0||0||0|
Appendix – LT Test Results
|Threads||Req. Per Sec||Avg Rows Per Second||Avg MB Per Minute|
|Web||Basic||Std 1||Web||Basic||Std 1|
|Threads||Req. Per Sec||SQL Avg Log Write %||SQL Avg CPU %|
|Web||Basic||Std 1||Web||Basic||Std 1|
The Requests per Second maximum applies in total, not per thread.