New TPC-E Results for SQL Server 2016

There have been two recent TPC-E OLTP benchmark results published for SQL Server 2016. These include one from Fujitsu and one from Lenovo.

The most recent result, from July 12, 2016 is for a four-socket FUJITSU Server PRIMERGY RX4770 M3 server that is using the latest generation, 14nm 2.2GHz Intel Xeon E7-8890 v4 processor (Broadwell-EX), with a TPC-E throughput score of 8,796.42. As is always the case with TPC-E benchmarks, the hardware vendor used the “flagship”, highest core count processor available from the latest processor family, in this case, a 24-core processor. This helps achieve the highest possible TPC-E throughput score (which is a measure of the total processor capacity of the system), at the cost of quite high SQL Server 2016 licensing costs, since you would have to purchase 96 SQL Server 2016 Enterprise Edition core licenses. This would cost about $684K at full retail price. Fujitsu priced the SQL Server 2016 licenses at $647K in the Executive Summary report.

Another recent result from May 31, 2016 is for a four-socket Lenovo System x3850 X6 that is using the same Intel Xeon E7-8890 v4 processor. This system gets a TPC-E throughput score of 9,068.00, which is about 3% higher than the Fujitsu system. Both systems use a 36TB initial database size, while the Fujitsu system uses 2TB of RAM and the Lenovo system uses 4TB of RAM (which is the license limit for Windows Server 2012 R2). Both systems use all flash storage, with 2.5” SAS SSDs.

Unlike the old TPC-C OLTP benchmark, TPC-E does not require an unrealistically expensive storage subsystem to get good scores. As long as the storage subsystem is “good enough” so that it does not become a bottleneck, then the ultimate TPC-E bottleneck becomes processor performance.

Earlier this year, there were two competing results for two-socket systems from Lenovo and Fujitsu. On March 30, 2016, Fujitsu published a result for a two-socket FUJITSU Server PRIMERGY RX2540 M2 system using the latest generation, 14nm 2.2GHz Intel Xeon E5-2699 v4 processor (Broadwell-EP), with a TPC-E throughput score of 4,734.87. The Intel Xeon E5-2699 v4 has 22 physical cores, so the two-socket system has a total of 44 physical cores that would need SQL Server 2016 licenses that would cost about $313K at full retail price. Fujitsu priced the SQL Server 2016 licenses at $296K in the Executive Summary report.

On March 24, 2016, Lenovo published a result for a Lenovo System x3650 M5 system using the same Intel Xeon E5-2699 v4 processors, with a TPC-E throughput score of 4,938.14, which is about 4% higher than the Fujitsu system. In this case, the Fujitsu system uses 1TB of RAM (with a 19TB initial database size), while the Lenovo system uses 512GB of RAM (with a 20TB initial database size). Both systems use all flash storage, with 2.5” SAS SSDs.

I know that this is a lot of numbers to be throwing around, so a summary of these four systems is shown in Table 1.

 

SystemProcessorRaw ScoreTotal CoresScore/Core
Lenovo System x3850 X6E7-8890 v49,068.009694.45
Fujitsu PRIMERGY RX4770 M3E7-8890 v48,796.429691.63
Lenovo System x3650 M5E5-2699 v44,938.1444112.23
Fujitsu PRIMERGY RX2540 M2E5-2699 v44,734.8744107.61

Table 1: Recent TPC-E Score Highlights

 

This shows that four-socket Broadwell-EX systems scale relatively well compared to older Xeon E7 processor families, meaning that the drop in single-threaded performance compared to equivalent two-socket Xeon E5 processor families is not as large as it used to be. There is still a gap though, which means that you are losing some scalability as you make the jump from a two-socket system to a four-socket system. If you can split your workload across two database servers, you would be better off to have two, two-socket servers rather than one, four-socket server. You would have more total processor capacity, better single-threaded performance, more PCIe expansion slots and lower SQL Server license costs.

An even better alternative for most people would be to use a lower core count, “frequency optimized” processor, instead of the flagship processor. For example, if you used the eight-core, 3.2 GHz Intel Xeon E5-2667 v4 processor in a two-socket server, you would get the estimated results shown in Table 2.

 

SystemProcessorRaw ScoreTotal CoresScore/Core
Estimated Two-Socket SystemE5-2667 v42611.9116163.24

Table 2: Estimated TPC-E Results

If you had four, two-socket systems with the eight-core, 3.2 GHz Intel Xeon E5-2667 v4 processor, instead of one, four-socket system with the 24-core 2.2 GHz Intel Xeon E7-8890 v4 processor, you would have about 15.2% more total processor capacity, about 72.9% better single-threaded performance, and a 33.3% lower SQL Server 2016 licensing cost (which would be about $227K in license savings). You would have the same total memory capacity, and more than three times the number of PCIe slots.

Recent SQL Server 2012 and 2014 Updates

Microsoft has released a number of SQL Server Cumulative Updates and Service Packs over the past several weeks. For SQL Server 2014, these include:

June 20, 2016    SQL Server 2014 RTM CU14 (12.0.2569)

June 20, 2016    SQL Server 2014 SP1 CU7 (12.0.4459)

July 11, 2016     SQL Server 2014 SP2 RTM (12.0.5000)

SQL Server 2014 RTM CU14 will be the last cumulative update for the SQL Server 2014 RTM branch, and it is now an “unsupported service pack”. If you are still on the RTM branch, you should be be planning on moving to either SP1 or preferably SP2. SQL Server 2014 SP2 RTM has all of the fixes that are in SQL Server 2014 SP1 CU7, so there is no need to wait for SQL Server 2014 SP2 CU1 in order to “catch up” to the previous branches. It also has a number of new features and performance improvements (which you can read about here), so I think people are going to want to move to the SP2 branch relatively soon.

You can find the official Microsoft Build list for SQL Server 2014 here:

SQL Server 2014 Build Versions

 

For SQL Server 2012, we have these updates:

July 18, 2016     SQL Server 2012 SP2 CU13  (11.0.5655)

July 18, 2016     SQL Server 2012 SP3 CU4    (11.0.6540)

As always, I think you are better off to be on the latest Service Pack for whatever version of SQL Server you are using. For SQL Server 2012, the RTM and SP1 branches are both considered “unsupported service packs”. You need to be on either SP2 or SP3, preferably SP3.

You can find the official Microsoft Build lists for SQL Server 2012 SP3 and SP2 here:

SQL Server 2012 SP3 build versions

SQL Server 2012 SP2 build versions

Finally, if you or your organization are still reluctant to deploy SQL Server Cumulative Updates, you should read the current official guidance from Microsoft about this. One of the key points is “we now recommend ongoing, proactive installation of CU’s as they become available”. This does not mean that you just blindly deploy a cumulative update to Production the day it is released. Rather, you should have a good testing and deployment plan that you go through before you deploy to Production. You can read the full Microsoft guidance here:

Announcing updates to the SQL Server Incremental Servicing Model (ISM)

Speaking at PASS Summit 2016

I will be presenting two sessions at the PASS Summit 2016 in Seattle, WA, which is being held October 25-28, 2016.

On Monday, October 24, 2016, I will be doing an all-day, Pre-Conference session on how to interpret my SQL Server diagnostic information queries. I have done many shorter versions of this session (such as 60 minutes, 75 minutes, or even a half-day) before, but I have always felt a little rushed as I went through the complete set of diagnostic queries, explaining how to interpret the results of each one, and also talking about related information that is relevant to each query.

Now, I will have a full day to go into more detail, without having to hurry to cover everything. I will be using the SQL Server 2016 version of the diagnostic queries, which have even more useful information, including information about many new SQL Server 2016 features. If you are on an older version of SQL Server, most of the queries will still be relevant (depending on how old of a version of SQL Server you are using).

Based on past experience and feedback, Dr. DMV has always been a very popular session that people really seem to enjoy. This all-day, expanded version is going to be really fun and useful, and I hope to see you there!

Dr. DMV: How to Use DMVs to Diagnose Performance Problems

SQL Server 2005 introduced Dynamic Management Views (DMVs) that allow you to see exactly what is happening inside your SQL Server instances and databases with much more detail than ever before. SQL Server 2016 adds even more capability in this area. You can discover your top wait types, most CPU intensive stored procedures, find missing indexes, and identify unused indexes, to name just a few examples. This session (which is applicable to SQL 2005-2016), presents and explains over seventy DMV queries that you can quickly and easily use to detect and diagnose performance issues in your environment. If you have ever been responsible for a mission critical database, you have probably been faced with a high stress, emergency situation where a database issue is causing unacceptable application performance, resulting in angry users and hovering managers and executives. If this hasn’t happened to you yet, thank your lucky stars, but start getting prepared for your time in the hot seat. This session will show you how to use DMV queries to quickly detect and diagnose the problem, starting at the server and instance level, and then progressing down to the database and object level. This session will show you how to properly analyze and interpret the results of every single query in the set, along with lots of information on how to properly configure your instance and databases.

 

This is a regular, 75 minute session that will be all new content, going into much deeper detail about the SQL Server related factors of current server hardware, and how to go about selecting the best database server hardware for your workload and budget. I want to show you how to pick hardware that gives you the best performance possible while minimizing your SQL Server license costs, saving your organization a huge amount of money!

Hardware 301: Diving Deeper into Database Hardware

Making the right hardware selection decisions is extremely important for database scalability. Having properly sized and configured hardware can both increase application performance and reduce capital expenses dramatically. Unfortunately, there are so many different choices and options available when it comes to selecting hardware and storage subsystems, it is very easy to make bad choices based on outdated conventional wisdom. This session will give you a framework for how to pick the right hardware and storage subsystem for your workload type. You will learn how to evaluate and compare key hardware components, such as processors, chipsets, and memory. You will also learn how to evaluate and compare different types of storage subsystems for various database workload types. Gain the knowledge you need to get the best performance and scalability possible from your hardware budget!

The PASS Summit is always a fun and very useful and educational event. It is a great way to get to know more people in the SQL Server community and to connect with people that you may only know online.

You can register for the PASS Summit 2016 here.