A a database professional, you should be most interested in the dual socket Xeon E5-2600 family, which has models with 2, 4, 6, and 8 CPU cores (plus hyper-threading). If you are considering upgrading to SQL Server 2012 (which also went RTM yesterday), you would not want a dual-core version, since the new core based licensing model for SQL Server 2012 Enterprise Edition requires you to pay for a minimum of four cores per processor socket. If you need to be particularly cost sensitive about your SQL Server 2012 licensing costs, but still want the best performance possible, you should be looking at the Xeon E5-2643, which has four-cores (plus hyperthreading), a 3.3GHz base clock speed, and a 10MB L3 cache.

If you want the most scalability, regardless of the SQL Server 2012 license costs, you should be looking at the Xeon E5-2690, which has eight cores (plus hyperthreading), a 2.9GHz base clock speed (with turbo boost to 3.8GHz), and a 20MB L3 cache. If your server platform offers it, you also might consider the Xeon E5-2687W, which has a 150W TDP and a base clock speed of 3.1GHz, but is otherwise identical to the E5-2690.

As usual, Anandtech has a good, early review, complete with some SQL Server 2008 R2 benchmark results. They also include SQL Server 2008 R2 results comparing performance under the Windows Balanced and High Performance Power Plans. Tom’s Hardware has another review, that is more focused on high-end workstation performance, since you can buy two socket workstations with Sandy Bridge-EP processors.

Of course in order to use this new processor family, you are going to need a new server platform that supports it, so the Tier 1 server vendors have all released new server models for the Sandy Bridge-EP. For example, Dell has the 12th generation PowerEdge R720 and R720xd models, while HP has the DL380p Gen 8 line.  I am pretty excited by the specifications available in these new models. For example, the Dell PowerEdge R720xd has 24 memory slots (that can support 32GB DIMMs if you have deep enough pockets), 26 2.5 inch internal drive bays, and six PCI-E 3.0 slots (two x16, and four x8). You can also get 10Gb Ethernet support with a daughtercard.  Honestly, if you can’t build a very fast two socket SQL Server database server with one of these, you are not really trying!

In Q1 of 2012, you will want to keep an eye on the quad-core, 3.3GHz Xeon E5-2643 with 10MB of L3 cache, which is part of the 32nm Sandy Bridge-EP family. Even though they have a slightly lower clock speed and a smaller L3 cache than the Xeon X5687, they should do much better for database workloads because of their improved quad-channel memory controller and support for PCI-E 3.0, among other things.

On the Intel desktop processor front, there has been a lot of leaked information about the upcoming (Q2 2012) 22nm Ivy Bridge processors over the past few days. Ivy Bridge is a Tick release that takes the existing Sandy Bridge architecture and shrinks it down with a move to the 22nm manufacturing process. Some initial benchmarks show anywhere from a 7%-25% application performance increase at the same clock speed, while power usage is also reduced. There are much larger increases in integrated graphics performance, which is plenty good enough for non-3D gaming usage. If you are thinking about buying a new desktop system, you might want to hold off until April 2012 (unless you really need something now).

This Tick-Tock release strategy benefits the DBA in a number of ways. It offers better predictability regarding when major (Tock) and minor (Tick) releases will be available. This helps the DBA plan upgrades.

Tick releases are usually socket-compatible with the previous year’s Tock release, which makes it easier for the system manufacturer to make the latest Tick release processor available in existing server models quickly, without completely redesigning the system. In most cases, only a BIOS update is required to allow an existing system to use a newer Tick release processor. This makes it easier for the DBA to maintain servers that are using the same model number (such as a Dell PowerEdge R710 server), since the server model will have a longer manufacturing life span.

As a DBA, you need to know where a particular processor falls in Intel’s processor family tree if you want to be able to meaningfully compare the relative performance of two different processors. Historically, processor performance has nearly doubled with each new Tock release, while performance usually goes up by 20-25% with a Tick release.

Some of the recent Intel Tick-Tock releases are shown in Figure 1.

Figure 1: Intel’s Tick-Tock Release Strategy

The manufacturing process technology refers to the size of the individual circuits and transistors on the chip. The Intel 4004 (released in 1971) series used a 10-micron process; the smallest feature on the processor was 10 millionths of a meter across. By contrast, the Intel Xeon “Westmere” 5600 series (released in 2010) uses a 32nm process. For comparison, a nanometer is one billionth of a meter, so 10-microns would be 10000 nanometers! This ever-shrinking manufacturing process is important for two main reasons:

Increased performance and lower power usage – even at the speed of light, distance matters, so having smaller components that are closer together on a processor means better performance and lower power usage.
Lower manufacturing costs – since you can produce more processors from a standard silicon wafer. This helps make more powerful and more power efficient processors available at a lower cost, which is beneficial to everyone, but especially for the database administrator.

The first Tock release was the Intel Core microarchitecture, which was introduced as the dual-core “Woodcrest” (Xeon 5100 series) in 2006, with a 65nm process technology. This was followed up by a shrink to 45nm process technology in the dual-core “Wolfdale”  (Xeon 5200 series) and quad-core “Harpertown” processors (Xeon 5400 series) in late 2007, both of which were Tick releases.

The next Tock release was the Intel “Nehalem” microarchitecture (Xeon 5500 series), which used a 45nm process technology, introduced in late 2008. In 2010, Intel released a Tick release, code-named “Westmere” (Xeon 5600 series) that shrank to 32nm process technology in the server space. In 2011, the Sandy Bridge Tock release debuted with the E3-1200 series for single socket servers and workstations.  All of these other examples are for two socket servers, but Intel uses Tick Tock for all of their processors. Figure 2 shows the recent and upcoming Tick-Tock releases in the two socket space.

Figure 2: Intel’s Tick Tock Milestones

When the operating system requests the highest processor performance state (P0), individual cores of the processor have their core speed increased temporarily by varying amounts, based on how many cores are busy, the total power usage of the processor, the ambient temperature of the system, etc.

This core speed increase is very helpful for OLTP workloads, especially if they are processor dependent. The amount of the possible speed increase varies according to the model of the processor, with the newer models with Turbo Boost 2.0 typically getting a 400MHz speed boost over the base clock speed.

I really don’t see any possible downside to this. The base clock speed on Intel processors is rated quite conservatively, with lots of thermal headroom. Having various algorithms in place to temporarily increase the clock speed of individual cores poses very little risk of causing any problems, such as overheating or shorter life for the processor.

One free tool you can use to monitor the effect of Turbo Boost in real-time is called TMonitor, which is available from cpuid.com. A screenshot of the tool in action is shown in Figure 1.

Figure 1: TMonitor running while the system is under a load

Figure 2 shows the Core Speed of Core #0 running at 2932MHz (which is the same as 2.93GHz), even though the rated base clock of my Core i7 930 is only 2.8GHz. The Core i7 930 only has the 1st generation Turbo Boost Technology, which is much less aggressive than the newer Turbo Boost Technology 2.0 in the Sandy Bridge processors.

Figure 2: CPU-Z showing Core #0 running above rated speed

Figure 1: New Intel Xeon Processor Numbering System

The E3 Product Line is for single processor server or workstations. The first generation of this line (E3-1200 series)is essentially the same as the desktop Sandy Bridge processors that were released in January 2011. The E7 Product Line (the Westmere-EX) has different models that are meant for two socket servers, four socket servers, and eight socket and above servers. The E7-2800 Series is for two socket servers, the E7-4800 Series is for four socket servers, while the E7-8800 Series is for eight socket and above servers. Later this year, Intel will release the E5 Product Line, which is the Sandy Bridge-EP.

After the Product Line designation, you have a four digit number that tells you more details about the particular processor. The first digit is the “wayness”, which is the number of physical CPUs that are allowed in a “node” (which is a physical server). This first digit can be 1, 2, 4, or 8. The second digit is the socket type, in terms of its physical and electrical characteristics. The last two digits are the processor SKU, with higher numbers generally being higher performance. Finally, you may have an L at the end, which is for energy efficient, low electrical power processors.

Knowing how to decode the processor model number is a very handy skill to have when you want to be able to understand the capabilities, relative age, and relative performance of a particular processor. An example of an Intel processor number is shown in Figure 1.

Figure 1: Classic Intel Xeon Processor Numbering

Intel Xeon processor numbers are categorized in four digit numerical sequences, plus an alpha prefix to indicate electrical power usage and performance. The alpha prefixes are as follows:

   •  X   meaning Performance
    •  E   meaning Mainstream
    •  L   meaning Power-Optimized

The model number will start with a 3, 5, or 7, depending on the server form factor that the processor is designed for. If the processor number starts with a 3, it is designed for a single socket server,if it starts with a 5, it is designed for a two socket server, and if it starts with a 7, it is designed for a four socket or more server.

The second digit of the model number designates the generation, or relative age of a processor. For example the Xeon 5100 series was launched in Q2 2006, while the Xeon 5300 series was launched in Q4 2006, and the Xeon 5400 series was launched in Q4 2007.

For a more complete example, a Xeon X7560 is a high-end Performance processor for multi-processor systems, an Intel Xeon E5540 is a Mainstream processor for dual-processor systems, while an Intel Xeon L5530 is a Power-Optimized processor for dual-processor systems. The final three digits denote the generation and performance of the processor; for example, a Xeon X7560 processor would be newer and probably more capable than a Xeon X7460 processor. Higher numbers for the last three digits of the model number mean a newer generation in the family, i.e. 560 is a newer generation than 460, in this example.

In my opinion, you should always choose the Performance models, with the X model prefix, for SQL Server usage. The additional cost of an X series Xeon processor, compared to an E series, is minimal compared to the overall hardware and SQL Server license cost of a database server system.

You should also avoid the power-optimized L series, since they can reduce processor performance by 20-30% while only saving 20-30 watts of power per processor, which is pretty insignificant compared to the overall electrical power usage of a typical database server (with its cooling fans, internal drives, power supplies, etc.). Of course, it would be a different story if you needed dozens or hundreds of web servers instead of a small number of mission critical database servers, since the overall power savings would be pretty significant.

My basic premise is that for a database server, you want the very best processor available for each physical socket in the server (since SQL Server Processor licenses are relatively expensive). Unlike a laptop or web server, you don’t want to pick a processor for a database server that is one or two models down from the most expensive, “top of the line” model”.

You will most likely be stuck with whatever processor you choose for the life of the server, since it rarely makes economic sense to upgrade the processors in an existing server. You can also use “excess” processor capacity for things like data compression or backup compression, to reduce the pressure on your I/O subsystem. Trading CPU utilization for I/O utilization is usually a net win, especially if you have a modern, multi-core processor that can shrug off the extra work.

So here is my recommended Intel processor list:

One socket server (OLTP)
Xeon E3-1280 (32nm Sandy Bridge)
•    3.50GHz, 8MB L3 Cache, 5.0 GT/s Intel QPI
•    Four-cores, Turbo Boost 2.0 (3.90GHz), hyper-threading
•    Two memory channels

One socket server (DW/DSS)
Xeon W3690 (32nm Westmere)
•    3.46GHz, 12MB L3 Cache, 6.40 GT/s Intel QPI
•    Six-cores, Turbo Boost (3.73GHz), hyper-threading
•    Three memory channels

Two socket server (OLTP)
Xeon X5690 (32nm Westmere-EP)
•    3.46GHz, 12MB L3 Cache, 6.40 GT/s Intel QPI
•    Six-cores, Turbo Boost (3.73GHz), hyper-threading     
•    Three memory channels

Two socket server (DW/DSS)
Xeon E7-2870 (32nm Westmere-EX)
•    2.40GHz, 30MB L3 Cache, 6.40 GT/s Intel QPI
•    Ten-cores, Turbo Boost 2.0 (2.8GHz), hyper-threading
•    Four memory channels

Four socket server (Any workload type)
Xeon E7-4870 (32nm Westmere-EX)
•    2.40GHz, 30MB L3 Cache, 6.40 GT/s Intel QPI
•    Ten-cores, Turbo Boost 2.0 (2.8GHz), hyper-threading
•    Four memory channels

Eight socket server (Any workload type)
Xeon E7-8870 (32nm Westmere-EX)
•    2.40GHz, 30MB L3 Cache, 6.40 GT/s Intel QPI
•    Ten-cores, Turbo Boost 2.0 (2.8GHz), hyper-threading
•    Four memory channels

First, I hate to see something like this happen to any company. It not only affects Intel, but their OEMs, retailers, and end-user customers. Many people will probably over-react and decide to skip Sandy Bridge, which I think would be a big mistake. It will also delay the widespread introduction of Sandy Bridge notebooks and the initial introduction of Sandy Bridge servers.

AnandTech does their usual good job of explaining the the issue here. It turns out that the issue (which is in the chipset on the motherboard, not the actual processor) only affects the four older 3Gbps SATA II ports on the motherboard (shown in light blue on the ASUS motherboard below). What may happen on a certain percentage of motherboards over time is that the connectivity of these ports might degrade and eventually stop working entirely. The two 6Gbps SATA III ports that are natively supported by the chipset are unaffected (they are light grey on the motherboard below). Many higher end motherboards have an additional Marvell controller that gives you two more 6Gbps SATA III ports (shown in dark blue below).

I have already built two Sandy Bridge Desktop systems, and I don’t plan on doing anything. I might swap out the motherboards when replacements are available, but I might not (since I am only using the 6Gbps SATA III ports).

This means that you can just use the 6Gbps SATA III ports (even for older 3Gbps SATA II devices) and you will be completely unaffected. Replacement motherboards should be available in perhaps four to six weeks, so you should be able to swap it out then, which will be a hassle. I have to give Intel some credit for how fast they have reacted to this problem.

The Sandy Bridge processor is a huge improvement over the older Lynnfield/Clarksfield/Clarkdale/Arrandale processors, with much better performance and lower power usage, which means better battery life for notebooks.

If you are in the market for a new notebook computer, I would still wait for a Sandy Bridge system with a Core i7 2820QM processor from your favorite vendor to be available. It will give you more than twice the performance of a Core i5 480M and roughly twice the performance of a Core i7 840QM system, with far better battery life.

It will also have native 6Gbps SATA III support, so you will be able to use one of the upcoming 2nd generation 6Gbps SSD drives that have sequential read and write speeds of over 400MB/sec.  This will absolutely humiliate the most fire-breathing, expensive, previous generation Core i7 system for both performance and battery life.

I ended up buying all of the parts needed to build a complete desktop system, including all this:

Intel Core i7 2600K processor

ASUS P8P67 Deluxe motherboard

VisionTek Radeon HD6850 1GB video card

(2) Corsair 8GB 1600MHz DDR3 kits (16GB of total RAM)

Crucial 128GB RealSSD C300 SSD

Western Digital 1.0TB Black 6Gbps hard drive

Thermaltake V9 BlacX Edition ATX case

SeaSonic X660 660 watt modular power supply

OEM 22X SATA DVD Recorder

The ASUS motherboard has four 6 Gbps SATA III ports (two Intel and two Marvel) and four USB 3.0 ports. Three of those USB 3.0 ports are on the front of the case. The Thermaltake case has two SATA II docking stations on the top, where you can just plug in a bare SATA drive. This system gets a 7.7 Windows Experience Index (WEI) score, due to the relatively weak video card, while the processor (at stock speed) comes in at 7.8. The memory score is 7.9 and the hard disk score is also 7.9, courtesy of the Crucial RealSSD.

Here is CPU-Z. I am still messing around with the BIOS settings. BTW, the ASUS motherboard has an EFI BIOS, which has a very nice GUI, and lets you use a mouse to navigate the BIOS setup screens. ASUS bumps the Bus Speed up to 103.0 Mhz by default.

Geekbench comes in at 12544, which is very impressive. That is about 50% better than a Core i7 950 system, and in the ballpark for a two socket, Xeon X5550 system.  I think the upcoming Sandy Bridge EP processors are going to be wicked fast for SQL Server workloads when they become available later this year.

Here is how the 128GB Crucial RealSSD C300 SSD does on Crystal DiskMark, plugged into one of the Intel 6 Gbps SATA III ports on the motherboard.  The 256GB Crucial RealSSD C300 was even better write performance than this. There were also several upcoming consumer SSDs announced at CES that will have substantially better performance than this.

Here is how the 1.0TB Western Digital Black 6Gbps drive does on Crystal DiskMark, plugged into one the other Intel 6 Gbps SATA III ports on the motherboard. This is the newer WD Black, with a 64MB cache.

This single socket desktop system has more CPU horsepower than most four socket database servers.