I have been working as a full-time performance engineer at Intel for 6 years. I started by benchmarking server products for performance validation and now I focus on the TPC-C and TPC-E OLTP server benchmarks. I have used a variety of workloads in this job and spent time optimizing each level of the performance hierarchy: application, system, and processor. I, like many of you, have learned the "tricks of the trade" the hard way: by trial, error, and success. I'm sharing now, so you can all benefit from the things I've picked up along the way.

Let's start with some general methodologies to follow when tuning performance, whether you do it full-time, as a hobby, or just in your spare cycles after getting your "regular work" done. I will follow up with a more detailed post on each habit individually.

1. Ask the right question: Why are you tuning your platform? What level of performance are you hoping to achieve? What do you (or your users) care most about: raw performance, cost/performance, performance/watt, or something else?

2. Start at the top: The first and easiest part of your application server to tune is the hardware itself. Move on to the software and workload only after you feel confident that you have removed any system-level bottlenecks.

3. Know your Platform: This should be where you begin your system (hardware) tuning. The first thing, which I can't stress enough, is to get a block diagram of your platform. Then study it!

4. Know your BIOS: Server BIOSes these days come with more and more options. Be sure to give your new platform's BIOS a once-over. Pay particular attention to options relating to performance and power.

5. Know your Workload: To quantify performance, you need a workload! Some examples: web server response time, boot time, frames rendered per second, simultaneous connections supported, etc. Understand as much as possible about how the work gets done.

6. Try one thing at a time: Little changes that seem harmless can significantly alter the behavior of your system. Or worse, they can interact with each other to wreak havoc. Always try one change at a time, and for goodness' sake, do habit number 7.

7. Document and Archive: When you change something, log it! For each experiment you do, store your hardware and software configuration, performance level, and any collected data.

8. Use the right tool for the job: There are free data collection tools out there for various levels of the tuning process. System tuning tools include such as Performance Monitor for Windows or Sar for Linux. Application-level tools include Intel ® VTune^TM for both Windows and Linux.

9. Don't break the law: Amdahl's Law, that is. Amdahl's Law tells us the maximum amount of performance improvement we will get from a particular enhancement. Amdahl can help you set your expectations properly and clue you in to when you should be suspicious.

10. Compare apples to apples: Todd Christ reminds us of this habit in the last paragraph of this post. Don't compare the performance of mis-matched systems. If you must do it, know exactly what the differences are: the processor, memory type/speed/vendor, a software component, chipset, etc. Dig into the configuration details!

So now you have the high-level list! Stay tuned to The Server Room for more information about each habit in the coming weeks.

Ok, so you have bought into the potential benefits of simultaneous workflows from my previous blogs where organizations may choose work differently and develop parallel workflows in order to potentially:

• compress the time between an idea and a product or,
• innovate faster and deliver finished animations in less time
• or simply work on larger more complex assemblies or analyses

Ok so now it's time to say I doubt the performance is really available to get this done.

The honest answer is it depends on how much you want to stress the system, because there are limits. But Russ Sagert, Petrel Portfolio Manager from Schlumberger Information Systems suggests "workstations based on the new Intel® Xeon® 5400 series processors change the game for our customers. These new workstations featuring faster processors, larger memories, two new advanced PCI Gen2 graphics adapters give users a remarkably interactive experience and an opportunity to employ parallel geophysical workflows that combine complex large scale attribute analysis and volume visualization. Net result: Higher quality results and, better more informed decisions in less time than ever before."

Alright that is mouthful, but today's Intel® Xeon® based workstations do present the performance necessary to potentially change your game and help you innovate faster - are you game?

So what is the secret behind the performance?

The "Core" is. Not 4 cores, not 2 cores, or 8 even cores -- just Intel's new Core microarchitecture. Introduced in 2006, and extended with the availability of the Intel® Xeon® 5400 series, Intel's new Intel's new Core microarchitecture is the foundation that enables demanding users with the processing capacity and visualization capability to change the way they work and use simultaneous workflows to continuously iterate through design ideas, rapidly test via simulation, and find design errors earlier. The Intel Core Microarchitecture helps deliver the power to innovate.

Here is a quick look at some of the Intel innovations that are changing the performance game - we will discuss them more in future blogs:

• Intel® Wide Dynamic Execution - Increases performance and energy efficiency by enabling the delivery of more instructions per clock cycle and enhancing the efficiency of instruction execution.
• Intel® Advanced Digital Media Boost - Effectively doubles the instructions' execution speed compared to previous Intel based solutions and accelerates a broad range of applications, including financial, engineering, scientific video, speech and image, photo processing, and encryption software.
• Intel® Advanced Smart Cache, which is optimized for multi-core processing and reduces latency to frequently used data
• Intel® Smart Memory Access optimizes the use of the available data bandwidth from the memory subsystem and
• Intel® Intelligent Power Capability improves energy efficient performance and reduces noise and energy consumption.
• Macro Fusion combines instructions into a single instruction for execution, delivering more efficient processing and faster throughput.

Net result is Intel's new Core Microarchitecture can help you hit a high note in performance and an even higher note in innovation when you move to simultaneous workflows.

Server Management: How much is good enough?

Jackson He

Digital Enterprise Group, Intel Corporation

Manageability is a hot topic for IT managers. There have been so many solutions out there from different software vendors and hardware vendors alike. It seems to be "the more, the better". Is that true for all cases? Not really. IT managers also realized that more management also means increase of complexity and cost. It begs the question "How much server management is good enough?", "What are the basic needs for server management?", "Is there a way to balance the paradox between more complex management and simpler/cheaper options?"

We have done many researches and visited datacenters around the world to understand top server management challenges. Some of our observations are as the following:

• Datacenter management too complex & too expensive: Typically a datacenter has several separate management systems for server platforms, for network, and for security infrastructure. For server management alone, there is a dedicated management module (BMC - baseboard management controller) on each server and management console software to centrally manage the hundreds and thousands of servers in the datacenter. Each BMC adds extra cost to each server, the management software also cost a lot to build or buy, and deploy. The datacenter management cost and utility expense (power and cooling) can be 4~5 times more than the acquisition cost of the server hardware through the life of the server (~4 years).
• Server management implementations inconsistent among platform providers: To make things worse, server management is not uniform across servers from different OEMs. That is servers from HP and Dell have different BMC and therefore, need different management interfaces and extended software to manage - added time and cost for software development. This aggravates the management cost pressure and limits the choices of platforms from different OEMs. As a result, server management becomes even more complex and expensive in the name of more powerful management, a vicious cycle that gets worse as the datacenter grows.
• Packaged server management solutions not quite fit for IPDC environment: Packaged server management solutions from ISVs, like IBM Tivoli, HP Openview, CA, etc. are designed for enterprise where total number of servers is relatively low, applications are diverse, and dedicated IT engineers are looking for turnkey solutions. For the case of IPDC (internet portal datacenter), they have fewer specialized applications distributed across a large number of servers. Most of the IPDC datacenters develop their own server management software. However, they have to customize the management software for specific BMC for the servers of their choice. Furthermore, IPDC datacenters do not have as much the same level of server management granularity like the enterprise. They typically care a lot of system status and availability monitoring and means to remotely power on/off servers, but not as much remote diagnosis and repair as their enterprise counterparts. IPDC datacenters have more homogenous application and larger pool of servers. It will be more flexible and cost effective to simply redistribute workload and turn off troubled servers and fix them in bathes later. As a result, most of the IPDC customers write their own server management software that interact with a few basic management functions on the server - they really don't need a full-featured BMC.

Based on these finding, it makes us wonder if there are ways to make server management simpler. Is there an opportunity that we can achieve something that is *"less is better than more"*? Can we break the large complex server management solutions that trying to have "a size fits all" into something that is tiered with simpler/common server management at the bottom and more complex/customized server management layered on top of the basic server management?

Such a modular way of offering server management solutions that address some basic common needs across most datacenters. In such as way, we can standardize basic server management features and make them available to all Intel Architecture servers, while allowing OEMs and ISVs to build differentiations on top of them. It also gives the customers more flexibility to pick and chose based on their management needs. For IPDC customer, they may chose the basic set of server management that comes with servers regardless of which OEMs they come from - more vendors of choice, lower cost. For enterprise customers, they can chose more complex and customized solutions from particular OEMs and ISVs that best fit their datacenter management needs. Such as division of functions for server management may provide a relief of the ever increasing datacenter server management complexity and cost.

If this is true, where do we draw the line for basic server management? What is the minimum "good enough" server management feature set? I'd like to listen to you ideas around this as well. I will talk more about our thought around this in my next blog. Stay tuned.

Innovation can hit a wall - ever wonder why you really want a "real" workstation?

At a recent ISV event last week in Las Vegas, I met several end users who wished they had purchased a "real" workstation - what they had bought was a high-end desktop computer with a recommended high-end graphics card. What they got was disillusionment, sub-par performance and countless support nightmares. Workstations are designed and tested for performance, stability and expandability. The software application works and it delivers on performance because it is designed to.

To level set everyone, I view the workstation as the essential technology tool for professional creators that can quickly and efficiently transform complex data into actionable information. Key words transform data and actionable information. High quality decisions can be made without delay.

Workstations, for me, represent the single most important innovation engine used by professionals to generate end customer value. As such, they are the workhorse. They must be dependable, scalable and powerful enough to get out of the way of end user innovation.

So, what gets in the way of innovation? I have my top 3.

1) First is the dreaded hour glass or the innovation inhibitor as I like to call it. To me the hour glass is one of the most dangerous innovation inhibitors. It has no place in a workstation solution. If the hour glass appears, innovation can be quickly lost or sidetracked. Creative thought is disrupted and a good idea --- well it just got away while we wait for the workstation to respond to our request.

2) Data size. Until recently, data sizes have been limited and have forced workstation users to work with small sub sets of data. The result, users often miss seeing important trends that occur in larger assemblies and or models. These missed trends, while not halting innovation, certainly play a key role in extending the time for innovation to occur as smaller data sizes limit a more complete awareness of strategic differences.

3) Perhaps the most dangerous innovation inhibitor is the one that has the least to do with technology and the most to do with how companies work. New technologies found in today's workstations have transformed these tools in to powerful workstation supercomputers. This new breed is capable of delivering near supercomputer performance at an individual's desk enabling users to quickly iterate through ideas and potentially innovate faster than ever before.

There is an old poster I remember as a kid, I think it plays here as well. The saying went something like this --- I have been rich and I have poor; rich is better. Well the same is true in workstations - I have been fast and I have been slow - fast is better. More accurately workstations with certified applications and graphics cards can help deliver the performance, stability and scalability you need to innovate faster. They can deliver the technology that gets out of the way of your users capability to innovate. Enabling them to create value faster than ever before on Intel based workstations.

High-end desktops with powerful graphics may not really be all that you want or need. In fact high-end desktop computers, while marginally less expensive, may not be, in the end, what you can afford as you potentially experience less than adequate performance and below industry average innovation.

Which can you afford?