Blog Posts From Wired Ethernet Tagged With server

Making Room for Networking at the Open Compute Summit - 2013

webadmin@intel.com — Wed, 16 Jan 2013 18:56:55 GMT

On most days, I talk with people who are focused on the latest in open networking in the data center. But this week I’m seeing that data center openness means different things if you design servers.

That’s because I’m at Open Compute Summit with the Intel® SeaCliff Trail 10G/40G top-of-rack switch reference design – tucked in to a small corner of an large Intel booth filled with boards and server designs, including microservers. It feels like I’m one of the few networking folks in a sea of server experts.

The annual event is a Facebook-initiated effort to establish open hardware standards for data center servers. Some key initiatives include new open standards for virtual IO, storage, hardware management, data center racks and power supplies. There’s also an OCP standard for an Intel motherboard. Server virtualization has increased the efficiency of data center servers, and the Open Compute projects are designed to extend that efficiency to every element of the hardware design.

I think the Seacliff Trail will be a great fit at this show. Servers and switches are the two essential elements in the data center, so it makes sense for them to be together at OCP Summit. And Seacliff Trail is a very open switch design – most notably with support for software-defined networking.

Intel® Flexible Port Partitioning (FPP) Using SR-IOV Overview & Demonstration

webadmin@intel.com — Mon, 19 Sep 2011 17:15:04 GMT

So, were you one of the ones that were able to visit me at booth #905 at the Intel Developer Forum in San Francisco last week and watch the Intel Flexible Port Partitioning demonstration? Well, if you weren’t, or if you are looking to see the demonstration again, it’s your lucky day!

Intel Flexible Port Partitioning (FPP) is the ability to use SR-IOV Virtual Functions, which have up until now been thought of as a strictly virtualization technology, and use them in a bare-metal (or mixed) Open Source OS. This provides a way to very flexibly and efficiently carve up your Ethernet ports.

I was so taken aback with the overwhelmingly positive response to the demonstration and the two chalk talks, as well as the standard session Brian Johnson and I presented last week, that when I came home from San Francisco, I combined our session material with the demonstration and produced a “from-the-hip” video explaining the technology and showing a demonstration.

I have a 15 minute cap in my YouTube account; the video is 14 minutes and 59 seconds. Hope you will excuse the occasional ‘uhm’

I hope you find the demo as interesting as the folks that visited me in the booth did. Here is a sample of some of the comments I received:

“Wow – you guys are killing everybody else! Nobody else is doing this kind of thing!”

“Hey – I saw some cool video a while back with little Ethernet packets and guys moving all over the screen that explained SR-IOV very very well – do you know where I can find that?” – was referring to my YouTube video

“Holy cow, that’s cool! I’ve got to hook you guys up with our Ethernet architect so we can get even more support for that!”

“Oh my gosh! That is exactly what we need! You just made my day and will make my architects very ,very happy”

“Very nice demo, on a great OS. Really shows the power of FPP; we need to get a joint paper out.”

“Wow, that is pretty slick – I’m going to have to go talk to some folks when I get back to the office.”

The whitepaper I promised is in the final stages as well and should be published within the next couple of weeks–so keep checking back here for the announcement.

Not sure what SR-IOV is? Click on the link to see a video explanation. I also have the PCI-SIG SR-IOV Primer you might want to check out.

Enjoy,

- Patrick

Energy Efficient Ethernet: Technology, Application and Why You Should Care

webadmin@intel.com — Thu, 05 May 2011 18:52:24 GMT

Introduction

A friend who labels himself as “technologically adverse” recently took a big step and reluctantly upgraded from an “antique” flip-phone he had for 3 years to a new “high-tech” smartphone. He has joined the estimated 60 million other smartphone users in the US who can access their email, check stocks, listen to streaming music, watch streaming videos, etc. with the flick of a finger.

So what does this have to do with IEEE802.3az: Energy Efficient Ethernet or “EEE” (pronounced “triple-E”)? As the computing continuum of wireless and connected devices increases, consequently the datacenter must also scale to meet the increased data demand. The datacenter and its networking equipment is not the only target market; so is networked equipment. Consumer and enterprise computers, consumer electronics like set-top boxes, routers and other Ethernet applications are part of the ecosystem.

As the world becomes increasingly connected, there is the tradeoff of increased demand for power. In order to help curb this need, EEE is seen as an incremental step to reduce power consumption. A Lawrence Berkeley Labs researcher estimates that adoption of EEE in the networking and networked device markets would result in energy savings of $400 million or more per year in the US (Merritt, 2008).

What is EEE?

Energy Efficient Ethernet allows for an Ethernet device operating over twisted-pair cabling (100BASE-TX, 1000BASE-T and 10GBASE-T) or electrical backplanes (1000BASE-KX, XAUI, 10GBASE-KX4 or 10GBASE-KR) to dynamically scale power consumption based on traffic load. This ability is the value proposition of 802.3az and is accomplished by a communication protocol called Low Power Idle (LPI) which is negotiated between a remote (server) and local (client) PHY at power up. LPI with be further explained in the next section.

Once LPI has been negotiated, the local PHY is allowed to enter a link state referred to as simply “LPI”. Conversely a non-EEE compliant PHY will remain in an idle state and, depending on the link speed, will consume up to 10 times more power than an EEE compliant PHY. But there is one caveat. In order to utilize EEE, the local and remote PHYs must both be EEE compliant.

The Intel® 82579 Gigabit Ethernet PHY is an 802.3az compliant part with the following power characteristics:

System State	Link State	Device Power (mW)
S0	1000Mbps Active	619
	1000Mbps Idle	509
	1000Mbps LPI	49
	100Mbps Active	315
	100Mbps Idle	207
	100Mbps LPI	53

Figure 1: 82579 Power Measurements

How does it Work?

The power savings from a purely percentage basis is significant, but how does EEE work? Let’s start off with the example of traffic flowing between a remote (server) and local (client) PHYs with network and switching equipment between them. As stated before, both devices will advertise and handshake the capability to support 802.3az at power up. Once the handshake is completed, the remote PHY knows it can initiate a LPI signature (figure 2) to the local PHY.

Figure 2: LPI Signature

From figure 2, the LPI sleep and wake signals are initiated by the remote PHY between periods of activity. Once the local PHY is asleep, periodic refreshes are sent to the client in order to update adaptive filters and timing circuits in order to maintain link integrity. It is during the low-power state where power savings are realized at the local PHY because it is allowed to switch off part of the receive and transmit circuitry. Once there is data to be transmitted, the remote PHY sends a wake signal and data packets are then transmitted.

AreThere Any Drawbacks?

Performance using EEE with respect to latency is impacted. The total amount of latency for a sleep plus a wake is about 7.36µs. By comparison, at 10Gbps it takes 1.2µs to send a 1500 byte frame. The latency is created by the overhead of sleep and wake transitions does not make this ideal for certain applications such as financial service servers.

Another drawback is the inability for EEE compliant PHYs to utilize power saving technology when linked with non-EEE compliant PHY. First generation silicon that supports EEE was released in 2010 and will be fully functional with non-EEE compliant silicon, but in order for LPI to be utilized there must be EEE compliant infrastructure to support the technology.

Summary

Smart energy use has become a focal point of customer sentiment, international laws and regulations, and corporate environmental stewardship and will continue to be a focal point of continuous improvement. Energy Efficient Ethernet is another step in allowing Ethernet connected devices to be more energy efficient. This technology may not be appropriate for every application, but there is a clear economic, social and environmental benefit.

Bibliography

Merritt, R. (2008, 5 8). Energy-efficient Ethernet standard gains traction. Retrieved 4 4, 2011, from EE Times: http://www.eetimes.com

Note about Edit on 5/9/2011: Earlier I fixed a typo by adding an "e" changing "servic" to "service". I neglected to add a note about the change I made. I am adding this not now so no one has to spend a lot of time trying to figure out what changed.

Mark H

Latest Intel Ethernet Virtualization Paper for Your Reading Pleasure!!

webadmin@intel.com — Wed, 08 Dec 2010 23:56:57 GMT

My partner-in-crime, Brian Johnson, and I have just finished a new White Paper detailing some best practices when using 10 Gigabit Ethernet with VMware* ESX/ESXi. This is the 3^rd paper in a series that began with Brian’s well-received and many times re-branded paper entitled Simplify VMware vSphere* 4 Networking with Intel® Ethernet 10 Gigabit Server Adapters that I blogged about here in July. This paper detailed some best practices regarding how to improve efficiency and simplicity by moving to two 10 Gigabit Ethernet connections on a virtualized server rather than multiple 1 Gigabit ports.

We continued the series with another White Paper entitled Virtual Switches Demand Rethinking Connectivity for Servers, which discusses how the virtual switch within VMM can now be thought of as having uplinks from your Virtualized Server in a VM to the virtual switch, which in turn has physical uplinks to the top-of-rack switch; this paper is located here.

This latest paper modifies some of the best practices discussed in the 1^st paper to utilize some of the new Quality of Service capabilities in the Virtual Distributed Switch in VMware vSphere 4.1. We discuss how and why those practices have been updated and provide some additional best practices that can be applied to a virtualized environment.

We hope you find the latest paper useful and encourage feedback. The white paper is located here.