Gone in 98 Nanoseconds

scottcschweitzer HFT, networking, TCP Offload Engine

Imagine a daily race with hundreds of top fuel dragsters all lined up rumbling along in parallel waiting for the same green Christmas tree light before launching off the line. In some electronic markets, with specific products, every weekday morning this is exactly what happens. It’s a race where being the fastest is the primary attribute used to determine if you’re going to be doing business. On any given day only the top finishers are rewarded with trades. Those who transmit their first orders of the day the fastest receive a favorable position at the head of the queue and are likely to do some business that day. In this market, EVERY nanosecond (a billionth of a second) of delay matters, and can be monetized. Last week the new benchmark was set at 98 nanoseconds, plus your trading algorithm, in some cases 150 nanoseconds total tick to trade.

“Latency” is the industry term for the unavoidable network delays, and “Tick to Trade Latency” aggregates together the network travel time for a UDP market data signal to arrive at a trading system, and for that trading system to transmit a TCP order into the exchange. Last year Solarflare introduced Application Nanosecond TCP Send (ANTS) and lowered the “Tick to Trade Latency” bar to 350 Nanoseconds. ANTS executes in collaboration with Solarflare’s Application Onload Engine (AOE) based on an Altera Stratix FPGA. Solarflare further advanced this high-speed trading platform to achieve 250 Nanoseconds. Then in the spring of 2017 Solarflare collaborated with LDA Technologies. LDA brought their Lightspeed TCP cores to the table and replaced the AOE with a Xilinx FPGA board once again lowering the “Tick to Trade Latency” to 120 Nanoseconds. Now through further advances, and moving to the latest Penguin Computing Skylake computing platform, all three partners just announced a STAC-T0 qualified benchmark of 98 nanoseconds “Tick to Trade Latency!”

There was even a unique case in this STAC-T0 testing where the latency was measured at negative 68 nanoseconds, meaning that a trade could be injected into the exchange before the market data from the exchange had even been completely received. Compared to traditional trading systems which require that the whole market data network packet to be received before ANY processing can be done, these advanced FPGA systems receive the market data in the packet in four-byte chunks and can begin processing that data while it is arriving. Imagine showing up in the kitchen before you wife even finishes calling your name for dinner. There could be both good and bad side effects of such rapid action, you have a moment or two to taste a few things before the table is set, or you may get some last minute chores. The same holds true for such aggressive trading.

Last week, in a Podcast with the same name we had a discussion with Vahan Sardaryan, CEO of LDA Technologies, where we went into this in more detail.

Penguin Computing is also productizing the complete platform, including Solarflare’s ANTS technology and NIC, LDA Technologies Lightspeed TCP, along with a high-performance Xilinx FPGA to provide the Ultimate Trading Machine.

The Ultimate Trading Machine

Security Entirely Chimerical, SEC

scottcschweitzer HFT, Security

On September 20th SEC Chairman Jay Claton released a “Statement on Cybersecurity.” It is an extremely dry read, but if one suffers through it they’ll find several interesting points.

“I recognize that even the most diligent cybersecurity efforts will not address all cyber risks that enterprises face. That stark reality makes adequate disclosure no less important.”

How does the SEC define “adequate disclosure?” The federal government has requirements that in some extreme breach cases require a report within one hour to DHS’s CERT. When faced with this class of breach recently it was found that the SEC waited 14 days, is this adequate disclosure? Much further down in the SEC Statement they disclosed the following.

“In August 2017, the Commission learned that an incident previously detected in 2016 may have provided the basis for illicit gain through trading. Specifically, a software vulnerability in the test filing component of our EDGAR system, which was patched promptly after discovery, was exploited and resulted in access to nonpublic information.”

So in the best case, the SEC waited only eight months to inform the public of this breach, but it could have been as much as 20 months. Unlike the publicly traded companies, the SEC regulates it isn’t legally required to tell investors or the public if it is ever breached. It is ONLY required to inform a law enforcement agency. EDGAR was also breached in 2014, but that saw little attention.

Now it’s one thing to breach an entity and remove data, but how about intentionally leaving false data behind for the purpose of capitalizing on that deposit? In at least two cases over the past few years, false business acquisition reports for Avon and the Rocky Mountain Chocolate Factory have been inserted into EDGAR. In the Avon case, the stock ran up 10 points. Does the SEC own up to these, well kinda of…

“As another example, our Division of Enforcement has investigated and filed cases against individuals who we allege placed fake SEC filings on our EDGAR system in an effort to profit from the resulting market movements.”

Ok, so EDGAR is a 30-year-old piece of swiss cheese riddled with potential attack surfaces some by design, others by just not keeping current on penetration testing of their systems. What about their physical assets?

“For example, a 2014 internal review by the SEC’s Office of Inspector General (“OIG”), an independent office within the agency, found that certain SEC laptops that may have contained nonpublic information could not be located.”

All the above quotes were from the Wednesday SEC Statement, but in a 2016 GAO report on the SEC, it stated that the SEC:

“…wasn’t always using encryption, supported software, well-tuned firewalls, and other key security tools while going about its business.”

Banking, in fact, our financial market structure as a whole is based on a singular concept, TRUST. The SEC was created in the wake of the Great Depression in 1934 as a way to restore trust in the markets. Technology savvy individuals will always attempt to exploit this trust for their own gain, it’s a part of how the game is played. In our financial system, the SEC plays the role of the gambling commission to ensure that the players, dealers, pit bosses, and the house are all working from the same set of published public rules. To his credit Chairman Clayton is working within the system in an attempt to shine daylight on an agency in trouble and out of touch with the technology driving the markets its charged with regulating. Today it is now possible to trade a stock based on a tick (a signal that something moved) within 150 billionths of a second, but it takes the SEC 1.2 million seconds (14 days) to report a serious breach of their security to law enforcement. Clearly, work remains to be done.

Equifax & Micro Segmentation

scottcschweitzer Containers, Microsegmentation, networking, Security, Website

Earlier this week it was reported that an Equifax web service was hacked creating a breach that existed for about 10 weeks. During that time the attackers used that breach to drain 143 million people’s private information. The precise technical details of the breach, which Equifax claims was detected and closed on July 29, has yet to be revealed. While it says it’s seen no other criminal activity on its main services since July 29th that’s of little concern as Elvis has left the building. At 143 million that means a majority of the adults in the US have been compromised. Outside of Equifax specific code vulnerabilities or further database hardening what could Equifax have done to thwart these attackers?

Most detection and preventative countermeasures that could have minimized Equifax’s exposure employ some variation of behavior detection at one network layer. They then shunt suspect traffic to a sideband queue for further detailed human analysis. Today the marketing trend to attract Venture Capital investment is to call these behavior detection algorithms Artificial Intelligence or Machine Learning. How intelligent they are, and to what degree they learn is something for a future blog post. While at the NGINX Conference this week we saw several companies selling NGINX layer-7 (application layer) plugins which analyzed traffic prior to passing it to NGINX’s HTML code evaluation engine. These plugins receive the entire HTML request after the OS stack has assembled it from multiple network packets. They then do a rapid analysis the request to determine if it poses a threat. If not then the request is passed back to NGINX for the web application to respond to. Along the way, the plugin abstracts metadata from the request and in parallel, it shoots that up to their cloud service for further evaluation. This metadata is then compared against prior history and other real-time customer data from with similar services to extract new potential threat vectors. As they are detected rules are then pushed back down into the plugin that can be applied to future packets.

Everything discussed above is layer-7, the application layer, traffic analysis, and mitigation. What does layer-7 have to do with network micro-segmentation? Nothing, what’s mentioned above is the current prevailing wisdom instantiated in several solutions that are all the rage today. There are several problems with a layer-7 solution. First, it competes with your web application for host CPU cycles. Second, if the traffic is determined to be malicious you’ve already invested tens of thousands of CPU instructions, perhaps even in excess of one hundred thousand instructions to make this determination, all that computer time is lost once the message is dropped. Third, the attack is now deep inside your web server and whose to say the attacker hasn’t learned what he needed to move to a lower layer attack vector to evade detection. Layer-7 while convenient, easy to use, and even easier to understand is very inefficient.

So what is network micro-segmentation, and how does it fit in? Network segmentation is the act of altering the flow of traffic such that only what you want is permitted to pass. Imagine the factory that makes M&Ms. These days they use high-speed cameras and other analytics that look for deformed M&Ms and when they see one they steer it away from the packaging system. They are in fact segmenting the flow of M&Ms to ensure that only perfect candy-coated pieces ever make it into our mouths. The same is true for network traffic, segmentation is the process of only allowing network packets to flow into or out of a given device via a specific policy or set of policies. Micro-segmentation is doing that down to the application level. At Layer-3, the network layer, that means separating traffic by the source and destination network address and port, while also taking into account the protocol (this is known as “the five-tuple”, a set of five elements). When we focus on filtering traffic by network port we can say that we are doing application level filtering because ports are used to map network traffic to applications. When we also take into account the local IP address for filtering then we can also say we filter by the local container (ex. Docker) or Virtual Machine (VM) as these can often get their own local IP address. Both of these items together can really define a very specific network micro-segmentation strategy.

So now imagine a firewall inside a smart network interface card (NIC) that can filter both inbound and outbound packets using this network micro-segmentation. This is at layer-3, the Network, micro-segmentation within the smart NIC. When detection is moved into the NIC no x86 CPU cycles are consumed when evaluating the traffic, and no host resources are lost if the packet is deemed malicious and is dropped. Furthermore, if it is a malicious packet and it’s stopped by a firewall in the NIC then the threat has never entered the host CPU complex, and as such, the system’s integrity is preserved. Consider how this can improve an enterprise’s security as it scales out both with new servers, as well as adding containers and VMs. So how can this be done?

Solarflare has been shipping its 8000 line of smart NICs since June of 2016, and later this fall they will release a new firmware called ServerLock(TM). ServerLock is a first generation firewall in the smart NIC that is centrally managed. Every second it sends a summary of network flows through the NIC, in both directions, to a central ServerLock Manager system. This system then allows administrators to view these network flows graphically and easily turn them into security roles and policies that can then be deployed. Policies can then be deployed to a specific local IP address, a collection of addresses (think Docker containers or VMs) called an “IP Set”, a host or host groups. When deployed policies can be placed in Monitor or Enforce mode. Monitor Mode will allow all traffic to flow, but it will generate alerts for all traffic outside of all the defined policies for a local IP address. In Enforce mode, ONLY traffic conforming to the defined policies will be permitted. Traffic outside of those policies will generate an alert and be dropped. Once a network device begins to drop traffic on purpose we say that that device is segmenting the network. So in Enforce mode, ServerLock smart NICs will actively segment that server’s network by only passing traffic for supported applications, only those for which a policy exists. This applies to traffic in both directions, so for example, if an administrator walks into the data center, grabs a keyboard and elects to Secure Copy (SCP) a file from a database server to his workstation things will get interesting. If the ServerLock smart NIC in that database server doesn’t have a policy supporting SCP (port 22) his outbound request from that database server to his workstation will be dropped in the NIC. Likely unknown to him an alert will be generated on the central ServerLock Manager console calling out the application and both the database server and his workstation, and he’ll have some explaining to do.

ServerLock begins shipping this fall so while it’s too late for Equifax it’s not too late for the next Equifax. So how would this help moving forward? Simple, if every server, including web servers and database servers, has a ServerLock smart NIC then every second these servers would report their flow data to the central Solarflare ServerLock Manager for further analysis. Solarflare is working with Cloudwick to do real-time analysis of this layer-3 traffic so that Cloudwick can then proactively suggest in real time back to ServerLock administrators new roles and policies to proactively protect servers against all sorts of threats. More to come as this product is released.

9/11/17 Update – It was released over the weekend that Equifax is now pointing the blame at an Apache Struts module. The exact module has yet to be disclosed, but it could be any one of the following that has been previously addressed. On Saturday The Apache group replied pointing to other sources that believe it might have been caused by exploiting a remote code execution bug in their REST plugin as outlined in CVE-2017-9805. More to come.

9/12/17 Update – Alert Logic has the best analysis thus far.

Get Three Times More From NGINX

scottcschweitzer networking

Recently Solarflare re-ran some tests with Nginx that measured the amount of traffic it could respond to before it started dropping requests. We then scaled up the number of cores provided to Nginx to see how additional compute resources impacted the servicing of web page requests, and this was the resulting graph:

click for larger image

As you can see from the above graph most NIC implementations require about six cores to achieve 80% wire-rate. The major difference highlighted in this graph though is that with a Solarflare adapter, and their OpenOnload OS Bypass driver (also known as UKB, Universal Kernel Bypass) they can achieve 90% wire-rate performance utilizing ONLY two cores versus six. Note that this is with Intel’s most current 10G NIC the x710.

What’s interesting here though is that OpenOnload internally can bond together up to six 10G links, before a configuration file change is required to support more.  This could mean that a single 12 core server, running a single Nginx instance should be able to adequately service 90% wire-rate across all six 10G links, or theoretically 54Gbps of web page traffic. Now, of course, this is assuming everything is in memory, and the rest of the system is properly tuned. Viewed another way this is 4.5Gbps/core of web traffic serviced by Nginx running with OpenOnload on a Solarflare adapter compared to 1.4Gbps/core of web traffic with an Intel 10G NIC. This is a 3X gain in performance for Solarflare over Intel, how is the possible?

Simple, OpenOnload is a userspace stack that communicates directly with the network adapter in the most efficient manner possible to service UDP & TCP requests. The latest version of OpenOnload has also been tuned to address the C10K problem. What’s important to note, is that by bypassing the Linux OS to service these communication requests Solarflare reduces the number of kernel context switches/core, memory copies, and can more effectively utilize the processor cache. All of this translates to more available cycles for Nginx on each and every core.

To further drive this point home we did an additional test just showing the performance gains OOL delivered to Nginx on 40GbE. Here you can see that the OS limits Nginx on a 10-core system to servicing about 15Gbps. With the addition of just OpenOnload to Nginx, that number jumps to 45Gbps. Again another 3X gain in performance.

If you have web servers today running Nginx, and you want to give them a gargantuan boost in performance please consider Solarflare and their OpenOnload technology. Imagine taking your existing web server today which has been running on a single Intel x520 dual port 10G card, replacing that with a Solarflare SFN7122F card, installing their OpenOnload drivers and seeing a 3X boost in performance. This is a fantastic way to breathe new life into existing installed web servers. Please consider contacting Solarflare today do a 10G OpenOnload proof of concept so you can see these performance gains for yourself first hand.

R.I.P. TCP Offload Engine NICs (TOEs)

scottcschweitzer HFT, networking, RDMA

Solarflare Delivers Smart NICs for the Masses: Software Definable,  Ultra-Scalable, Full Network Telemetry with Built-in Firewall for True Application Segmentation, Standard Ethernet TCP/UDP Compliant​

As this blog post by Michael C. Bazarewsky states, Microsoft quietly pulled support for TCP Chimney in its Windows 10 operating system. Chimney was an architecture for offloading the state and responsibility of a TCP connection to a NIC that supported it. The piece cited numerous technical issues and lack of adoption, and Michael’s analysis hits the nail on the head. Goodbye TOE NICs.

During the early years of this millennium, Silicon Valley venture capitalists dumped hundreds of millions of dollars into start-ups that would deliver the next generation of network interface cards at 10Gb/sec using TCP offload engines. Many of these companies failed under their weight of trying to develop expensive, complicated silicon that just did not work. Others received a big surprise in 2005 when Microsoft settled with Alacritech over patents they held describing Microsoft’s Chimney architecture. In a cross-license arrangement with Microsoft and Broadcom, Alacritech received many tens of millions of dollars in licensing fees. Alacritech would later get tens of millions of more fees from nearly every other NIC vendor implementing a TOE in their design. At the time, Broadcom was desperate to pave the way for their acquisition of Israeli based Siloquent. Due to server OEM pressure, the settlement was a small price to pay for the certain business Broadcom would garner from sales of the Siloquent device. At 1Gb/sec, Broadcom owned an astounding 100% of the server LAN-on-Motherboard (LOM) market, and yet their position was threatened by the onslaught of new, well-funded 10Gb start-ups.

In fact, the feature list for new “Ethernet” enhancements got so full of great ideas that most vendor’s designs relied on a complex “sea of cores” promising extreme flexibility that ultimately proved to be very difficult to qualify at the server OEMs. Any minor change to one code set would cause the entire design to fail in ways that were extremely difficult to debug, not to mention being miserably poor in performance. Most notably, Netxen, another 10Gb TOE NIC vendor, quickly failed after winning major design-ins at the three big OEMs, ultimately ending in a fire sale to Qlogic. Emulex saw the same pot of gold in its acquisition of ServerEngines.

That new impetus was a move by Cisco to introduce Fibre Channel Over Ethernet (FCoE) as a standard to converge networking and storage traffic. Cisco let Qlogic and Emulex (Q & E) inside the tent before their Unified Computing System (UCS) server introduction. But the setup took some time. It required a new set of Ethernet standards, now more commonly known as Data Center Bridging (DCB). DCB was a set of physical layer requirements that attempted to emulate the reliability of TCP by injecting wire protocols that would allow “lossless” transmission of packets. What a break for Q & E! Given the duopoly’s control over the Fibre Channel market, this would surely put both companies in the pole position to take over the Ethernet NIC market. Even Broadcom spent untold millions to develop a Fiber Channel driver that would run on their NIC.

Q & E quickly released what many called the “Frankenstein NIC,” a kluge of Applied-Specified Integrated Circuits (ASIC) designed to get a product to market even while struggling to develop a single ASIC, a skill at which neither company excelled. Barely achieving its targeted functionality, no design saw much traction. Through all of our customer interactions (over 1,650), we could find only one that had implemented FCoE. This large bank has since retracted its support for FCoE and in fact, showed a presentation slide several years ago stating they were “moving from FCoE to Ethernet,” an acknowledgment that FCoE was indeed NOT Ethernet.

In conjunction with TOEs, the industry pundits believed that RDMA (Remote Direct Memory Access) was another required feature to reduce latency, and not just for High-Frequency Trading (HFT), another acknowledgment that lowering latency was critical to the hyper-scale cloud, big data, and storage architectures. However, once again, while intellectually stimulating, using RDMA in any environment proved to be complex and simply not compatible with customers’ applications or existing infrastructures.

The latest RDMA push is to position it as the underlying fabric for Non-Volatile Memory Express (NVMeF). Why? Flash has already reduced the latency of storage access by an order of magnitude, and the next generation of flash devices will reduce latency and increase capacity even further. Whenever there’s a step function in the performance of a particular block of computer architecture, developers come up with new ways to use that capability to drive efficiencies and introduce new, and more interesting applications. Much like Moore’s Law, rotating magnetic memory is on its last legs. Several of our most significant customers have already stopped buying rotating memory in favor of Flash SSDs.

Well… here we go again. RDMA is NOT Ethernet. Despite the “fake news” about running RDMA, RoCE and iWARP on Ethernet, the largest cloud companies, and our large financial services customers have declared that they cannot and will not implement NVMeF using RDMA. It just doesn’t fit in their infrastructures or applications. They want low-latency standard Ethernet.

Since our company’s beginning, we’ve never implemented TOEs, RDMA or FCoE or any of the other great and technically sound ideas for changing Ethernet. Sticking to our guns, we decided to go directly to the market and create the pull for our products. The first market to embrace our approach was High-Frequency Trading (HFT). Over 99% of the world’s volume of Electronic trading, in all instruments, runs on our company’s NICs. Why? Customers could test and run our NICs without any application modifications or changes to their infrastructure and realize enormous benefits in latency, Jitter, message rate and robustness… it’s standard Ethernet, and our kernel bypass software has become the industry’s default standard.

It’s not that there isn’t room for innovation in server networking, it’s that you have to consider the customer’s ability to adapt and manage that change in a way that’s not inconsistent or disruptive to their infrastructure, while at the same time, delivering highly valued capabilities.

  • If companies are looking for innovation in server networking, they need to look for a company that can provide the following: Best-in-class PTP synchronization
  • Ultra-high resolution time stamps for every packet at every line rate
  • A method for lossless, unobtrusive, packet capture and analysis
  • Significant performance improvement in NGINX and LXC Containers
  • A firewall NIC and Application Micro-Segmentation that can control every app, VM, or container with unique security profiles
  • Real, extensive Software Definable Networking (SDN) without agents

In summary, while it’s taken a long time for the industry to realize its inertia, logic eventually prevailed.  Today, companies can now benefit from innovations in silicon and software architecture that are in deployment and have been validated by the market.   Innovative approaches such as neural-scale networking, which is designed to respond to the high-bandwidth, ultra-low-latency, hardware-based security, telemetry, and massive connectivity needs of ultra-scale computing, is likely the only strategy to achieve a next-generation cloud and data center architecture that can scale, be easily managed, and maybe most importantly secured.

— Russell Stern, CEO Solarflare

Cloaked Data Lakes

scottcschweitzer Microsegmentation, networking, Security

Once Jessie James was asked why he robbed banks and answered: “Because that’s where the money is?” Today a corporation’s most valuable asset, aside from its people, is its data. For those folks who are Star Trek fans imagine if you could engage your data lake’s network cloaking device just before deployment? It would waver out of view then totally disappear from your enterprise network to all but those who are responsible for extracting value from it. Your key data scientists and applications could still see and interact with your cloaked data lakes, but to others exploring and scanning the network, it would be entirely transparent as if it were not even there.

Imagine if you will that a Klingon Bird of Prey is cloaked and patrolling the Neutral Zone. Along comes the Federation Starship Enterprise, also patrolling the Neutral Zone, but the Federation is actively scanning the quadrant. Since the Klingon ship is Cloaked the Federation can’t detect them, but the moment the Enterprises scanners pass over the Bird of Prey it automatically jumps to red alert, energizes its weapons systems and alters course to shadow the Federation ship. Imagine if the same could be true of an insider threat or an internal breach via say a phishing attack that is seeking out your companies data. The moment someone pings a system or executes a port scan of even one IP addresses of the servers within your data lake alarm bells are set off, and no reply is returned. The scanner would see no answer, and expect that nothing exists, little would they know the hell that would soon reign down on them.

Your network administrators would then be alerted that their new server orchestration system had raised an alert. They’ll quickly see that the attacker is another admin’s workstation, someone that has been suspected of being an insider threat, but they’ve been too cagey to nail down. Now it’s 9 PM at night, and he’s port scanning the exact range of internal network addresses that were set aside a week earlier for this new data lake. He then moves on to softer targets exfiltrating data from older systems. Little does he know though that every server he’s touched the past week has been tracking and reporting every network flow back to his workstation. Management was just waiting for the perfect piece of evidence and this attempted port scan, along with all the other network flows was the final straw.

His plan had been to finish out the week, then quit on Friday and sell all his companies data to its competitors. He had decided to stay on an extra two weeks when he heard they were standing up a new Hadoop cluster. He figured that it would make a juicy soft target with tons of the newest aggregated data which could be enormously valuable. What he didn’t know, because he wasn’t invited to those planning meetings, was that the cluster included a new stealth security feature from Solarflare called Active Cloaking. He also wasn’t aware that that feature was the driving reason why many of his companies servers over the past two weeks had been upgraded to new Solarflare 10GbE NICs with ServerLock.

Since he was a server administrator responsible for some of the older legacy systems he wasn’t involved in the latest round of network upgrades. While he had noticed that lately some of the newer servers were no longer accessible to him via SSH, what he wasn’t aware of was that every server he touched was now reporting his every move. What would prove even more damning though was that some of those older servers, which had been upgraded with Solarflare ServerLock enabled NICs, were left as internal SSH/SCP honeypots with old legacy data that held little if any real value, but would prove damning evidence once compromised. Tonight had proved to be his downfall, his manager, and his VP, along with building security had just entered his cubical and stated that the police were on their way.

At Black Hat last month both Solarflare and Cloudwick (CDL) demonstrated ServerLock and data lake cloaking. In September several huge enterprises will begin testing SeverLock, and if you’re an insider threat consider yourself warned!

1st Ever Firewall in a NIC

scottcschweitzer Microsegmentation, networking, Security

Last week at Black Hat Solarflare issued a press release debuting their SolarSecure solution, which places a firewall directly in your server’s Network Interface Card (NIC). This NIC based firewall called ServerLock, not only provides security, but it also offers agentless server-based network visibility. This visibility enables you to see all the applications running on every ServerLock enabled server. You can then quickly and easily develop security policies which can be used for compliance or enforcement. During the Black Hat show setup, we took a 10-minute break to have an on-camera interview with Security Guy Radio that covered some of the key aspects of SolarSecure.

SolarSecure has several very unique features not found in any other solution:

  • Security and visibility are entirely handled by the NIC hardware and firmware, there are NO server side software agents, and as such, the solution is entirely OS independent.
  • Once the NIC is bound to the centralized manager it begins reporting traffic flows to the manager which then represents those graphically for the admins to easily turn into security policies. Policies can be created for specific applications, enabling application level network segmentation.
  • Every NIC maintains separate firewall tables for each local IP address hosted on the NIC to avoid potential conflicts from multiple VMs or Containers sharing the same NIC.
  • Each NIC is capable of handling over 5,000 filter table rules along with another 1,000 packet counters that can be attached to rules.
  • Packets transit the rules engine between 50 and 250 nanoseconds so the latency hit is negligible.
  • The NIC filters both inbound and outbound packets. Packets which are dropped as a result of a match to a firewall rule generate an alert on the management console and inbound packets consume ZERO host CPU cycles.

Here is a brief animated explainer video which was produced prior to the show that sets up the problem and explains Solarflare’s solution. We also produced a one-minute demonstration of the management application and its capabilities.

Storage Over TCP in a Flash

scottcschweitzer networking, NVMe

By Patrick Dehkordi

Recently Solarflare delivered a TCP transport for Non-Volatile Memory Express (NVMe). The big deal with NVMe is that it’s FLASH memory based, and often multi-ported so when these “disk blocks” are transferred over the network, even with TCP, they often arrive 100 times faster than they would if they were coming off spinning media. We’re talking 100 microseconds versus average 15K RPM disk seek times measured in milliseconds. Unlike RoCE or iWARP a TCP transport provides storage over Ethernet without requiring ANY network infrastructure changes.

It should be noted that this should work for ANY NIC and does not require RDMA, RoCE, iWARP or any special NIC offload technology. Furthermore, since this is generic TCP/IP over Ethernet you don’t need to touch your switches to setup Data Center Bridging. Also, you don’t need Data Center Ethernet, Converged Ethernet, or Converged Enhanced Ethernet, just plain old Ethernet.  Nor do you need to set things up to use Pause Frames or Priority Flow Control. This is industry changing stuff, and yet not hard to implement for testing so I’ve included a recipe for how to make this function in your lab below, it is also cross-posted in the GitHub.

At present this is a fork of the v4.11 kernel. This adds two new kconfig options:

  • NVME_TCP : enable initiator support
  • NVME_TARGET_TCP : enable target support

The target requires the nvmet module to be loaded. Configuration is identical to RDMA, except "tcp" should be used for addr_trtype.

The host requires the nvme, nvme_core and nvme_fabrics modules to be loaded. Again, the configuration is identical to RDMA, except -t tcp should be passed to the nvme command line utility instead of -t rdma. This requires a patched version of nvme-cli.

Example assumptions

This is assuming a target IP of 10.0.0.1, a subsytem name of ramdisk and an underlying block device /dev/ram0. This is further assuming an existing system with RedHat/Centos Distribution built on 3.x kernel.

Building the Linux kernel

For more info refer to https://kernelnewbies.org/KernelBuild

Install or confirm the following packages are installed

yum install gcc make git ctags ncurses-devel openssl-devel

Download, unzip or clone the repo into a local directory

git clone https://github.com/solarflarecommunications/nvme-of-tcp/tree/v4.11-nvme-of-tcp
cd nvme-of-tcp-4.11

Create a .config file or copy the existing .config file into the build directory

scp /boot/config-3.10.0-327.el7.x86_64 .config

Modify the .config to include the relevant NVMe modules

make menuconfig

Under “Block Devices” at a minimum select “NVM Express block device” “NVMe over Fabrics TCP host support” “NVMe over Fabrics TCP target support” Save and Exit the text based kernel configuration utility.

Confirm the changes

grep NVME_ .config

Compile and install the kernel

(To save time you can utilize multiple CPUs by including the j option)

make -j 16
make -j 16 modules_install install

Confirm that the build is included in the boot menu entry

(This is dependent on the bootloader being used, for GRUB2)

cat /boot/grub2/grub.cfg | grep menuentry

Set the build as the default boot option

grub2-set-default 'Red Hat Enterprise Linux Server (4.11.0) 7.x (Maipo)’

Reboot the system

reboot

Confirm that the kernel has been updated:

uname -a 
Linux host.name 4.11.0 #1 SMP date  x86_64 GNU/Linux

NVMe CLI Update

Download the correct version of the NVMe CLI utility that includes TCP:

git clone https://github.com/solarflarecommunications/nvme-cli

Update the NVMe CLI utility:

cd nvme-cli
make
make install

Target setup

Load the target driver

This should automatically load the dependencies,nvmenvme_core and nvme_fabrics

modprobe nvmet_tcp

Set up storage subsystem

mkdir /sys/kernel/config/nvmet/subsystems/ramdisk
echo 1 > /sys/kernel/config/nvmet/subsystems/ramdisk/attr_allow_any_host
mkdir /sys/kernel/config/nvmet/subsystems/ramdisk/namespaces/1
echo -n /dev/ram0 > /sys/kernel/config/nvmet/subsystems/ramdisk/namespaces/1/device_path
echo 1 > /sys/kernel/config/nvmet/subsystems/ramdisk/namespaces/1/enable

Setup port

mkdir /sys/kernel/config/nvmet/ports/1
echo "ipv4" > /sys/kernel/config/nvmet/ports/1/addr_adrfam
echo "tcp" > /sys/kernel/config/nvmet/ports/1/addr_trtype
echo "11345" > /sys/kernel/config/nvmet/ports/1/addr_trsvcid
echo "10.0.0.1" > /sys/kernel/config/nvmet/ports/1/addr_traddr

Associate subsystem with port

ln -s /sys/kernel/config/nvmet/subsystems/ramdisk /sys/kernel/config/nvmet/ports/1/subsystems/ramdisk

Initiator setup

Load the initiator driver

This should automatically load the dependencies,nvmenvme_core and nvme_fabrics.

modprobe nvme_tcp

Use the NVMe CLI utility to connect the initiator to the target:

nvme connect -t tcp -a 10.0.0.1 -s 11345 -n ramdisk

lsblk should now show an NVMe device.

What’s a Smart NIC?

scottcschweitzer Microsegmentation, networking, Security

While reading these words, it’s not just your brain doing the processing required to make this feat possible. We’ve all seen over and under exposed photos and can appreciate the decision making necessary to achieve a perfect light balanced photo. In the laboratory, we observed that the optic nerve connecting the eye to the brain is responsible for measuring the intensity of the light hitting the back of your eye. In response to this data, each optic nerve dynamically adjusts the aperture of the iris in your eye connected to this nerve to optimize these levels. For those with some photography experience, you might recall that there is a direct relationship between aperture (f-stop) and focal length. It also turns out that your optic nerve, after years of training as a child, has come to realize you’re reading text up close, so it is now also responsible for modifying the muscles around that eye to sharpen your focus on this text. All this data processing is completed before your brain has even registered the first word in the title. Imagine if your brain was responsible for processing all the data and actions that are required for your body to function properly?

Much like your optic nerve, the difference between a standard Network Interface Card (NIC) and a smart NIC is how much processing the Smart NIC offloads from the host CPU. Until recently Smart NICs were designed around Field Programmable Gate Array (FPGA) platforms costing thousands of dollars. As their name implies, FPGAs are designed to accept localized programming that can be easily updated once installed. Now a new breed of Smart NIC is emerging that while it isn’t nearly as flexible as an FPGA, they contain several sophisticated capabilities not previously found in NICs costing only a few hundred dollars. These new affordable Smart NICs can include a firewall for security, a layer 2/3 switch for traffic steering, several performance acceleration techniques, and network visibility with possibly remote management.

The firewall mentioned above filters all network packets against a table built specifically for each local Internet Protocol (IP) address under control. An application processing network traffic is required to register a numerical network port. This port then becomes the internal address to send and receive network traffic. Filtering at the application level then becomes a simple process of only permitting traffic for specific numeric network ports. The industry has labeled this “application network segmentation,” and in this instance, it is done entirely in the NIC. So How does this assist the host x86 CPU? It turns out that by the point at which operating system software filtering kicks in the host CPU has often expended over 10K CPU cycles to process a packet. If the packet is dropped the cost of that drop is 10K lost host CPU cycles. If that filtering was done in the NIC, and the packet was then dropped there would be NO host CPU impact.

Smart NICs also often have an internal switch which is used to steer packets within the server rapidly. This steering enables the NIC to move packets to and from interfaces and virtual NIC buffers which can be mapped to applications, virtual machines or containers. Efficiently steering packets is another offload method that can dramatically reduce host CPU overhead.

Improving overall server performance, often through kernel bypass, has been the providence of High-Performance Computing (HPC) for decades. Now it’s available for generic Ethernet and can be applied to existing and off the shelf applications. As an example, Solarflare has labeled its family of Kernel Bypass accelerations techniques Universal Kernel Bypass (UKB). There are two classes of traffic to accelerate: network packet and application sockets based. To speed up network packets UKB includes an implementation of the Data Plane Development Kit (DPDK) and EtherFabric VirtualInterface (EF_VI), both are designed to deliver high volumes of packets, well into the 10s of millions per second, to applications familiar with these Application Programming Interfaces (APIs). For more standard off-the-shelf applications there are several sockets based acceleration libraries included with UKB: ScaleOut Onload, Onload, and TCPDirect. While ScaleOut Onload (SOO) is free and comes with all Solarflare 8000 series NICs, Onload (OOL) and TCPDirect require an additional license as they provide micro-second and sub-microsecond 1/2 round trip network latencies. By comparison, SOO delivers 2-3 microsecond latency, but the real value proposition of SOO is the dramatic reduction in host CPU resources required to move network data. SOO is classified as “zero-copy” because network data is copied once directly into or out of your application’s buffer. SOO saves the host CPU thousands of instructions, multiple memory copies, and one or more CPU context switches, all dramatically improve application performance, often 2-3X, depending on how network intense an application is.

Finally, Smart NICs can also securely report NIC network traffic flows, and packet counts off the NIC to a centralized controller. This controller can then graphically display for network administrators everything that is going on within every server under its management. This is real enterprise visibility, and since only flow metadata and packet counts are being shipped off NIC over a secure TLS link the impact on the enterprise network is negligible. Imagine all the NICs in all your servers reporting in their traffic flows, and allowing you to manage and secure those streams in real time, with ZERO host CPU impact. That’s one Smart NIC!

What are Neural Class Networks?

scottcschweitzer networking

In 1971 Intel released the 4004. The 4004 was their first 4-bit single core processor, and for the next 34 years, that’s pretty much how x86 computing progressed. Sure they bumped up the architecture and speed as designs and processes improved, but one thing remained constant a single processing engine. Under pressure in the 1990s from Unix workstations driven by Reduced Instruction Set Computing (RISC) with multi-core pipeline architectures like IBM’s PowerPC, Sun’s Ultra-SPARC and the MIPS R3000 Intel began exploring multi-core architectures.

So from 1971 until 2005, every x86 processor had a single core, life was simple. Intel even provided reference designs so system builders could even put two of these processors into the same system. Sure there were fringe companies that developed Symmetrical Multi-Processing (SMP) systems (ex. Sequent & NEC) with more than two CPU sockets, but they were large frame expensive custom servers not found in general mainstream use. So if you wanted to scale out your computational capacity to tackle a tough problem, you had to rack more servers.

This single core challenge largely drove commodity Linux clustering making it all the rage by the turn of the century, particularly in high-performance computing (HPC). It wasn’t uncommon to tightly couple 1,000 or more dual socket single core systems together to tackle a tough computational problem. Government agencies leveraged large clusters to model our nuclear stockpile and computationally secure it. Auto companies crashed dozens of virtual car designs on a daily basis, and oil companies crunched seismic data to compute untapped oil reserves. Then the game changed, and x86 shifted to multi-core processors. Yesterday Intel announced the availability of their new Skylake server platform, but the industry is already refocusing on 2018’s Cascade Lake 32 core, 64 thread, server chip. So why does all this matter?

As a general rule, Google doesn’t publish or confirm the computation capacity of their data centers. If we pick a specific example, their Oregon data center, we can apply particular assumptions and project that it’s roughly 100,000 servers. Again assuming they are using dual socket eight core hyper-threaded processors this translates to 3.2 million parallel threads of computation tightly coupled in one physical location potentially addressing a single problem. Structures like this are quickly approximating what took nature millions of years to develop, an organic brain based on the neuron. Neurons on average have 7,000 connections to both local and remote neurons within their system, that’s a considerable amount of networking per single computational unit.

By contrast, the common cockroach has one million neurons, and that frog you played with as a kid sports 16 million. So if we equate a neuron to a single thread of execution on an x86 that would put a Google data center somewhere between the cockroach and a frog. If in 2018 Google were to upgrade the Oregon facility to Cascade Lakes it would still be only 12.8 million threads, still less than a frog. Given the geometric core growth though it will only be a few decades before Google is deploying data centers approaching the capability of the 86 billion neurons found in the human brain. Oh, and that’s assuming an x86 thread, and a neuron is even computationally similar.

So what is Neural Class Networking? As mentioned above a neuron on average is connected to 7,000 other neurons. Imagine if every hardware thread of execution in your server were networked to even 64 external threads of execution on related systems working on the same problem, that’s the start of neural class networking. Today we have servers with typically 32 threads of execution. Solarflare’s newest generation of XtremeScale Smart NICs provides 2,048 virtual NICs, so each thread of those 32 threads has the capability to sustain 64 dedicated hardware paths to other external threads. That’s the start of Neural Class Networking.