Last week we hosted the Open vSwitch 2014 Fall Conference, which was another great opportunity to demonstrate our continued investment in leading open source technologies. To get a sense of the energy and enthusiasm at the event, take a quick view of this video we captured with attendees.
I’ve been thinking about the key takeaways from everything I saw and everyone I spoke with.
First, there’s huge interest in Open vSwitch performance, both in terms of measurement and improvement. The talks from Rackspace and Noiro Networks/Cisco led me to believe that we’ve reached the point where Open vSwitch performance is good enough on hypervisors for most applications, and often faster than competing software solutions such as the Linux bridge.
Talks from Intel and one from Luigi Rizzo at the University of Pisa demonstrated that by bypassing the kernel entirely through DPDK or netmap, respectively, we haven’t reached the limits of software forwarding performance. Based on a conversation I had with Chris Wright from Red Hat, this work is helping the Linux kernel community look into reducing the overhead of the kernel, so that we can see improved performance without losing the functionality provided by the kernel.
Johann Tönsing from Netronome also presented a talk describing all the ways that Netronome’s NPU hardware can accelerate OpenFlow and Open vSwitch; I’ve talked to Johann many times before, but I had never realized how many different configurations their hardware supports, so this was an eye-opening talk for me.
Next, enhancing Open vSwitch capabilities at L4 through L7 is another exciting area. Our own Justin Pettit was joined by Thomas Graf from Noiro to talk about the ongoing project to add support for NAT and tracking L4 connections, which is key to making Open vSwitch capable of implementing high-quality firewalls. A later talk by Franck Baudin from Qosmos presented L7 enhancements to this capability.
The final area that I saw highlighted at the conference is existing applications for Open vSwitch today. Peter Phaal from InMon, for example, demonstrated applications for sFlow in Open vSwitch. I found his talk interesting because although I knew about sFlow and had talked to Peter before, I hadn’t realized all of the varied uses for sFlow monitoring data. Vikram Dham also showed his uses for MPLS in Open vSwitch and Radhika Hirannaiah her use case for OpenFlow and Open vSwitch in traffic engineering.
I want to thank all of our participants and the organizing committee for helping to put together such an amazing event.
This week, VMware will be hosting the Open vSwitch 2014 Fall Conference, with more than 200 attendees and nearly two dozen talks on a variety of subjects from a key participants. The full schedule is available here, and we’ll be doing a wrap up of some of the takeaways from the conference a bit later.
For the uninitiated, Open vSwitch is a production quality, multilayer virtual switch licensed under the open source Apache 2.0 license. It is designed to enable massive network automation through programmatic extension, while still supporting standard management interfaces and protocols (e.g. NetFlow, sFlow, IPFIX, RSPAN, CLI, LACP, 802.1ag). In addition, it is designed to support distribution across multiple physical servers similar to VMware’s vDS or Cisco’s Nexus 1000V. See full feature list here
For more information on OVS, I encourage you to check out the OVS website.
In the mean time, take a read about latest Open vSwitch developments in this post on Network Heresy by OVS core contributors Justin Pettit, Ben Pfaff, and Ethan Jackson.
This post was written by Roie Ben Haim and Max Ardica, with a special thanks to Jerome Catrouillet, Michael Haines, Tiran Efrat and Ofir Nissim for their valuable input.
The modern data center design is changing, following a shift in the habits of consumers using mobile devices, the number of new applications that appear every day and the rate of end-user browsing which has grown exponentially. Planning a new data center requires meeting certain fundamental design guidelines. The principal goals in data center design are: Scalability, Redundancy and High-bandwidth.
In this blog we will describe the Equal Cost Multi-Path functionality (ECMP) introduced in VMware NSX release 6.1 and discuss how it addresses the requirements of scalability, redundancy and high bandwidth. ECMP has the potential to offer substantial increases in bandwidth by load-balancing traffic over multiple paths as well as providing fault tolerance for failed paths. This is a feature which is available on physical networks but we are now introducing this capability for virtual networking as well. ECMP uses a dynamic routing protocol to learn the next-hop towards a final destination and to converge in case of failures. For a great demo of how this works, you can start by watching this video, which walks you through these capabilities in VMware NSX.
Scalability and Redundancy and ECMP
To keep pace with the growing demand for bandwidth, the data center must meet scale out requirements, which provide the capability for a business or technology to accept increased volume without redesign of the overall infrastructure. The ultimate goal is avoiding the “rip and replace” of the existing physical infrastructure in order to keep up with the growing demands of the applications. Data centers running business critical applications need to achieve near 100 percent uptime. In order to achieve this goal, we need the ability to quickly recover from failures affecting the main core components. Recovery from catastrophic events needs to be transparent to end user experiences.
ECMP with VMware NSX 6.1 allows you to use upto a maximum of 8 ECMP Paths simultaneously. In a specific VMware NSX deployment, those scalability and resilience improvements are applied to the “on-ramp/off-ramp” routing function offered by the Edge Services Gateway (ESG) functional component, which allows communication between the logical networks and the external physical infrastructure.
External user’s traffic arriving from the physical core routers can use up to 8 different paths (E1-E8) to reach the virtual servers (Web, App, DB).
In the same way, traffic returning from the virtual server’s hit the Distributed Logical Router (DLR), which can choose up to 8 different paths to get to the core network.
How is the path determined:
NSX for vSphere Edge Services Gateway device:
When a traffic flow needs to be routed, the round robin algorithm is used to pick up one of the links as the path for all traffic of this flow. The algorithm ensures to keep in order all the packets related to this flow by sending them through the same path. Once the next-hop is selected for a particular Source IP and Destination IP pair, the route cache stores this. Once a path has been chosen, all packets related to this flow will follow the same path.
There is a default IPv4 route cache timeout, which is 300 seconds. If an entry is inactive for this period of time, it is then eligible to be removed from route cache. Note that these settings can be tuned for your environment.
Distributed Logical Router (DLR):
The DLR will choose a path based on a Hashing algorithm of Source IP and Destination IP.
What happens in case of a failure on one of Edge Devices?
In order to work with ECMP the requirement is to use a dynamic routing protocol: OSPF or BGP. If we take OSPF for example, the main factor influencing the traffic outage experience is the tuning of the OSPF timers.
OSPF will send hello messages between neighbors, the OSPF “Hello” protocol is used and determines the Interval as to how often an OSPF Hello is sent.
Another OSPF timer called “Dead” Interval is used, which is how long to wait before we consider an OSPF neighbor as “down”. The OSPF Dead Interval is the main factor that influences the convergence time. Dead Interval is usually 4 times the Hello Interval but the OSPF (and BGP) timers can be set as low as 1 second (for Hello interval) and 3 seconds (for Dead interval) to speed up the traffic recovery.
In the example above, the E1 NSX Edge has a failure; the physical routers and DLR detect E1 as Dead at the expiration of the Dead timer and remove their OSPF neighborship with him. As a consequence, the DLR and the physical router remove the routing table entries that originally pointed to the specific next-hop IP address of the failed ESG.
As a result, all corresponding flows on the affected path are re-hashed through the remaining active units. It’s important to emphasize that network traffic that was forwarded across the non-affected paths remains unaffected.
Troubleshooting and visibility
With ECMP it’s important to have introspection and visibility tools in order to troubleshoot optional point of failure. Let’s look at the following topology.
A user outside our Data Center would like to access the Web Server service inside the Data Center. The user IP address is 192.168.100.86 and the web server IP address is 172.16.10.10.
This User traffic will hit the Physical Router (R1), which has established OSPF adjacencies with E1 and E2 (the Edge devices). As a result R1 will learn how to get to the Web server from both E1 and E2 and will get two different active paths towards 172.16.10.10. R1 will pick one of the paths to forward the traffic to reach the Web server and will advertise the user network subnet 192.168.100.0/24 to both E1 and E2 with OSPF.
E1 and E2 are NSX for vSphere Edge devices that also establish OSPF adjacencies with the DLR. E1 and E2 will learn how to get to the Web server via OSPF control plane communication with the DLR.
From the DLR perspective, it acts as a default gateway for the Web server. This DLR will form an OSPF adjacency with E1 and E2 and have 2 different OSPF routes to reach the user network.
From the DLR we can verify OSPF adjacency with E1, E2.
We can use the command: “show ip ospf neighbor”
From this output we can see that the DLR has two Edge neighbors: 184.108.40.206 and 192.168.100.10.The next step will be to verify that ECMP is actually working.
We can use the command: “show ip route”
The output from this command shows that the DLR learned the user network 192.168.100.0/24 via two different paths, one via E1 = 192.168.10.1 and the other via E2 = 192.168.10.10.
Now we want to display all the packets which were captured by an NSX for vSphere Edge interface.
In the example below and in order to display the traffic passing through interface vNic_1, and which is not OSPF protocol control packets, we need to type this command: “debug packet display interface vNic_1 not_ip_proto_ospf”
We can see an example with a ping running from host 192.168.100.86 to host 172.16.10.10
If we would like to display the captured traffic to a specific ip address 172.16.10.10, the command capture would look like: “debug packet display interface vNic_1 dst_172.16.10.10”
Useful CLI for Debugging ECMP
To check which ECMP path is chosen for a flow
debug packet display interface IFNAME
To check the ECMP configuration
show configuration routing-global
To check the routing table
show ip route
To check the forwarding table
show ip forwarding
Useful CLI for Dynamic Routing
show ip ospf neighbor
show ip ospf database
show ip ospf interface
show ip bgp neighbors
show ip bgp
ECMP Deployment Consideration
ECMP currently implies stateless behavior. This means that there is no support for stateful services such as the Firewall, Load Balancing or NAT on the NSX Edge Services Gateway. The Edge Firewall gets automatically disabled on ESG when ECMP is enabled. In the current NSX 6.1 release, the Edge Firewall and ECMP cannot be turned on at the same time on NSX edge device. Note however, that the Distributed Firewall (DFW) is unaffected by this.
Roie Ben Haim works as a professional services consultant at VMware, focusing on design and implementation of VMware’s software-defined data center products. Roie has more than 12 years in data center architecture, with a focus on network and security solutions for global enterprises. An enthusiastic M.Sc. graduate, Roie holds a wide range of industry leading certifications including Cisco CCIE x2 # 22755 (Data Center, CCIE Security), Juniper Networks JNCIE – Service Provider #849, and VMware vExpert 2014, VCP-NV, VCP-DCV. Follow his personal blog at http://roie9876.wordpress.com/
Max Ardica is a senior technical product manager in VMware’s networking and security business unit (NSBU). Certified as VCDX #171, his primary task is helping to drive the evolution of the VMware NSX platform, building the VMware NSX architecture and providing validated design guidance for the software-defined data center, specifically focusing on network virtualization. Prior to joining VMware, Max worked for almost 15 years at Cisco, covering different roles, from software development to product management. Max owns also a CCIE certification (#13808).
This post was co-authored by Justin Pettit, Staff Engineer, Networking & Security Business Unit at VMware, and Ravi Shekhar, Distinguished Engineer, S3BU at Juniper Networks.
As discussed in other blog posts and presentations, long-lived, high-bandwidth flows (elephants) can negatively affect short-lived flows (mice). Elephant flows send more data, which can lead to queuing delays for latency-sensitive mice.
VMware demonstrated the ability to use a central controller to manage all the forwarding elements in the underlay when elephant flows are detected. In environments that do not have an SDN-controlled fabric, an alternate approach is needed. Ideally, the edge can identify elephants in such a way that the fabric can use existing mechanisms to treat mice and elephants differently.
Differentiated services (diffserv) were introduced to bring scalable service discrimination to IP traffic. This is done using Differentiated Services Code Point (DSCP) bits in the IP header to signal different classes of service (CoS). There is wide support in network fabrics to treat traffic differently based on the DSCP value.
A modified version of Open vSwitch allows us to identify elephant flows and mark the DSCP value of the outer IP header. The fabric is then configured to handle packets with the “elephant” DSCP value differently from the mice.
Figure 1: Elephants are detected at the edge of the network and signaled to the fabric through DSCP. Based on these code points, the fabric can treat elephant traffic differently from mice
Detecting and Marking Elephants with Open vSwitch
Open vSwitch’s location at the edge of the network gives it visibility into every packet in and out of each guest. As such, the vSwitch is in the ideal location to make per-flow decisions such as elephant flow detection. Because environments are different, our approach provides multiple detection mechanisms and actions so that they can be used and evolve independently.
An obvious approach to detection is to just keep track of how many bytes each flow has generated. By this definition, if a flow has sent a large amount of data, it is an elephant. In Open vSwitch, the number of bytes and an optional duration can be configured. By using a duration, we can ensure that we don’t classify very short-lived flows as elephants. We can also avoid identifying low-bandwidth but long-lived flows as elephants.
An alternate approach looks at the size of the packet that is being given to the NIC. Most NICs today support TCP Segmentation Offload (TSO), which allows the transmitter (e.g., the guest) to give the NIC TCP segments up to 64KB, which the NIC chops into MSS-sized packets to be placed on the wire.
Because of TCP’s slow start, the transmitter does not immediately begin sending maximum-sized packets to the NIC. Due to our unique location, we can see the TCP window as it opens, and tag elephants earlier and more definitively. This is not possible at the top-of-rack (TOR) or anywhere else in the fabric, since they only see the segmented version of the traffic.
Open vSwitch may be configured to track all flows with packets of a specified size. For example, by looking for only packets larger than 32KB (which is much larger than jumbo frames), we know the transmitter is out of slow-start and making use of TSO. There is also an optional count, which will trigger when the configured number of packets with the specified size is seen.
Some new networking hardware provides some elephant flow mitigation by giving higher priority to small flows. This is achieved by tracking all flows and placing new flows in a special high-priority queue. When the number of packets in the flow has crossed a threshold, the flow’s packets from then on are placed into the standard priority queue.
This same effect can be achieved using the modified Open vSwitch and a standard fabric. For example, by choosing a packet size of zero and threshold of ten packets, each flow will be tracked in a hash table in the kernel and tagged with the configured DSCP value when that flow has generated ten packets. Whether mice are given a high priority or elephants are given a low priority, the same effect is achieved without the need to replace the entire fabric.
Handling Elephants with Juniper Devices
Juniper TOR devices (such as QFX5100) and aggregation devices (such as MX, EX9200) provide a rich diffserv model CoS to to achieve these goals in the underlay. These include:
Elaborate controls for packet admittance with dedicated and shared limits. Dedicated limits provide a minimum service guarantee, and shared limits allow statistical sharing of buffers across different ports and priorities.
A large number of flexibly assigned queues; up to 2960 unicast queues at the TOR and 512K at the aggregation device.
Enhanced and varied scheduling methods to drain these queues: strict and round-robin scheduling with up to 4-levels of hierarchical schedulers.
Shaping and metering to control the rate of injection of traffic from different queues of a TOR in the underlay network. By doing this, bursty traffic at the edge of the physical network can be leveled out before it reaches the more centrally shared aggregation devices.
Sophisticated controls to detect and notify congestion, and set drop thresholds. These mechanisms detect possible congestion in the network sooner and notify the source to slow down (e.g. using ECN).
With this level of flexibility, it is possible to configure these devices to:
Enforce minimum bandwidth allocation for mice flows and/or maximum bandwidth allocation for elephant flows on a shared link.
When experiencing congestion, drop (or ECN mark) packets of elephant flows more aggressively than mice flows. This will result in TCP connections of elephant flows to back off sooner, which alleviates congestion in the network.
Take a different forwarding path for elephant flows from that of mice flows. For example, a TOR can forward elephant flows towards aggregation switches with big buffers and spread mice flows towards multiple aggregation switches that support low-latency forwarding.
By inserting some intelligence at the edge and using diffserv, network operators can use their existing fabric to differentiate between elephant flows and mice. Most networking gear provides some capabilities, and Juniper, in particular, provides a rich set of operations that can be used based on the DSCP. Thus, it is possible to reduce the impact of heavy hitters without the need to replace hardware. Decoupling detection from mitigation allows each to evolve independently without requiring wholesale hardware upgrades.
Adoption of network virtualization and SDN technologies from VMware and Arista Networks simplifies cloud infrastructure and enables automation to reduce timescales of cloud and network service provisioning
Offering the largest enterprise-class cloud footprint in Europe, Colt, an established leader in delivering integrated network, data center, voice and IT services, has implemented software- defined networking [SDN] and network virtualization to simplify how its managed IT and cloud-based networking environment is deployed, managed and scaled throughout its data centers.
Following an extensive review, Colt selected Arista to provide high speed 10 and 40 gigabit Ethernet cloud-centric switches as an underlay network fabric and VMware NSX™ network virtualization to deliver a fully decoupled software network overlay.
SDN paves the way for automated cloud service delivery
The shift to SDN will provide a flexible, scalable, efficient and cost effective way to support the delivery of Colt’s managed IT services, including cloud based services. This makes Colt one of the first service providers in Europe to adopt SDN in a production environment to remove automate cloud service delivery.
As a result of deploying a new network architecture based on Arista and VMware networking technologies, the time for Colt to add, change or modify services will now take minutes rather than days, and will enable Colt to onboard customers faster and expand its service portfolio quicker.
The big transformation in IT in recent years has been the development of cloud services with IT capacity purchased on-demand. In contrast, networking has remained relatively static. The adoption of server virtualization over the past decade as the foundation for cloud computing and IT-as-a-service have resulted in a completely new operational model for provisioning and managing application workloads. However, the operating model of the network to which dynamic virtualized services are connected has not evolved to help businesses achieve the full benefits of mobile-cloud.
Mirko Voltolini, VP Technology and Architecture at Colt says, “The excitement around SDN and network virtualization is that – for the first time – networking is becoming more software orientated so we’re able to dynamically orchestrate service modification and activation in real-time. In other words, network connectivity can now keep up when virtual machines and associated compute and storage change or are moved within distributed data centers. Ultimately this means that servers, storage and now the network are in synch so that we can meet the specific needs of our customers in a timescale they demand.”
With more than 25,000 customers worldwide, Colt offers an information delivery platform comprising network, voice, data center and IT services sold directly to its enterprise customers or indirectly via channel partners and operators. In Europe, it has invested significantly to create a pan European network spanning 22 countries, 195 connected cities, around 19,800 buildings, as well as operating 42 metropolitan area networks.
Turning to the specialist technology firms has really delivered
Colt first considered adopting network overlay technology three years ago. It went out to tender approaching only large, mainstream technology suppliers and was disappointed by the response received. The cost was too great and solutions not really mature enough to warrant changing. Eighteen months ago, it revisited the process given the technology had evolved, expanding the shortlist of suppliers asked to provide proposals to include specialist firms like Arista and VMware.
VMware NSX enables Colt to decouple the data center network from the underlying physical hardware to gain massive scale while simplifying network design and operations. With NSX, Colt is able to consolidate operations for four disparate physical networks running in the data center and manage these networks as a single logical network. Colt has developed a new data center architecture that leverages the scalability of a Layer 3 data center fabric and NSX’s overlay network virtualization platform.
Chris King, vice president, product marketing, networking and security business unit at VMware, said, “Colt is an all too common story of an organization that simply hit the limits of what the physical network could provide in a virtualized world. VLAN limitations prevented Colt’s ability to scale. They needed to simplify the physical infrastructure in order to gain flexibility which in turn would allow them to adapt quickly to the business environment. VMware NSX helped Colt successfully execute a data center re-architecture which can now operate at cloud scale with better performance, easier management and lower overall costs.”
In addition to wanting to capitalize on the all the benefits offered by network overlays, the requirement for a new switch supplier was driven by Colt’s need to replace its existing legacy switches which had reached end of life and are not supported anymore. Furthermore, the business wanted to reduce the total cost of ownership [TCO] of its networking equipment.
Voltolini explains, “Our target was to reduce the unit cost of our switches which includes the cost per port, along with maintenance, power, space and so on. We wanted a step change in TCO which we will now achieve working with Arista.”
VXLAN addresses the limitations of Spanning Tree
From a technical perspective, Colt also wanted to move away from legacy protocols like Spanning Tree protocol which requires ports to be available – but not used – to deliver service availability. This underutilizes switch assets and adds unnecessary cost to its operation. Moreover, Colt required new switches which could scale to support increased connectivity capabilities both in terms of the number of ports [so that more customers can be connected] as well as logical scale.
Voltolini says, “The new VXLAN protocol removes traditional Ethernet limitations which is crucial for a service provider so that we can handle multiple tenants per port across numerous physical locations.”
Ultimately Arista switches will be installed in all Colt data center locations, the roll out of which will be driven by service and capacity demands. The expectation is that this will happen over the next 18 to 24 months. Deployment is made straightforward as all Arista switches – irrespective of port count or speed – feature the same network operating system, the Arista EOS.
Mark Foss, VP Global Operations and Marketing, concludes, “It is important to stress that this project is one of collaboration. Being an innovative nimble company, we’re accommodating Colt’s requirements and helping shape their service design, while they’re guiding us in terms of our future product roadmap so we develop features pertinent to all cloud service providers.”
VMware NSX and Palo Alto Networks are transforming the data center by combining the fast provisioning of network and security services with next-generation security protection for East-West traffic. At VMworld, John Spiegel, Global IS Communications Manager for Columbia Sportswear will take the stage to discuss their architecture, their micro-segmentation use case and their experience. This is session SEC1977 taking place on Tuesday, Aug 26, 2:30-3:30 p.m.
Micro-segmentation is quickly emerging as one of the primary drivers for the adoption of NSX. Below, John shares Columbia’s security journey ahead of VMworld
When I started at Columbia, we were about a $500 million company. Now we’re closing in on $2 billion and hoping to get to $3 billion rather quickly. So as you can imagine, our IT infrastructure has to scale with the business. In 2009, we embarked on a huge project to add a redundant data center for disaster recovery. As part of the project, we partnered with VMware and quickly created a nearly 100% virtualized datacenter. It was a huge success. But something was missing; a security solution that matched our virtualized data center. There just wasn’t a great way to insert security in order to address east-west traffic between VMs, nor have the security tied to the applications as they moved around dynamically.
We set out looking for a solution to bridge that gap.
To address our security needs in the data center, we looked at several different strategies and at that time, there really weren’t any good solutions. Many of the solutions were physical in nature. They required us to do some crazy configurations to apply security. We looked at the Cisco 6500 firewall blades, Juniper’s virtual solution and a few other lightweight security offerings, but they just didn’t have what we needed. The solutions at the time didn’t have what we needed. We kept looking.
At VMworld last year, we were introduced to VMware NSX. I saw the power of the platform, and it all started to click. And when Palo Alto Networks (our perimeter firewall vendor) announced they were a major partner, and that their technology integrated with NSX to give us an additional level of security, things really came together for us. The ability to drive security down into the infrastructure, down to the kernel level, and then take advantage of Palo Alto Networks next generation security was very attractive. Doing micro-segmentation with NSX, and then having the option of inserting next generation firewalling services from Palo Alto Networks in those areas of the business that require them, will really help us improve our overall security posture. A solution like this is where we need to be. These tools give us the ability to manage both physical and virtual security policies centrally with Palo Alto Networks management tool Panorama. I know that when workloads move the security and policies follow the workloads.
To me, that’s what it is about – advanced security inside the data center, plus automation via software that’s completely independent of the underlying physical infrastructure. With solutions such as NSX and the integration with Palo Alto Networks to provide advanced security services, we are going put security back in the data center, the right way.=
Nicolas Vermandé (VCDX#055) is practice lead for Private Cloud & Infrastructure at Kelway, a VMware partner. Nicolas covers the Software-Defined Data Center on his blog www.my-sddc.om,
This series of posts describes a specific use case for VMware NSX in the context of Disaster Recovery. The goal is to demonstrate the routing and programmability capabilities through a lab scenario. This first part presents the NSX components and details the use case. The second part will show how to deploy the lab and the third part will deal with APIs and show how to use python to execute REST API calls to recreate the required NSX components at the recovery site.
When considering dual datacenter strategy with VMs recovery in mind, one important decision is whether to adopt an active/active or active/standby model. The former is generally much more complex to manage because it requires double the work in terms of procedures, testing and change controls. In addition, capacity management becomes challenging as you need to accommodate physical resources to be able to to run all workloads within whatever site. On top of that, stretched VLANs are sometimes deployed across datacenters so that recovered VMs can keep their IP addresses. This is generally very costly if you want to leverage proper L2 extension technology, such as Cisco OTV.
Alternatively, in a SDDC environment, you can leverage VMware NSX to efficiently manage connectivity and network changes required in the event of a full site failover. NSX gives you the ability to maintain the same IP address scheme for all you workloads by leveraging APIs, with little effort. Or with more granularity, you could even move a single subnet as part of a specific recovery plan. NSX will make this possible by providing:
An overlay network that allows you to decouple the backend VM network from the physical network. NSX-V is using VXLAN, each ESXi host acting as a VTEP.
Programmability through RESTful APIs that allows you to provision Logical Switches and modify Logical Routers configuration in seconds.
Dynamic routing protocol (OSPF, IS-IS, BGP) that will advertise VM subnets to your enterprise network, making them accessible for end users or other applications (North-South or East-West traffic)
As many NSX introduction blog posts can be found on the web (like here or here), I’m not gonna spend much time on this topic. NSX components are:
NSX Manager: it’s the single point of configuration and the REST API (and UI) interface. It is provided as a VM appliance and is actually the only appliance you have to manually install. There is a 1:1 mapping between the vCenter Server an the NSX Manager. The manager is responsible for deploying NSX Controllers, NSX Edge Gateways and Logical Router Controllers. It also installs the Distributed Routing and the firewall kernel modules on ESXi hosts, as well as the User World Agent (UWA). NSX configuration is accessible through vCenter once you’ve installed the NSX plugin.
NSX Controller: it provides a control plane to distribute VXLAN Logical Routing and Switching network information to ESXi hosts. It also enables ARP suppression to reduce flooding. It is typically implemented as a 3-node cluster and maintains MAC, ARP and VTEP tables. It is finally responsible for installing routes into each ESXi host.
Logical Switch (LS): it acts as the L2 domain boundary for VMs, identified by a VXLAN ID (VNI) and associated with a specific subnet. Its vCenter representation is a distributed Portgroup with specific capabilities.
Distributed Logical Router (DLR): it’s the distributed L3 first-hop for VM traffic. As its name suggests, it’s completely distributed. You can think about it as an anycast gateway, where each ESXi corresponds to a node, sharing a single virtual IP and virtual MAC address. The data-path routing process runs within each ESXi in vmkernel space and enables East-West traffic optimisation, avoiding well-known hair-pinning effects when VMs want to talk to their default gateway.
Logical Router Control VM: it provides the DLR with a control plane and can be deployed as a redundant pair of VM appliances, in an active/standby fashion. It supports both OSPF and BGP as dynamic routing protocols. The Control VM receives its initial configuration from the NSX Manager.
Edge Services Gateway (ESG): it provides network perimeter services to the virtual environment. It is intended for North-South communication, i.e. between the physical and the virtual network or at the edge of your tenant. It is NOT distributed, meaning that its placement is critical. It can run in HA-mode, where the appliances are deployed in an active/standby fashion. The HA mechanism doesn’t rely on VMware HA (as some people at Cisco seem to think), but with minimum common sense, you’re gonna create a DRS anti-affinity rule to separate active and stanby VMs. Depending on specific requirements, the edge gateway can be deployed with several sizes:
Compact (1vCPU – 512MB RAM)
Large (2 vCPUs – 1GB RAM)
Quad-Large (4 vCPUs – 1GB RAM)
X-Large (6 vCPUs – 8GB RAM).
Available services include: Firewall, NAT, DHCP, Routing, Load-Balancing, Site-to-Site VPN, SSL VPN and L2VPN.
Distributed Firewall (DFW): It enables distributed security capabilities at VM NIC level as an East-West L2-L4 stateful firewall. The module is present on each ESXi host as a kernel module and therefore removes any form of bottleneck. If you need more bandwidth, just add a new host! It also includes the Service Composer feature, which allows you to create specific services by integrating additional 3rd party capabilities to the firewall, such as endpoint services (e.g. Anitivirus, Data Security) and deep packet inspection (Palo Alto). I have to say that this feature is one of the most compelling to me!
The following picture shows how those components fit together:
To understand NSX concepts, it’s useful to map vSphere network components to NSX components:
In a traditional vSphere environment, a VM wishing to communicate with the outside world first hits a virtual port on the virtual switch. This virtual port is part of a Portgroup, which is basically a group of virtual network ports tagged with a specific VLAN ID. In the NSX world, when a VM is part of a Logical Switch, it hits a virtual port member of a Portgroup specifically created by the NSX Manager. It is created on every host member of the VDS, like a traditional distributed Portgroup. However, the difference is that all egress frames hitting this Portgroup will be forwarded inside a VXLAN tunnel, tagged with a specific external VLAN ID to transport the VXLAN frames on the physical network .
The role of the Logical Router is to connect two or more Logical Switches together, enabling routing between the corresponding subnets (you can assume 1 LS = 1 subnet). It also advertises (and learns) prefixes and routes to its neighbor(s) if a dynamic routing protocol has been activated. Alternatively, you can also configure static routes.
As an example, the following diagram shows the DLR establishing adjacency with the ESG, which is also running a dynamic routing protocol, and advertises VM subnets to the physical world. The ESG has its internal interface connected to a VXLAN and its uplink connected to a VLAN. As a result, the physical network can learn about the virtual network, and vice-versa.
Now that you’ve had a basic introduction to NSX principles, I can detail my scenario.In my lab environment, I’ve simulated the following architecture:
I didn’t actually deploy two sets of controllers and two managers linked to two different Virtual Centers in separate physical datacenters. Instead I’ve created logical containers called “Transport Zones” to make both virtual datacenters completely independent from a data-plane standpoint. The goal here is to demonstrate how to integrate virtual network operations into an orchestrated Disaster Recovery Plan with NSX. The only requirement is the ability to run a script as part of your DR procedures. This may be ultimately be achieved by VMware Site Recovery Manager, or another orchestration tool.
This architecture represents a traditional dual datacenter environment, connected over a L3 IP cloud. In a standard network environment, it basically means that you have to change VMs IP addresses upon recovery. (There are other alternatives, like host routes, RHI and NAT, but these solutions come at a certain complexity cost).
The main goal of the scenario is to show how to provide a flexible orchestrated Disaster Recovery solution without having to change VMs IP addresses. Let’s see how we can achieve this with NSX. The order of operations would be:
Disconnect LS1 and LS2 in DC1.
Create new LS in DC2: DR_LS1 and DR_LS2 (Or pre-create them without connecting the upstream DLR).
Add two new interfaces to DLR2 in DC2, with the same IP addresses as previously used by DLR1 to connect LS1 and LS2. In this way, we don’t have to change the default gateway of the recovered VMs.
Connect those interfaces to the corresponding LS.
Recover VMs in DC2.
Connect VMs to the appropriate LS.
Boot VMs and test connectivity.
Check route updates on the physical network.
Note: I’m assuming here that security devices configuration are synchronized between datacenters.
Because OSPF is running within the virtual network on both DLR1 and DLR2, routing updates will be sent up to the IP cloud to reflect that DR_LS1 and DR_LS2 subnets are now reachable through DC2. In the same way, because LS1 and LS2 have been disconnected from DLR1, corresponding routes will be removed to reflect that LS1 and LS2 subnets are not reachable in DC1 anymore. Magic??!! No, just awesome technology
The next post will focus on how to deploy this lab environment.
[This post was co-authored by VMware's Bruce Davie and Ken Duda from Arista Networks, and originally appeared on Network Heresy]
Almost a year ago, we wrote a first post about our efforts to build virtual networks that span both virtual and physical resources. As we’ve moved beyond the first proofs of concept to customer trials for our combined solution, this post serves to provide an update on where we see the interaction between virtual and physical worlds heading.
Our overall approach to connecting physical and virtual resources can be viewed in two main categories:
terminating the overlay on physical devices, such as top-of-rack switches, routers, appliances, etc.
managing interactions between the overlay and the physical devices that provide the underlay.
The latter topic is something we’ve addressed in some other recent posts (here, here and here) — in this blog we’ll focus more on how we deal with physical devices at the edge of the overlay. Continue reading →
In this post, Bruce Davie and T. Sridhar of VMware’s Networking and Security Business Unit take a look at a proposed a new encapsulation protocol that would standardize how traffic is tunneled over the physical infrastructure by networkoverlay software.
For as long as we’ve been doing Network Virtualization, there has been debate about how best to encapsulate the data. As we pointed out in an earlier post, it’s entirely reasonable for multiple encapsulations (e.g. VXLAN and STT) to co-exist in a single network. With the recent publication of “Geneve”, a new proposed encapsulation co-authored by VMware, Microsoft, Red Hat and Intel, we thought it would be helpful to clarify a few points regarding encapsulation for network virtualization. First, with all the investment made by us and our partners in developing support for VXLAN (described here), we very much intend to continue supporting VXLAN — indeed, we’ll be enhancing our VXLAN capabilities. Second, we want to explain why we believe Geneve is a necessary and useful addition to the network virtualization landscape.
The advantages a software-defined data center, using network virtualization as a core underpinning, include service delivery speed, operational efficiency, reduced hardware dependency and lower cost. However, by far the most popular use case by customers thus far has been the use of NSX for network microsegmentation. Why? Because perimeter-centric network security has proven insufficient, and micro-segmentation has to date been operationally and economically infeasible. With NSX, security teams, in partnership with their network and virtualization teams, are benefiting from network micro-segmentation to begin to transform their data center security architecture. Then read the VMware SDDC Micro-Segmentation White Paper.