Tag Archives: SDN

Tech

Microsoft needs SDN for Azure cloud

Leave a comment

Microsoft needs SDN for Azure cloud

Couldn’t scale without it, Azure CTO says
The Microsoft cloud, through which the company’s software products are delivered, has 22 hyper-scale regions around the world. Azure storage and compute usage is doubling every six months, and Azure lines up 90,000 new subscribers a month.

Six TED Talks that can change your career
Of the hundreds of TED talks available online, many are geared toward helping people view life in a new

Fifty-seven percent of the Fortune 500 use Azure and the number of hosts quickly grew from 100,000 to millions, said CTO Mark Russinovich during his Open Network Summit keynote address here this week. Azure needs a virtualized, partitioned and scale-out design, delivered through software, in order to keep up with that kind of growth.

“When we started to build these networks and started to see these types of requirements, the scale we were operating at, you can’t have humans provisioning things,” Russinovich said. “You’ve got to have systems that are very flexible and also delivering functionality very quickly. This meant we couldn’t go to the Web and do an Internet search for a scalable cloud controller that supports this kind of functionality. It just didn’t exist.”

Microsoft wrote all of the software code for Azure’s SDN. A description of it can be found here.
Microsoft uses virtual networks (Vnets) built from overlays and Network Functions Virtualization services running as software on commodity servers. Vnets are partitioned through Azure controllers established as a set of interconnected services, and each service is partitioned to scale and run protocols on multiple instances for high availability.

Controllers are established in regions where there could be 100,000 to 500,000 hosts. Within those regions are smaller clustered controllers which act as stateless caches for up to 1,000 hosts.
Related

Why is Microsoft killing off Internet Explorer?
Microsoft builds these controllers using an internally developed Service Fabric for Azure. Service Fabric has what Microsoft calls a microservices-based architecture that allows customers to update individual application components without having to update the entire application.

Microsoft makes the Azure Service Fabric SDK available here.
Much of the programmability of the Azure SDN is performed on the host server with hardware assist. A Virtual Filtering Platform (VFP) in Hyper-V hosts enable Azure’s data plane to act as a Hyper-V virtual network programmable switch for network agents that work on behalf of controllers for Vnet and other functions, like load balancing.

Packet processing is done at the host where a NIC with a Field Programmable Gate Array offloads network processing from the host CPU to scale the Azure data plane from 1Gbps to 40Gbps and beyond. That helps retain host CPU cycles for processing customer VMs, Microsoft says.

Remote Direct Memory Access is employed for the high-performance storage back-end to Azure.
Though SDNs and open source go hand-in-hand, there’s no open source software content in the Azure SDN. That’s because the functionality required for Azure was not offered through open source communities, Russinovich says.

“As these requirements were hitting us, there was no open source out there able to meet them,” he says. “And once you start on a path where you’re starting to build out infrastructure and system, even if there’s something else that comes along and addresses those requirements the switching cost is pretty huge. It’s not an aversion to it; it’s that we haven’t seen open source out there that really meets our needs, and there’s a switching cost that we have to take into account, which will slow us down.”

Microsoft is, however, considering contributing the Azure Service Fabric architecture to the open source community, Russinovich said. But there has to be some symbiosis.

“What’s secret sauce, what’s not; what’s the cost of contributing to open source, what’s the benefit to customers of open source, what’s the benefit to us penetrating markets,” he says. “It’s a constant evaluation.”

Some of the challenges in constructing the Azure SDN were retrofitting existing controllers into the Service Fabric, Russinovich says. That resulted in some scaling issues.
Resources

7 Critical Questions to Demystify DRaaS
“Some of the original controllers were written not using Service Fabric so they were not microservice oriented,” he says. “We immediately started to run into scale challenges with that. Existing ones are being (rewritten) onto Service Fabric.

“Another one is this evolution of the VFP and how it does packet processing. That is not something that we sat down initially and said, ‘it’s connections, not flows.’ We need to make sure that packet processing on every packet after the connection is set up needs to be highly efficient. It’s been the challenge of being able to operate efficiently, scale it up quickly, being able to deliver features into it quickly, and being able to take the load off the server so we can run VMs on it.”

What’s next for the Azure SDN? Preparing for more explosive growth of the Microsoft cloud, Russinovich says.

“It’s a constant evolution in terms of functionality and features,” he says. “You’re going to see us get more richer and powerful abstractions at the network level from a customer API perspective. We’re going to see 10X scale in a few years.”

MCTS Training, MCITP Trainnig

Best Microsoft MCTS Certification, Microsoft MCITP Training at certkingdom.com

Tech

SDN in 2014: A year of non-stop action

Leave a comment

Review of dozens of SDN moves may hint at what’s in store for 2015

The past year was a frantic one in the SDN industry as many players made strategic and tactical moves to either get out ahead of the curve on software-defined networking, or try to offset its momentum.

December
Juniper unveils a version of its Junos operating system for Open Compute Platform switches, commencing a disaggregation strategy that’s expected to be followed by at least a handful of other major data center switching players in an effort to appeal to white box customers.

November
Cisco declares “game over” for SDN competitors, and perhaps the movement itself, prompting reaction from two industry groups that the game has just begun; Alcatel-Lucent and Juniper also virtualize their routers for Network Functions Virtualization (NFV) requirements; AT&T and other unveil ONOS, an open source SDN operating system viewed as an alternative to the OpenDaylight Project’s code.

October
Cisco joins the Open Compute Project, 16 months after criticizing it as a one-trick white box commodity pony that has “weaknesses” and is destined to “lose;” Internet2 demonstrates a nationwide virtualized multitenant network, formed from SDN and 100G, that operates as multiple discrete, private networks; increased competition, largely as a result of VMware’s $1.26 billion acquisition of network virtualization start-up Nicira, goads Cisco into selling most of its stake in the VCE joint venture to EMC; Dell increases its participation in OpenDaylight after initially having doubts about the organization’s motivations; Start-up SocketPlane emerges to establish DevOps-defined networking; Cisco invests $80 into a cloud venture with Chinese telecom vendor TCL.

September
Cisco boosts its Intercloud initiative, an effort to interconnect global cloud networks, with 30+ new partners, 250 more data centers, and products to facilitate workload mobility between different cloud providers; HP opens its SDN App Store; Brocade becomes perhaps the first vendor to unveil an OpenDaylight-based SDN controller; Cisco loses two key officials in its Application Centric Infrastructure and OpenStack efforts; Cisco acquires OpenStack cloud provider Metacloud; Infonetics Research says the SDN market could hit $18 billion by 2018; SDN’s contribution to the Internet of Things becomes clearer.

July
A Juniper Networks sponsored study finds 52.5% yay, 47.5% nay on implementing SDNs; Cisco ships its ACI controller, and announces pricing and packaging of its programmable networking lineup; The IEEE forms a 25G Ethernet study group after a number of data center switching vendors with considerable operations in SDN and cloud form a consortium to pursue the technology; Big Switch Networks unveils its Cloud Fabric controller; The Open Networking User Group establishes working groups to address what it sees as the biggest pain points in networking, and issues a white paper describing the current challenges and future SDN needs; After initially claiming it wasn’t SDN, Cisco now says ACI is the “most complete” SDN; Cisco says its acquisition of cloud orchestrator Tail-f will complement its own Intelligent Automation for Cloud product.

June
Facebook unveils its homegrown “Wedge” SDN data center switch; Cisco acquires cloud orchestrator Tail-f, which gives it entrée into AT&T’s SDN project; HP unveils an SDN switch with a midplane-free chassis, similar to Cisco’s Nexus 9500; Market researchers find that SDN “hesitation” is slowing spending on routers and switches; Avaya, citing its experience at the Sochi Winter Olympic Games, describes a plan to ease implementation of SDN and other environments using its fabric technology.

May
HP clarifies its views on open source SDNs; A Goldman Sachs report concludes that Cisco’s ACI provides a 3X better total cost of ownership than VMware NSX; Cisco CEO Chambers dashes talk of Cisco acquiring cloud provider Rackspace; Cisco offers products to allow earlier generation Nexus switches to participate in a programmable ACI environment; SDN prompts more questions than answers at a Network World conference; Seven months after dismissing OpenDaylight and open source SDNs, HP raises its investment and participation in OpenDaylight; Cisco’s Noiro Networks open source project is revealed as a contributor to a policy blueprint approved for the OpenStack Neutron networking component.

April
CloudGenix debuts as the latest SDN start-up targeting enterprise WANs; Michael Dell shares his views on SDNs after his namesake company allies with SDN companies Big Switch Networks and Cumulus Networks; Juniper appears ready to accept OpenDaylight after initially dismissing it when it develops a plugin to link its own OpenContrail SDN controller to the open source code; Cisco and VMware take the SDN battle to the policy arena; Cisco unveils the OpFlex policy protocol, largely viewed as an alternative to OpenFlow and other southbound protocols, for ACI and SDNs.

March
New certifications are expected as SDN takes hold in the networking industry; three years after pledging not to enter cloud services and compete with its customers, Cisco enters cloud services through its $1 billion Intercloud initiative; Dell unveils a fabric switch and SDN controller designed to scale and automate OpenStack clouds; Cisco rolls out new chassis configurations for its Nexus 9000 switches, the hardware underlay of its ACI programmable networking response to SDN; OpenDaylight commissioned study concludes that everyone wants open source SDNs; Cumulus garners additional support for its bare metal NOS; SDN preparation may require 11 steps; Goldman Sachs says there’s nothing really new to SDNs; AT&T, NTT and others share SDN implementation experiences at Open Networking Summit 2014; Brocade becomes an early provider of OpenFlow 1.3; NEC looks to scale OpenFlow SDNs.

February
HP Networking head Bethany Mayer is tapped to lead the company’s new Network Functions Virtualization effort; Juniper expands its carrier SDN portfolio with controller and management products at Mobile World Congress; Research finds that enterprise adoption of SDNs lags that of service providers due to several factors, primarily the criticality of the network itself; Big Switch explains why it is optimistic after rebooting its SDN business; OpenDaylight announces that its “Hydrogen” SDN release is now available, after a delay; SDN start-up Pluribus Networks ships its server-switch product.

January
IBM is reported to be looking to sell its SDN business for $1 billion; JP Morgan downgrades Cisco stock based on challenges in emerging markets, and on the potential impact of SDNs; Cisco announces ACI Enterprise Module, a version of its ACI SDN controller for enterprise access and WAN programmability; ACG Research finds that sales of SDN products for live service provider deployments will reach $15.6 billion by 2018, while those that have live deployment potential will reach $29.5 billion; SDN startup Anuta Networks unveils a network services virtualization system for midsize and large enterprises; Reports surface that an SDN schism has developed at Juniper, pitting Junos and OpenDaylight programmers against CTO and Founder Pradeep Sindhu and prompting the exit of many engineers; AT&T determines that Cisco’s ACI is too complex and proprietary for its Domain 2.0 SDN project, according to an investment firm’s report.

MCTS Training, MCITP Trainnig

Best Microsoft MCTS Certification, Microsoft MCITP Training at certkingdom.com

 

 

Tech

Weighing the IT implications of implementing SDNs

Software-defined anything has myriad issues for data centers to consider before implementation

Software Defined Networks should make IT execs think about a lot of key factors before implementation.

Issues such as technology maturity, cost efficiencies, security implications, policy establishment and enforcement, interoperability and operational change weigh heavily on IT departments considering software-defined data centers. But perhaps the biggest consideration in software-defining your IT environment is, why would you do it?
Prove to me there’s a reason we should go do this, particularly if we already own all of the equipment and packets are flowing. We would need a compelling use case for it.
— Ron Sackman, chief network architect at Boeing

“We have to present a pretty convincing story of, why do you want to do this in the first place?” said Ron Sackman, chief network architect at Boeing, at the recent Software Defined Data Center Symposium in Santa Clara. “If it ain’t broke, don’t fix it. Prove to me there’s a reason we should go do this, particularly if we already own all of the equipment and packets are flowing. We would need a compelling use case for it.”

[WHERE IT’S ALL GOING: VMware adds networking, storage to its virtual data center stack]

And if that compelling use case is established, the next task is to get everyone onboard and comfortable with the notion of a software-defined IT environment.

“The willingness to accept abstraction is kind of a trade-off between control of people and hardware vs. control of software,” says Andy Brown, Group CTO at UBS, speaking on the same SDDC Symposium panel. “Most operations people will tell you they don’t trust software. So one of the things you have to do is win enough trust to get them to be able to adopt.”

Trust might start with assuring the IT department and its users that a software-defined network or data center is secure, at least as secure as the environment it is replacing or founded on. Boeing is looking at SDN from a security perspective trying to determine if it’s something it can objectively recommend to its internal users.

“If you look at it from a security perspective, the best security for a network environment is a good design of the network itself,” Sackman says. “Things like Layer 2 and Layer 3 VPNs backstop your network security, and they have not historically been a big cyberattack surface. So my concern is, are the capex and opex savings going to justify the risk that you’re taking by opening up a bigger cyberattack surface, something that hasn’t been a problem to this point?”

Another concern Sackman has is in the actual software development itself, especially if a significant amount of open source is used.

“What sort of assurance does someone have – particularly if this is open source software – that the software you’re integrating into your solution is going to be secure,” he asks. “How do you scan that? There’s a big development time security vector that doesn’t really exist at this point.”

Policy might be the key to ensuring security and other operational aspects in place pre-SDN/SDDC are not disrupted post implementation. Policy-based orchestration, automation and operational execution is touted as one of SDN’s chief benefits.

“I believe that policy will become the most important factor in the implementation of a software-defined data center because if you build it without policy, you’re pretty much giving up on the configuration strategy, the security strategy, the risk management strategy, that have served us so well in the siloed world of the last 20 years,” UBS’ Brown says.

Software Defined Data Center’s also promise to break down those silos through cross-function orchestration of the compute, storage, network and application elements in an IT shop. But that’s easier said than done, Brown notes – interoperability is not a guarantee in the software-defined world.

“Information protection and data obviously have to interoperate extremely carefully,” he says. The success of software defined workload management – aka, virtualization and cloud – in a way has created a set of children, not all of which can necessarily be implemented in parallel, but all of which are required to get to the end state of the software defined data center.

“Now when you think of all the other software abstraction we’re trying to introduce in parallel, someone’s going to cry uncle. So all of these things need to interoperate with each other.”

So are the purported capital and operational cost savings of implementing SDN/SDDCs worth the undertaking? Do those cost savings even exist?

Brown believes they exist in some areas and not in others.
We’ve got massive cost targets by the end of 2015 and if I were backing horses, my favorite horse would be software-defined storage rather than software-defined networks.
— Andy Brown

“There’s a huge amount of cost take-out in software-defined storage that isn’t necessarily there in SDN right now,” he said. “And the reason it’s not there in SDN is because people aren’t ripping out the expensive under network and replacing it with SDN. Software-defined storage probably has more legs than SDN because of the cost pressure. We’ve got massive cost targets by the end of 2015 and if I were backing horses, my favorite horse would be software-defined storage rather than software-defined networks.”

Sackman believes the overall savings are there in SDN/SDDCs but again, the security uncertainty may make those benefits not currently worth the risk.

“The capex and opex savings are very compelling, and there are particular use cases specifically for SDN that I think would be great if we could solve specific pain points and problems that we’re seeing,” he says. “But I think, in general, security is a big concern, particularly if you think about competitors co-existing as tenants in the same data center — if someone develops code that’s going to poke a hole in the L2 VPN in that data center and export data from Coke to Pepsi.

“We just won a proposal for a security operations center for a foreign government, and I’m thinking can we offer a better price point on our next proposal if we offer an SDN switch solution vs. a vendor switch solution? A few things would have to happen before we feel comfortable doing that. I’d want to hear a compelling story around maturity before we would propose it.”

MCTS Training, MCITP Trainnig

Best Microsoft MCTS Certification, Microsoft MCITP Training at certkingdom.com

 

 

Juniper

Juniper kills MobileNext mobile packet product line

Juniper MobileNext was a high-profile competitor to Cisco’s Starent gateway that was designed to enable non-interrupted delivery of high-definition voice and video over 2G/3G and LTE mobile networks

Juniper has killed a high-profile product for the core of mobile operator networks after combining business units to focus on potential growth opportunities.

Juniper has exterminated or what it calls end-of-lifed (EOL) its MobileNext mobile packet core product line, software introduced in 2009 as part of “Project Falcon” for its MX edge routers that was designed to enable non-interrupted delivery of high-definition voice and video to users over 2G/3G and LTE mobile networks. MobileNext was launched at Mobile World Congress in early 2011 to allow Juniper’s MX 3D to function as a broadband gateway, an authentication and management control plane for 2G/3G and LTE mobile packet cores, and as a policy manager for subscriber management systems.

MobileNext was intended to compete with Cisco’s ASR 5000 LTE gateway, obtained from its acquisition of Starent. But the product was struggling to gain traction in the market and was one of a handful of new Juniper products straining company financials as they went through lengthy evaluation cycles with potential customers.

Juniper is killing the entire MobileNext offering, which consists of three products: the Mobile Broadband Gateway; the Mobile Control Gateway; and the Mobile Policy Manager. The company claims, however, that its mobility strategy for the operator core remains intact.
“We have made the decision to end-of-life the MobileNext solution,” a Juniper spokesperson says. “However, our strategy remains unchanged: to virtualize mobile networks and deliver innovation through our existing portfolio of backhaul, security, routing and edge services with products such as the MX Series 3D Universal Edge Routers, SRX Series Services Gateways and JunosV App Engine software virtualization platform. We will continue to work with our partners to deliver best-in-class solutions that help customers improve network economics and accelerate delivery of new mobile services.”

Juniper will now address mobile packet core requirements through software-defined network (SDN) and network functions virtualization (NFV) capabilities, according to an internal memo authored by Daniel Hua, senior vice president of Juniper’s Routing Business Unit, and obtained by Network World.

“Despite our decision to EOL MobileNext we remain committed to executing on all existing commitments to our customers and to the mobility space longer term. We believe we can meet the needs of our customers by providing the underlying virtualized mobile infrastructure (routing, switching, SDN and NFV to enable customers to make this transition as well as offer specific virtualized network functions.”

Indeed, Juniper earlier this year announced a virtualized, SDN version of the Mobile Control Gateway based on the JunosV App Engine, which is shipping now on the MX router.

MobileNext’s demise comes as Juniper merges its Edge Services Business Unit into its Routing Business Unit. Hua explains the rationale for this in his memo:

“The compelling reason driving this organization alignment is to increase synergy and focus under the umbrella of a single routing business unit. We believe this step will ensure close alignment of our embedded and virtual services with our market-leading MX and PTX platforms. Many of the network edge services were originally developed as extensions of the Junos OS within RBU. We are realigning these services back to its original function allowing us to strengthen and further innovate in the areas of our Access, Edge, and Core offerings through tighter integration of network services.”

Sources say Juniper is also scaling down development of its Junos Content video and media delivery product line, formerly known as Media Flow and obtained from the $100 million acquisition of Ankeena Networks in 2010. Junos Content is designed to optimize mobile and fixed networks for efficient video and media delivery to smartphones and other mobile devices.

 

MCTS Training, MCITP Trainnig

Best Microsoft MCTS Certification,
Microsoft MCITP Training at certkingdom.com