Network Virtualization-Global Digital Infrastructure Trends

Posted on 13. Dec, 2017 by in Cloud, desktop, Developement, Hosting, internet, linux, Programming, Uncategorized, Web Hosting, windows

Global Digital Infrastructure Trends

December 13, 2017


Quick Snapshot: Today’s Newsletter

Virtualizing the Network. While server and storage virtualization have both evolved significantly over the past several decades, physical switches have remained essentially the same since Cisco introduced its first network switches in 1993. The status quo is finally changing.

Virtualizing the Remote Branch. SDN is extending beyond the datacenter and out into the branch. Cisco and VMware are both jockeying to shore up their position in the SD-WAN marketplace.

Apple Holds its Lead in Tablet Market. The introduction of two new iPad models (9.7” iPad, 10.5” iPad Pro) and hardware upgrades to the 12.9” iPad Pro, along with significant productivity-facing updates in iOS 11, are helping Apple keep a tight grip on the market.

IBM Ups the Ante on Db2. IBM recently rolled out an updated version of Db2 for cloud and on-premises, as well as integrated tools. The new Integrated Analytics System complements the expanding Db2 line.


I. Software-Defined Datacenters: The Networking Angle

While server and storage virtualization have both evolved significantly over the past several decades, physical switches have remained essentially the same since Cisco introduced its first network switches in 1993.

However, when server virtualization began to gain traction in large datacenters in the mid-2000s, three serious networking challenges had to be addressed.

  • Underutilization. Physical switches typically ran at only a fraction of their potential capacity.
  • Virtualization. Switches weren’t designed to cope with virtualized resources and dynamic, moving workloads.
  • Architecture. For virtual and cloud infrastructure, the tiered networking approach was inefficient – there were too many hops between one VM and another.

Besides the traditional ‘north-south’ traffic up to top-of-rack switches and out over the wider network, scale-out infrastructure required support for ‘east-west’ server-to-server traffic to enable faster VM migration and to speed up cloud performance.

To meet these challenges, open source initiatives (e.g., OpenFlow, Open vSwitch) led the evolution to software-defined networking (SDN).

SDN can potentially:

  • Eliminate 75% of network connections
  • Divide bandwidth to suit demand and workloads
  • Provide an on-ramp to converged fabrics

Software-Defined Networking (SDN)

One of the main values of SDN is a simplification of the way networks are interconnected.

In physical networks, lashing systems together means chasing physical cabling between chassis and boxes. Keeping track of what went where creates a significant documentation burden, and changing connections can get complex.

Network virtualization provides the ability to change interconnections without physical access to systems, and enables VMs to remain independent of physical location and free from dependencies on the physical network infrastructure.

To achieve this, extensions to the networking stack must provide VM-level visibility and granular network configuration and control. An example of this is the Virtual Extensible LAN (VXLAN) protocol, which is a means of supporting virtual LANs through the use of overlay networks.

Network Functions Virtualization (NFV) vs. SDN

NFV is almost exclusively carrier-driven to virtualize network services. The first installations are now reaching production, although most are still in the early stages.

While SDN is about the abstraction of the control and data planes from networking equipment, NFV focuses on the separation of network functions from hardware so that the functions themselves can run in software (see Figure below).

These functions span the entire ecosystem of network technology and services, including:

  • Load balancers
  • Deep Packet Inspection (DPI)
  • Firewalls
  • Policy servers
  • Domain Name Systems (DNS)
  • Network address translation
  • IP Multimedia Subsystems (IMS)

Software-Defined WAN (SD-WAN)

Another prominent application for SDN is software-defined WAN (SD-WAN).

WANs connecting branch offices to larger sites have been around for decades, although they haven’t changed much in that time.

Their static, rigid and architecturally inflexible nature makes them ill-suited to the dynamic on-demand characteristics of cloud computing, mobility, IoT, and as-a-service network and security offerings.

WAN technology can benefit significantly from the added agility of a software-defined approach.

SD-WANs extend SDN to the network edge, and can be provided as cloud-based services or implemented on-premises across LTE and MPLS networks.

This provides superior functionality for adjusting bandwidth in real time, and can help businesses control costs by dynamically steering traffic to optimize bandwidth.

This makes SD-WAN a much better match for the needs of cloud and IoT use cases compared with non-software-defined approaches that can be expensive to provision.

Cisco’s Application Centric Infrastructure (ACI) and VMware’s Network Virtualization and Security Platform (NSX) are currently the most prominent commercial SDN products, providing network virtualization that emphasizes security and workflow automation. (For more information on VMware’s NSX, Cisco’s ACI and SD-WAN, see the next article in this newsletter.)

The OpenFlow standard was a strong early influence on SDN designs, and the Linux Foundation’s Open Daylight Project provides an open source reference.

A Work In Progress

Network incumbent Cisco has never seen such a broad-ranging challenge to its core networking business – from cloud, software-defined technologies and convergence right through to distributed IoT – and it’s adapted accordingly.

SDN concepts have begun to transform the enterprise WAN in many ways, but perhaps the most exciting software-defined transformation yet has been the SD-WAN, because of the clear and immediate benefits to the customer.

Meanwhile, a raft of telco equipment suppliers are looking to profit from industry-wide transformation, and to increase their ‘software IQ’ as buyer demand for service-driven business outcomes increases over traditional technology outcomes. Vendors coming from the network equipment side of the house, however, all face the key challenge of establishing credibility as a platform for delivering IT, as well as communication services.

II. VMware Plugs Gaps in its SD-WAN Portfolio

VMware is in the process of acquiring software-defined WAN (SD-WAN) startup VeloCloud Networks in a bid to bolster its network virtualization offerings in the red-hot SD-WAN market.

The acquisition should allow VMware to expand its networking, automation and security capabilities beyond the datacenter, to the enterprise edge and branch offices, and into the cloud.

The deal will ramp up VMware’s rivalry with Cisco in the SDN and network virtualization market.

Cisco bought VeloCloud competitor Viptela in May, and recently announced a partnership with Google Cloud for hybrid cloud integration.

VMware and Cisco are already bitter combatants in the datacenter SDN arena, and now they’re on a collision course in branch-office networking, as well.

Their common goal is to automate and enforce application and security policy from the datacenter to the cloud.

Deal Rationale

VMware has been eyeing the remote-office/branch-office opportunity for a couple of years, forging partnerships with VeloCloud, CloudGenix and other SD-WAN startups.

Most of these partnerships were aimed at extending and enforcing NSX micro-segmentation and security policies to remote sites.

The ultimate goal for SD-WAN is to become part of a continuum of connectivity that links all of an enterprise’s infrastructure – from branch offices through on-premises environments and on to cloud and SaaS providers.

VMware believes that the future of networking will be in software, and VeloCloud lines up closely with its vision.

Branch networking is the next big market for VMware, and it sees an opportunity to redefine branch offices through cloud delivery and management, much of it automated and orchestrated by NSX and vCloud.

VeloCloud will become a core component of VMware’s Cross-Cloud service management strategy, a fabric that encompasses datacenters, branch offices and clouds. VeloCloud has roughly 600 customers, including many tier one service providers, such as AT&T, Sprint and Telstra.

VMware expects to simplify the deployment of virtual network functions (VNFs) for applications such as security by combining VMware vCloud platform with a cloud-delivered SD-WAN platform.

VeloCloud was already certified as an SD-WAN VNF platform on operator networks, most recently Telstra’s.

The integrated portfolios of VMware and VeloCloud could offer providers a more complete VNF service that includes access capabilities and steps beyond dependence on specific hardware: The virtual customer premises equipment (vCPE) movement has been gaining support as operators look to minimize the complexity of service deployment with VNFs hosted on off-the-shelf x86 hardware.

Competition Heats Up

In the SDN space, VMware competes primarily with Cisco.

>VMware’s NSX and Cisco’s ACI are the leading datacenter SDN platforms, and the two companies have been bolstering their SDN and cloud capabilities over the past five years.

The most recent developments are VMware’s acquisition of VeloCloud and Cisco’s partnership with Google Cloud for hybrid public/private cloud integration (which could be viewed as a response to VMware’s partnership with AWS, which was announced in September).

Additionally, Cisco bought SD-WAN startup Viptela earlier this year for its cloud-based enterprise branch networking capabilities.

In addition to vying with Cisco’s Viptela in the SD-WAN space, VeloCloud competes with Versa Networks, FatPipe Networks, Cisco’s Meraki, Riverbed, Citrix, Silver Peak, Cradlepoint, Talari, Aryaka, Apcela, CloudGenix, Huawei, Nuage Networks, Gluware and Cato Networks.

The 451 View

Branch-office networking was a missing component in VMware’s SDN portfolio of datacenter and cloud offerings. VeloCloud should fill that gap.

The VeloCloud model suits VMware’s software networking ethos, and could be a natural extension of the NSX product architecture, if properly integrated.

VMware’s challenge will be to ensure a consistent experience from the datacenter to the cloud for enterprises and service providers via the integration of VeloCloud, NSX and vRealize Network Insight.

III. Apple iPads Drive Seasonal Uptick in Tablet Demand

Apple remains the leader in tablets as it drives a seasonal increase in buying – although overall consumer demand has weakened with each successive year. Interest in iPads is strong among planned tablet buyers over the next six months, and satisfaction among existing iPad owners is higher than the industry average.

The introduction of two new iPad models (9.7” iPad, 10.5” iPad Pro) and hardware upgrades to the 12.9” iPad Pro, along with significant productivity-facing updates in iOS 11, are helping Apple keep a tight grip on the market.

Our latest survey (completed August 21) of 1,920 primarily North American respondents from 451 Research’s Leading Indicator panel focused on key tablet market trends, including buying plans and satisfaction. The survey also took a close look at the use of 2-in-1 laptops and PC buying.

Consumer Tablets – Key Findings

In the latest survey, 6.4% of respondents say they will purchase a new tablet in the next six months. Although the findings show a slight seasonal uptick in tablet buying, the overall trend for the market remains negative – down 1.3 points year over year.

Leading Tablet Manufacturers. First-place Apple is attracting 72% of tablet buyers, with the 10.5” iPad Pro at the top of the list for 41% of planned Apple buyers.

Samsung (11%) and Microsoft (8%) are a distant second and third.

Tablet Customer Satisfaction. Apple’s iPad Pro leads the tablet space in satisfaction, registering the highest Very Satisfied ratings. Microsoft’s Surface Pro also posted solid ratings.

Tablet Purchase Channels. iPad and Surface tablets are largely purchased Directly from the manufacturer, while Wireless carriers are more important for Galaxy tablet buying than other brands.

Where did you purchase your current tablet?

Device Upgrade Cycle

The tablet upgrade cycle now sits at 4.1 years, with consumers holding onto their devices an additional 2.4 months compared with August 2016. Note that consumers are replacing their laptops every 4.5 years and their desktops every five years – similar to a year ago.

IV. IBM Pushes Beyond Data Warehousing with Integrated Analytics

IBM’s recently announced Integrated Analytics System is based on Db2 and Netezza technology, but it’s much more than a data-warehousing system – it includes a number of capabilities to enable advanced data science activities, including:

  • Apache Spark
  • The IBM Data Science Experience (advanced analytics and machine learning)
  • Potential integration with Hadoop-based systems


Even though Watson gets the lion’s share of attention, IBM has not forgotten its mainstay systems, especially the Db2 line.

The company recently rolled out an updated version of Db2 for the cloud and on-premises, as well as integrated tools. The former dashDB product line has been rebranded as Db2 on Cloud and Db2 Warehouse on Cloud.

The new Integrated Analytics System complements the expanding Db2 line.

However, it’s important to understand IBM’s broader strategy and how the Integrated Analytics System fits in.

At a high level, IBM is taking a hybrid approach, but it’s more than offering a choice of cloud and on-premises computing. Instead, IBM defines ‘hybrid’ as not only offering different environments (on-premises and cloud), but also offering different choices of technologies and enabling different data types.

With its hybrid strategy, IBM sees organizations leveraging a number of systems linked together by a set of common tools.

For instance, at the systems level, organizations can have Db2 as a:

  • Relational system deployed on-premises or in the cloud
  • Managed cloud service (Db2 on Cloud)
  • Data warehouse deployed in the cloud or on-premises (Db2 Warehouse on Cloud or Db2 Warehouse)
  • Analytics appliance on-premises (the Integrated Analytics System)

The systems are linked via the company’s Big SQL engine. Via virtualization, data residing on any system can be viewed and queried from any other system.

Furthermore, architectural commonalities between the systems enable analytics applications written on one system to be moved to another system without additional coding.

Integrated Analytics System

The Integrated Analytics System is an on-premises system optimized specifically for advanced analytical workloads, particularly machine learning. It includes a cloud-ready platform so that applications written for the platform can be deployed in cloud (public and private) environments.

The Integrated Analytics System combines elements of Db2 Warehouse on Cloud (formerly dashDB), including BLU Acceleration and PureData Systems for Analytics (formerly Netezza), which means transitioning from these systems is a fairly straightforward task.

However, to call the Integrated Analytics System the next-generation Netezza or dashDB is not accurate, mainly because of the additional analytical capabilities that have been added.

Specifically, it includes the IBM Data Science Experience along with Apache Spark, which enables advanced data science activities – such as machine learning – to be carried out directly within the platform, as opposed to moving them to a different system.

From a hardware standpoint, the Integrated Analytics System is based on IBM’s Power architecture running Linux. Flash memory is standard, with tiered storage due in 1H 2018, which will enable mixing operational and analytical workloads.


While cloud gets most of the attention, a number of vendors offer on-premises systems for business intelligence (BI) and advanced analytics. In this space, IBM will compete primarily with the following vendors:

On the cloud front, IBM squares off against a number of other competitors:

While IBM sees a few of its longtime competitors still delivering on-premises systems, perhaps its greatest challenge will come from cloud-specific platform providers that offer database services with integrations to data science services.

Rate this ➜

1 person likes this.

Tags: , , , , , , , , ,

No comments.

Leave a Reply