Close Menu
TechCentralTechCentral

    Subscribe to the newsletter

    Get the best South African technology news and analysis delivered to your e-mail inbox every morning.

    Facebook X (Twitter) YouTube LinkedIn
    WhatsApp Facebook X (Twitter) LinkedIn YouTube
    TechCentralTechCentral
    • News
      How Cell C shaped Blu Label's executive pay - Brett Levy Mark Levy

      How Cell C shaped Blu Label’s executive pay

      1 October 2026
      Google scraps Gemini 3.5 Pro and turns to Gemini 4

      After falling behind in AI race, Google turns to Gemini 4

      1 October 2026
      Huawei's Mate 90 bets on 3D chip design to get around US curbs

      Huawei’s Mate 90 bets on 3D chip design to get around US curbs

      1 October 2026
      Capitec's non-bank bet is paying off, big time

      Capitec’s non-bank bet is paying off, big time

      30 September 2026
      A shrinking JSE goes looking for its next tech listings - Patrick Krappie

      A shrinking JSE goes hunting for tech SMEs

      30 September 2026
    • World
      BMW restructuring plan bets on AI and new models

      BMW restructuring plan bets on AI and new models

      1 October 2026
      OpenAI's rogue agent problem keeps getting bigger - Sam Altman

      OpenAI’s rogue agent problem keeps getting bigger

      28 September 2026
      The new battle over the desktop

      The new battle over the desktop

      23 September 2026
      AMD is now worth a trillion dollars - Lisa Su

      AMD is now worth a trillion dollars

      22 September 2026
      Anthropic weighs new model launch to blunt OpenAI's Astra surge - Anthropic CEO Dario Amodei and OpenAI CEO Sam Altman

      Anthropic weighs new model launch to blunt OpenAI’s Astra surge

      21 September 2026
    • In-depth

      10 days that changed the course of AI

      21 September 2026
      Meta to the AI industry: slow down without us - Mark Zuckerberg

      Meta to the AI industry: slow down without us

      16 September 2026
      Google DeepMind CEO Demis Hassabis. Image: John Sears

      The plan to stop AI from breaking the world

      16 July 2026
      The internet has a Strait of Hormuz problem

      The internet has a Strait of Hormuz problem

      15 July 2026
      AI boom sparks rally, frenzy and fear

      AI boom sparks rally, frenzy and fear

      11 June 2026
    • TCS
      TCS | Dominic White and Adam Ely on AI agents going rogue

      TCS | Dominic White and Adam Ely on AI agents going rogue

      29 September 2026
      TCS | Octotel's Trevor van Zyl on the fibre merger question

      TCS | Octotel’s Trevor van Zyl on the MetroFibre merger question

      16 September 2026
      Meet the CIO | Shoprite's Chris Shortt on what a supermarket becomes

      Meet the CIO | Shoprite’s Chris Shortt on what a supermarket becomes

      9 September 2026
      Rubicon's EV charging network is profitable - and growing fast - Watts & Wheels

      W&W | Rubicon’s EV charging network is profitable – and growing fast

      8 September 2026
      Winstone Jordaan on building a national EV charging network

      Winstone Jordaan on building a national EV charging network

      2 September 2026
    • Opinion
      South Africa's next energy crisis is in the accounts department - Craig Holmes

      South Africa’s next energy crisis is in the accounts department

      29 September 2026
      The steam engine lesson AI doomsayers keep missing - Sam Clarke

      The steam engine lesson AI doomsayers keep missing

      28 September 2026
      Regulating AI: apply the laws we have first - Dirk de Vos

      Regulating AI: apply the laws we have first

      21 September 2026
      What Revolut can and cannot take from South Africa's banks - Pambos Soteriades

      What Revolut can and cannot take from South Africa’s banks

      15 September 2026
      The end is nigh, and the shares go on sale in October - Duncan McLeod

      The end is nigh, and the shares go on sale in October

      14 September 2026
    • Company Hubs
      • 1Stream
      • Africa Data Centres
      • AfriGIS
      • Altron Digital Business
      • Altron Document Solutions
      • Altron Group
      • Arctic Wolf
      • Ascent Technology
      • AvertITD
      • BBD
      • Braintree
      • CallMiner
      • CambriLearn
      • CM.com
      • Contactable
      • CYBER1 Solutions
      • Digicloud Africa
      • Digimune
      • Domains.co.za
      • ESET
      • Euphoria Telecom
      • HOSTAFRICA
      • Incredible Business
      • iONLINE
      • IQbusiness
      • Iris Network Systems
      • Kaspersky
      • LSD Open
      • Mitel
      • NEC XON
      • Netstar
      • Network Platforms
      • Next DLP
      • Ovations
      • Paracon
      • Paratus
      • Publishared
      • Q-KON
      • SevenC
      • SkyWire
      • Solid8 Technologies
      • Telit Cinterion
      • Telviva
      • Tenable
      • Vertiv
      • Videri Digital
      • Vodacom Business
      • Vox
      • Wipro
      • Workday
      • XLink
    • Sections
      • AI and machine learning
      • Banking
      • Broadcasting and Media
      • Cloud services
      • Contact centres and CX
      • Cryptocurrencies
      • Education and skills
      • Electronics and hardware
      • Energy and sustainability
      • Enterprise software
      • Financial services
      • HealthTech
      • Information security
      • Internet and connectivity
      • Internet of Things
      • Investment
      • IT services
      • Lifestyle
      • Policy and regulation
      • Public sector
      • Retail and e-commerce
      • Satellite communications
      • Science
      • SMEs and start-ups
      • Social media
      • Talent and leadership
      • Telecoms
      • Watts & Wheels
    • Events
    • Advertise
    TechCentralTechCentral
    Home » Company News » Software vs hardware SDN for automatic data centre network deployments

    Software vs hardware SDN for automatic data centre network deployments

    By Huawei South Africa20 August 2021
    Twitter LinkedIn Facebook WhatsApp Email Telegram Copy Link
    Get breaking news on WhatsApp

    As modern data centres continue to evolve towards software-defined networking (SDN), network planners have encountered new challenges during initial planning. In particular, SDN now has two possible development routes, with overlay control endpoints located either on physical switches or on virtual switches in servers. Consequently, hardware SDN (sometimes called hard SDN) and software-based SDN (soft SDN) solutions stem from each of these development options, which will be discussed in further detail.

    Before outlining the differences between the hardware and software SDN solutions, let us first review the problems that SDN has solved. Once we fully understand the advantages of software and hardware SDN solutions, as well as relevant project construction requirements, we will be better prepared to make educated decisions on their suitability.

    Why use SDN?

    The use of SDN is closely related to overlay technologies, mainly virtual extensible local-area network (VXLAN) technologies. As such, we will discuss them as a whole.

    Before the introduction of SDN, networks were manually configured through the use of command lines, which is an inefficient, error-prone and expensive process, offering relatively slow time to market. During in-depth development of compute resources towards virtualisation, the layer-2 domain scale of data centre networks was insufficient. As a result, layer-2 networks could not be constructed across regions, restricting possible deployment locations and flexible migration of service virtual machines, while also hindering the improvement of large-scale data centre automation capabilities.

    SDN introduces great changes to traditional command-line deployment, with many manufacturers providing intuitive graphical user interfaces. SDN integrates with VXLAN overlay technology to associate virtual network elements with compute resources after network virtualisation, providing service-orientated networks and removing the need for tedious configuration. The frequently used spine-leaf architecture also solves the problem of traditional tree topology being unable to adapt to the rapid growth of east-west traffic.

    SDN has made significant contributions to the structural scalability of network and service automation

    SDN has made significant contributions to the structural scalability of network and service automation. Looking forward, basic capabilities that network personnel must focus on — such as high reliability, high performance, open integration and more — must also be continuously strengthened.

    How do we evaluate the advantages and disadvantages of software and hardware SDN solutions?

    As SDN develops, two primary routes have emerged: the hardware SDN solution dominated by network device vendors and the software SDN solution driven by software vendors.

    Characteristics

    We will now compare and analyse the characteristics of software and hardware SDN solutions.

    From the perspective of automation flexibility, the degree of automation depends on vendors’ understanding of network services and the logical abstraction level; it is not closely related to software and hardware modes. For example, in SDN service provisioning, North American vendors use multi-level forms to deliver service logic, while manufacturers in China have focused on graphical interaction with a higher level of abstraction and more direct logical understanding. This is where the attention of manufacturers in China should be focused. Currently, hardware SDN vendors can utilise the IT service language to automatically configure networks, even achieving higher automation levels such as pre-event and post-event verification, whereas software SDN vendors lag in these areas.

    Software SDN relies on flexible and iterative software for service development, enabling capabilities to be quickly updated without depending on hardware. In actual commercial use, strict version policies are available, and features which severely affect performance — such as encryption and encapsulation — must be offloaded to network interface card. As such, while the software approach is more flexible, many constraints exist as it relies entirely on the CPU. Therefore, it is important to focus on the specific aims and techniques of different vendors in terms of automation capabilities, regardless of the software and hardware modes.

    Moving on, network scalability is identical across both solutions, as both support the construction of large-scale SDN networks, including multi-domain, multi-data centre cascading networks.

    Software SDN can flexibly replace network resources by consuming server resources, while specifications such as tenants and virtual private clouds may exceed the hardware resource limitations of hardware SDN on switches. However, the specifications of mainstream switches do not have bottlenecks, so this advantage is not reflected in actual projects. Both solutions are evenly matched in this regard.

    Reliability and stability are top priorities in actual network deployment, as an unstable network will not be tolerated. As such, hardware SDN has a stronger reputation within the industry, delivering commercial reliability advantages in the early stages. While the software SDN solution also provides failover capabilities through vSwitches, such features are dependent on software reliability. In addition, software SDN introduces new challenges to the network operations and maintenance (O&M) team. The O&M boundary must be extended to servers, which can lead to conflicts as it involves both network and service teams. As such, most Internet vendors who opt for software SDN possess the strong technical capabilities necessary to overcome this problem.

    The industry has not yet reached a consensus on whether some offload technologies can be put into large-scale commercial use

    In terms of performance differences between the two solutions, the hardware SDN solution is superior thanks to switch hardware and the chips used for processing. However, the vSwitch — which software SDN depends on — is making continuous strides in the area of performance and has overcome some of the hurdles by adding a data plane development kit and offloading intelligent network interface cards, while increasing the data volume of each vendor to more than 10GB. Despite these improvements, the capabilities of the software SDN solution are still significantly lower than those of the hardware SDN solution.

    Bottlenecks

    The industry has not yet reached a consensus on whether some offload technologies can be put into large-scale commercial use, as many are still in the innovation and experimental phase. In addition, introducing such auxiliary technologies extends the boundary and increases the complexity and cost of software SDN. A number of insurance companies in China, having initially opted for the software SDN solution, ultimately switched to the hardware SDN solution after experiencing unacceptable performance bottlenecks.

    Another point to consider is open integration capabilities. A large number of enterprise market segments exist, and based on their unique experiences, each enterprise has different preferences for specific components and suppliers, including server vendors and models, virtualisation platforms, and cloud platforms. The software SDN solution depends on server vSwitches and must be compatible with the underlying platform software or match the virtual server model. Due to the differences between various virtualisation-layer software solutions, a single vSwitch cannot match all virtualisation-layer platforms, necessitating independent products in some cases. In contrast, the hardware SDN solution avoids such lock-in. As the control point is on the physical switch, the hardware SDN solution can generally adapt to the IT resource ecosystem of lower-layer access, such as servers from individual vendors and different virtualisation and cloud platforms.

    Finally, cost is hugely important, with unanimous agreement within the industry that the two most critical elements of software and hardware SDN solutions are reliability and cost.

    Performance problems are also considered a factor of the overall cost. The software SDN solution uses servers to implement network resources, increasing expenses due to performance overlay and scale management. Meanwhile, the hardware capabilities of network device vendors have improved to offset the hardware costs involved in the hardware SDN solution.

    Currently, the hardware SDN solution has obvious advantages in terms of cost. According to information obtained from actual projects, switches in the hardware SDN solution have offset SDN costs, and can now be obtained at a price similar to non-SDN switches as a result of rapidly developing chip integration capabilities in recent years. In contrast, the software SDN solution remains significantly expensive due to the large number of vSwitches and annual O&M fees.

    Despite the overall increased expense, software SDN remains flexible across a range of scenarios. For example, in a traditional service data centre with no service growth, customers do not intend to invest significantly in network reconstruction, and old switches do not support SDN. In this case, software SDN is a good choice to replace SDN capabilities with server resources, as it is unnecessary to waste time and resources on services that are not growing. However, if services are expected to grow, it is recommended that the hardware SDN solution be used for construction or gradual reconstruction. Of course, implementing the solution early results in a greater return on investment.

    Generally speaking, software SDN offers unique value in certain segmented scenarios, while hardware SDN is the more commonly used option

    Typically, each vendor employs a different sales strategy, and the cost reduction and competition speed of servers and switches changes dynamically. As such, overall costs vary greatly with different scenarios and projects.

    To summarise, we compared and analysed software and hardware SDN solutions in terms of flexibility, scalability, reliability, performance, open integration capability and cost. The following table lists the evaluation results:

    Generally speaking, software SDN offers unique value in certain segmented scenarios, while hardware SDN is the more commonly used option, as seen in current industry practices. In addition to the ultimate flexibility pursued by some Internet companies, most enterprises — including finance, government, and production and manufacturing industries — use hardware SDN to deploy networks in order to quickly obtain business value in their own fields.

    Indeed, competition between software and hardware SDN solutions must be determined based on application scenarios and network planning elements. From a market scale perspective, the hardware SDN solution is more widely applicable.

    • This article was written by Yang Xiaofeng, a Huawei data centre networking expert who has extensive experience in the IP field
    • This promoted content was paid for by the party concerned
    Add TechCentral as a preferred source on GoogleFollow TechCentral on Google NewsGet breaking news on WhatsApp


    Huawei SDN Yang Xiaofeng
    WhatsApp YouTube
    Share. Facebook Twitter LinkedIn WhatsApp Telegram Email Copy Link
    Previous ArticleHow collaboration is transforming the prospects of business in SA
    Next Article Apple tried to poach search chief from Google

    Related Posts

    Huawei's Mate 90 bets on 3D chip design to get around US curbs

    Huawei’s Mate 90 bets on 3D chip design to get around US curbs

    1 October 2026
    MTN Uganda and Huawei bring LinkHome to the MyMTN app

    MTN Uganda and Huawei bring LinkHome to the MyMTN app

    30 September 2026
    Standard Bank and Huawei forge strategic cloud partnership

    Standard Bank and Huawei forge strategic cloud partnership

    29 September 2026
    Company News
    B2B buyers pick a winner long before the pitch - Publishared, Michelle Losco

    B2B buyers pick a winner long before the pitch

    1 October 2026
    MTN Uganda and Huawei bring LinkHome to the MyMTN app

    MTN Uganda and Huawei bring LinkHome to the MyMTN app

    30 September 2026
    How AI voice reduces cost and improves performance in South African contact centres - 1Stream

    How AI voice reduces cost and improves performance in South African contact centres

    30 September 2026
    Opinion
    South Africa's next energy crisis is in the accounts department - Craig Holmes

    South Africa’s next energy crisis is in the accounts department

    29 September 2026
    The steam engine lesson AI doomsayers keep missing - Sam Clarke

    The steam engine lesson AI doomsayers keep missing

    28 September 2026
    Regulating AI: apply the laws we have first - Dirk de Vos

    Regulating AI: apply the laws we have first

    21 September 2026

    Subscribe to Updates

    Get the best South African technology news and analysis delivered to your e-mail inbox every morning.

    Latest Posts
    How Cell C shaped Blu Label's executive pay - Brett Levy Mark Levy

    How Cell C shaped Blu Label’s executive pay

    1 October 2026
    Google scraps Gemini 3.5 Pro and turns to Gemini 4

    After falling behind in AI race, Google turns to Gemini 4

    1 October 2026
    Huawei's Mate 90 bets on 3D chip design to get around US curbs

    Huawei’s Mate 90 bets on 3D chip design to get around US curbs

    1 October 2026
    BMW restructuring plan bets on AI and new models

    BMW restructuring plan bets on AI and new models

    1 October 2026
    © 2009 - 2026 NewsCentral Media
    Built and maintained by Chronon
    • Cookie policy (ZA)
    • TechCentral – privacy and Popia

    Type above and press Enter to search. Press Esc to cancel.

    Manage consent

    TechCentral uses cookies to enhance its offerings. Consenting to these technologies allows us to serve you better. Not consenting or withdrawing consent may adversely affect certain features and functions of the website.

    Functional Always active
    The technical storage or access is strictly necessary for the legitimate purpose of enabling the use of a specific service explicitly requested by the subscriber or user, or for the sole purpose of carrying out the transmission of a communication over an electronic communications network.
    Preferences
    The technical storage or access is necessary for the legitimate purpose of storing preferences that are not requested by the subscriber or user.
    Statistics
    The technical storage or access that is used exclusively for statistical purposes. The technical storage or access that is used exclusively for anonymous statistical purposes. Without a subpoena, voluntary compliance on the part of your Internet Service Provider, or additional records from a third party, information stored or retrieved for this purpose alone cannot usually be used to identify you.
    Marketing
    The technical storage or access is required to create user profiles to send advertising, or to track the user on a website or across several websites for similar marketing purposes.
    • Manage options
    • Manage services
    • Manage {vendor_count} vendors
    • Read more about these purposes
    View preferences
    • {title}
    • {title}
    • {title}
    🇿🇦 Sign up to the TechCentral newsletter