Digital & Communication Technology
Baby Steps to better WAN performance
By Ricardo Belmar, senior director, global enterprise marketing, Infovista S
uccess for newly launched digital services can result in massive uptake which in turn can impact
performance. The factors that impair performance can include application design, stretched databases as well as connectivity. As more applications move from client based to centrally delivered from the data centre or cloud, the poor performance of the Wide Area Network can have a disproportionality negative impact on application responsiveness and user experience.
Less data and more bandwidth The perceived antidote has been to throw more bandwidth at the problem. And as the cost of WAN connectivity has declined, this has delivered some benefits but has increasingly hit a wall in terms of gains. As fast as organisations have expanded the WAN bandwidth, new applications and business drivers have arrived to fill up the spare capacity. For example, video conferencing takes up 20x the bandwidth of a voice call. While enterprise requirements such as real-time data backup from remote sites still represent a constant drain on WAN bandwidth. The first evolution of technologies to improve WAN performance focused purely on compression – based on the principle that a lot of the data is redundant and can be mathematically altered to consume less bandwidth without losing fidelity. In the early 2000’s, this paradigm ruled – and is still valuable today but again runs into technical limitation – and some trade-offs. Although lots of legacy WAN traffic can be compressed to use less bandwidth, elements such as video, voice, and many large file transfers are already compressed. As such, the utility of compression delivers
less value with modern data flows. The other issue is that certain types of traffic are more sensitive to interruption and latency which means that it needs to take priority when flowing across the network.
Added intelligence The next baby step to alleviate WAN performance issues were branded under network optimisation that combine technologies such as compression along with more intelligence around which type of traffic should take priority. WAN technologies such as Multiprotocol Label Switching (MPLS) already has a level of intelligence through the establishment of pre-determined, highly efficient routes for traffic to pass along. However, the technology was designed two decades ago, for an era when branch offices sent traffic back to a main headquarters or data centre, not for today’s world where branch offices need direct access to the cloud. However, MPLS networks are far from dead. The managed MPLS market was valued at USD 45.61 billion in 2018 and is estimated to register a CAGR of 6.5% during the forecast period (2019-2024). The reasons are multiple and centre on its reliability and security as a private, carrier provided network that is segregated from the internet. MPLS is also highly scalable and bandwidth efficient - yet, these
benefits come at much higher cost than business broadband.
The cost of MPLS is highly dependent on how much capacity you purchase and global location of the sites that must connect to the WAN. ADSL technologies that connect via the public internet may cost 100x less per MB than MPLS but are less able to deliver the base line predictability of MPLS.
Cloud and SaaS The most recent baby steps to WAN nirvana has aligned with two societal trends. The first is the growth of internet delivered applications and services – often shorthanded as Cloud and Software as a Service (SaaS). The other trend is the rise of software as a replacement for highly specialist physical appliances – often called Software Defined.
Many organisations that previously ran massive data centres delivering large monolithic applications across their own WANs are instead switching to applications like Salesforce, SAP and Microsoft Office that stream across the public internet. The big innovation in the quest to
improve this evolved WAN challenge is the recognition that all traffic is not equally important. And providing that security is maintained, a large portion of the typical WAN traffic that previously flowed over the expensive MPLS can instead flow over much lower cost business broadband links. To control what application data, flows across which links is increasingly becoming the job of software – hence the name of the most recent networking technology Software Defined WAN.
36 December/January 2020 Components in Electronics
When software met WAN SD-WAN as a concept is roughly a decade old and is more a philosophy than a set of internationally agreed technical standards that MPLS embodies. SD-WANs overarching aim is to separate the networking hardware from its control mechanism. At one end of the SD-WAN parading is the notion that MPLS is dead and organisations should ditch these private networks and instead deploy lots of bandwidth using multiple connection paths such as ADSL, FTTP, FTTC and 4G that route data over the open internet. Smart software acts as the traffic manager switching between different connections to deliver reliability, quality of services and performance.
Although a likely destination, the current evolution is more of a hybrid SD-WAN approach that recognises that larger enterprises value the reliability of the private MPLS networks. Instead, hybrid aims to supplement MPLS with additional lower cost connectivity with software dynamically managing the flow of traffic over the top of the various connections – but without eliminating the physical appliances from the mix. Hybrid is gaining ground faster and looking at the more distant future, may act as a testbed as a transition to a full SD-WAN migration. The criticality of WAN performance will only become more acute as the rise of cloud and SaaS accelerates. The notion that simply throwing more bandwidth at the problem may well solve the short-term issues while thinking more intelligently about the wider WAN strategy is likely to turn a baby step into a giant leap forward.
www.cieonline.co.uk
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56
