Playing the 'man' instead of the 'ball' does not add any credibility to your statement. If you believe there are any factual inaccuracies in the article, please substantiate.
No point in delving into details if it is going to be a monologue. Efficiency matters.
PS: Would help if your powers of observation were a bit keener, there is no 'u' in my surname as is plainly visible in the article.
My bad, I promise to pay closer attention next time. Touchy
As this article was written entirely from memory (experience), I would be very interested in what 'old text' you are referring to?
Old text that refers to MPLS as a method to improve performance by eliminating costly routing lookups. This is a myth, MPLS does not improve performance, e.g Cisco use the same express forwarding engine for IP and MPLS.
Roman said:
While the added flexibility and power provided by MPLS make it the de facto choice for next-generation network designs
Technically no, while MPLS enables certain services and it disables others. For example, the recently renamed MPLS QOS bits are only 3 bits long compared to the IP header's 8 TOS bits. It could be argued that IP provides increased QOS capability due to the 8 class limitation of MPLS. Intermediate path manipulation is another example of flexibility that is lost once labeled. To over come this limitation another add-on, MPLS traffic engineering, is introduced. That is the nature of MPLS once one service is added, more is lost than what is gained and a new work-around is developed to overcome the previous lost functionality. A vicious cycle of after thoughts. Just look at the IETF working group on MPLS, although more than a decade old, it is still work in progress. More about TE and NGN further down.
Roman said:
Unfortunately, not a simple task to distil the origins, scope, relevance & basic workings of a complex subject like MPLS into a 1000 words.
This is true, but a bit of misinformed journalism can be harmful, especially in the land of the blind where the one eyed man is king. I take jannievanzyl's positive comment serious as he is exactly the type of person (symbolic in playing a vital part in the internet landscape of SA) that reads articles like this and in the next technical meetings says: "thats it, we're rolling out mpls, I want a NGN."
Roman said:
MPLS achieves this by pre-calculating the forwarding information for every viable end-to-end path in the network...
MPLS has no calculation mechanisms and is completely dependent on what paths are already determined as best. Once a path and a binding match will it be labelled, where it is going and what path it will take it has no knowledge of and no need for.
Roman said:
MPLS-VPN is primarily built upon two other related technologies...VRF...The second is multi-protocol BGP
Often a misunderstood concept about mpls-vpn that cause MPLS and BGP VPN to be deployed on every device. In other to reduce the overall MPLS footprint, on small scale MPLS-VPN's can be deployed without MPLS or MBGP with Vrf-lite.
Roman said:
It is therefore mandatory for all communication between VRFs to use MPLS
see above. Consider also that l2tp could be used to provide communication between PE devices, eliminating the need for an end to end mpls network.
Roman said:
Ethernet over MPLS (EoMPLS) enables a port on one end of a network to exchange Ethernet packets exclusively with another on the other end.
It should be noted that EoMPLS or more accurately any transport over mpls does not require the technology to be Ethernet, end to end services could be different technologies e.g. ethernet to atm. More importantly mpls should not be deployed for this sole purpose as L2tp can achieve the same functionality without the MPLS requirement.
Roman said:
These mechanisms working in unison are collectively referred to as virtual private LAN services (VPLS)
VPLS is a completely misunderstood technology on its own and should be avoided if not absolutely required. The worst is when MPLS is deployed over ethernet medium in order to provide VPLS services. For now it will remain limitedly effective in distribution and core layers. The multicast limitation for IPTV and similar multicast services will continue to hinder the adoption of VPLS. In the future when multicast and mldp is fully supported and deployed will it be a 'maybe' in carrier access networks. Again L2tp can sufficiently provide the same service without the MPLS requirement.
Roman said:
While the service related enhancements to MPLS have garnered much of the attention, enhancements to its core forwarding intelligence and capabilities have also progressed greatly since inception. The most useful and powerful of these is MPLS traffic engineering (MPLS-TE).
Common myth about MPLS TE is that it is MPLS dependent. MPLS-TE is self efficient and does not require mpls to be deployed throughout. In other words, if MPLS-TE is required to band-aid network designs, MPLS is not required, the IP networks can still use MPLS-TE tunnels in order for path manipulation/optimization.
Roman said:
The additional structure/order imposed by MPLS-TE does however make it simpler to apply QoS policies in a more logical and effective manner
That is just completely false in every aspect.
Roman said:
In essence MPLS-TE provides the tools to squeeze ever last ounce of efficiency out of the network, ultimately leading to the maximisation of its cost effectiveness and resilience.
As mentioned the same forwarding engine is used, therefore MPLS is not more efficient at load balancing than IP is. Though correctly stated, additional knobs are provided, from a pure load balancing point MPLS-TE does not provide any additional efficiency over IP. While the additional options are provided in order to dynamically shape flow of traffic, one has to ask the question: "What is the cost/value ratio and at what point does it become viable?".
Roman said:
For example ever wonder how Telkom ensures traffic from various ISP remains segregated on a common ADSL access network? The answer of course is MPLS.
MPLS was chosen to ensure the ISP's are billed for inter client traffic to stop the 'uncapped' inter branch vpn solutions way back. Whether this was wise is debatable, from Telkom's business case perspective it was. I am not so sure it is in the long term as this design is technically inefficient and costly, which boils down to the high IP Connect costs and the Telkom network bursting at the seams. However the answer was never only MPLS, true bitstream was an option, since all the DSLAM's until recently were on ATM. Which brings me to the following quote:
Roman@MWEB said:
Personally, I find the wiki page quite daunting & confusing from the perspective of a non-technical person. The comparisons/references to technologies like ATM & Frame-Relay have no relevance in the year 2010.
Without understanding Frame and ATM one can not write articles with the pretense to be an authoritative voice on MPLS. ATM and Frame-relay at the core are the first two label switched technologies. Everything that is MPLS today evolved from the label concept of FR and ATM. Look at many poor network designs and it is clear ATM/SDH and even Ethernet can achieve/replace a lot of what MPLS is used for without loosing any flexibility and gaining all that is IP.
This is the main problem I have with these articles. Non-technical people 'mythbusting' technical concepts. The article is not just keeping with clearing the general misconceptions about mpls but further sells the technology as the 'de facto choice for next-generation network designs'. Coming from a reputable provider readers would accept this as fact and another myth is started. Claiming to clear myths further contributes to the perception that this is the gospel while in fact it is only creating more, hence my comment that the M for 'misunderstand' applies to the author. NGN is a buzz word used to sell products, the design could comprise of many different technologies. The NGN term evolved when the term 'Convergence' lost its marketing punch (SS7/PSTN moving to IP). MPLS is not the requirement for either, but could be one element of the complete solution. Telkom erroneously use the term NGN to describe their network when most of their Voice services are still PSTN. We need to be moderately concerned if every MPLS enabled network becomes so called NGN's. Soon vendors will invent and market a new term with equal enthusiasm and we will have the G4s vs G3.75 debate all over again.
My conclusion/agenda: In SA, MPLS, VPLS, MPLS-TE and NGN's are the nice to have fashion flavors of the moment. MPLS solutions are
hyped,sold and deployed when and where not required increasing the cost of networks in SA. Eventually a client will pay the bill for hot air which is counter productive to the whole objective of reducing costs, investing in expansion, growing the economy.
MPLS is an explosive technology and just like fuel it is most useful in the combustion chamber. Anywhere else you run the risk of setting more than just the fuel on fire.
felt like i was studying for my cisco exams all over again
more alphabet soup that helps (and some ..actually a whole LOT that doesn't) at 