Week4 Flashcards
(16 cards)
What are motivations for an enterprise wanting to offload NF processing to the Cloud?
- Leverage economy of scale to cut costs
- Simplify management (no need for training personnel, upgrades are handled by cloud provider, Low-level configuration of NFs is replaced by policy configurations - avoid failures due to misconfiguration)
- Elastic scaling - scale in/out works much better on cloud vs on premise (avoid failures due to overload).
What is the advantage of using bounce redirection?
Does not require any modification to the enterprise or the client applications
What is the disadvantage of using bounce redirection?
There is an extra round trip to the cloud.
Thus, can only be feasible if cloud point-of-presence is located close to enterprise
What is the basic idea of Bounce redirection?
It is the simplest form of redirection. Tunnel ingress and egress traffic to the cloud service from the enterprise.
- In more detail:
1. External traffic comes into the enterprise
2. Traffic is bounces to the cloud provider for ingress processing.
3. Traffic is bounced back to the enterprise gateway for the enterprise processing
4. After enterprise processing, the enterprise gateway will direct the traffic to the cloud provider for the egress processing that may need to be done.
5. Traffic will bounce back to the enterprise gateway.
6. Finally, enterprise returns the result back to the original external site that made the request in the first place.
What is the basic idea of IP redirection?
Similar to bounce redirection, but saves extra round-trip by sending client traffic directly to the cloud service.
How?
Cloud service announced IP prefix on behalf of the enterprise.
That way, when the client generate a traffic intended for the enterprise it’ll automatically get routed to the particular cloud site, where the network functions will get evaluated, and then the processed result will be sent to the enterprise for further processing.
What are the drawback of IP redirection?
- A cloud provider have multiple PoP (point of presence), hence you cannot ensure that same PoP receives both flows a->b and b->a (because all PoPs advertise the same IP address range) - a problem if flow-affinity or connection-affinity is required.
- Since traffic is directed using Border Gateway Protocol (BGP), no guarantee of selecting PoP that minimizes end-to-end latency
What is DNS-based redirection?
Cloud provider runs DNS resolution on behalf of enterprise.
In more detail:
- Cloud provider registers with a DNS server, and acts as the surrogate for the enterprise.
- Any external site that wants to send traffic to enterprise will do a DNS lookup of the enterprise and will automatically get redirected (unbeknownst to the client) to the cloud provider.
- The network functions will get run in this cloud provider.
- The resulting package will be sent to the enterprise GW.
- For the egress traffic we’ll use the same IP address of the cloud provider that is provided by the DNS service - same PoP is retrieved!
- Finally, the same PoP routs the traffic back to the eternal site.
What is the drawback of using DNS based redirection?
Loss of backwards compatibility in the sense that there are legacy enterprise applications that may actually be exposing IP addresses to external clients and not a DNS name.
*Nonetheless the state of the art is DNS based redirection
What is one factor that heavily influences the choice of a Cloud provider for offloading NF processing?
Geographical footprint expressed in terms of RTT from enterprise.
In other words, we want to choose a cloud PoP that minimizes:
latency(PoP, client) + latency(PoP, enterprise)
Explain the downside of moving certain network functions into the cloud.
Middleboxes like Web proxy (HTTP proxy) and WAN accelerator are used to limit WAN bandwith used by enterprise. For example, HTTP Proxy limits WAN bandwidth usage by caching web pages. If we move them to the cloud, WAN bandwidth becomes high for the enterprise.
- **Safest solution is to not migrate those types of middleboxes!
- **Alternatively,
What are the two categories of packet processing performed in cellular access networks?
- Analog radio function processing (RF processing): D2A/A2D converter, filtering and amplification of signal
- Digital signal processing: L1, L2 and L3 functionality (namely the physical link and the network layer functionality).
What are the benefits of separating the baseband processing from RF processing in 3G/4G networks?
- Lower power consumption since RF functionality can be placed on poles/rooftops => efficient cooling.
- Multiple BBU (Base Band Unit) can be placed together in a convenient location => cheaper maintenance.
- One BBU can server multiple RRHs (Remote Radio Head).
What are the limitations 3G/4G networks design?
- Static RRH-to-BBU assignment => Resource can be underutilized.
- BBUs are implemented as specialized hardware => poor scalability and failure handling
What are motivations to move from the current 3G/4G access networks design to a Cloud-RAN design?
- Allows elastic scaling of BBUs based on current workload
- BBU-RRH assignment is dynamic, leading to higher resource utilization.
- RAN stands for Radio Access Network.
- Cloud RAN virtualizes the BBUs in a BBU pool - and BBU now implemented as software running on general purpose servers.
Why does the BBU pool need to be located in close proximity to cellular radios?
For low latency, low jitter and high throughput
What does MEC stands for?
Multi-access Edge Computing
