Implementing QoS with Juniper

Hey folk, how you doing?!

In today's post we will be talking about integrating Quality of Service (QoS) in our Data Center solution. For that, we will be using some technology of our beloved enterprise Juniper. As we will be using this company for the integration of the QoS, we should firstly identify where we will be doing so. For that, I purpose one solution, and it is the following: it should be placed somewhere where the data leaves our place and starts to travel through the internet.

But, before we start, shall we explain why QoS is needed. This feature is mostly needed, as nowadays, internet is crowded of different network data and information, each one being more prioritary than the last one. So, how do we establish some rules to decide which traffic is more important than another? Using QoS. By using it, we can control the allocation of network attributes such as available bandwidth, latency, jitter, packet drop and bit rate errors, so that resources are managed to levels acceptable to your network customers and applications.

So, the star elements here are our MX240 edge routers, which are the last routers before leaving the data center network. So lets check how this MX240 routers implement that quality of service:

Routers MX240 allow two type of QoS implementation: Modular Port Concentrators (MPCs) and Dense Port Concentrators (DPCs). Both provide different levels of quality of service.

The Modular Port Concentrators are line modules for multiple advanced Ethernet services which they use high and modular 10/100 Gigabit Ethernet hardware. The MPCs house Packet Forwarding Engines deliver Layer 3 routing (IPv4 and IPv6), Layer 2 switching, inline services, and advanced hierarchical quality of service (H-QoS) per MX Series slot. MPCs also support Peak information Rate (PIR) and Committed Information Rate (CIR)

The most important QoS features that provide the MPCs are queue management, scheduler hierarchy, shaping, intelligent oversubscription, weighted round robin (WRR), random early detection (RED), and weighted random early detection (WRED).

On the other hand, we have Dense Port Concentrators, which are formed by DPCE-X, DPCE-Q, DPCE-R, all provide an amount of physical interfaces and Packet Packet Forwarding Engines on a single board that performs packet processing and forwarding. Each Packet Forwarding Engine consists of one I-chip for Layer 3 processing and one network processor for Layer 2. DPCE-Qs offer enhanced queuing capabilities and the QoS features of WRR, RED, and WRED.

MPCs are the newest generation hardware and have higher density and more interface options than DPC. The only thing currently that currently requires a DPC is if you need services like encryption, NAT that isn't 1:1, stateful firewall, and some other stuff.

So, we decided to use MPC for our solution as they are next-generation line modules for advanced Ethernet services, as we have already said. the MPC (Modular Port Concentrator) allows the services and features of the card to be handled separately from the physical port configuration, thus allowing greater range of configuration options for individual users.