Why Deep Buffer Matters for Next-Gen Metro Networks?
Why deep buffer matters.
Latency is crucial for Juniper’s vision to drive the next-gen Cloud Metro and provide a common framework to support multiple types of architecture. Check out this video from the Juniper Cloud Metro AMA series, where our experts answer your questions related to next-gen Cloud Metro network.
You’ll learn
How Virtual Output Queue (VOQ) provides both power density and great latency
How to optimize VOQ while supporting a variation of topologies
How microburst absorption can help to manage traffic flow
Who is this for?
Host
Guest speakers
Transcript
00:04 Welcome to Juniper Cloud Metro AMA series, where we have
00:07 experts to answer your questions, any questions related
00:11 to next-gen Metro networks. I'm your host today, Irene Zhang,
00:15 Director of Product Marketing at Juniper Networks. Joining me
00:18 today is Peter Chung, senior product manager at Juniper.
00:22 Peter, welcome. Today we have a question about why deep buffer
00:27 matters for next-gen metro networks?
00:30 Yeah, that's a great question that most of you frequently ask.
00:33 So, why we need a deep buffer for our next generation of the
00:37 metro networks. Our foundation of the building block is try to
00:41 provide very common framework. So, basically, this common
00:46 framework will give you the very flexibility to provide a
00:49 different variation for the topology, you can have a ring,
00:53 you can have a hub-spoke or leaf-and-spine architecture.
00:57 So our vision to drive the Next-Gen Cloud Metro. Latency is
01:01 very important.
01:02 So we picked the merchant silicon based on the VOQ
01:06 architecture called Virtual Output Queue. which give you the
01:11 great benefit for the optimization from the silicon to
01:15 keep your power density, but in the meantime, we'll provide you
01:19 great latency.
01:20 So, how to balance between the VOQ performance capacity and the
01:26 able to support variation of topology. So, keep in mind, our
01:32 Cloud Metro not just support for the traditional Metro topology,
01:36 but also to support 5G.
01:38 So, you can picture your scenario, your 5G network have
01:42 the very low speed application, but try to limit your new Metro
01:46 100GE or even for 400GE up speed uplink. So now the deep buffer
01:56 is very important.
01:58 Thanks to the silicon evolution, the latest generation chipset
02:03 gives you the delicate on chip resource, and more importantly,
02:06 is to offer you up to 8G HBM to memory.
02:11 So, those kinds of new enhancement for the silicon and
02:15 overall density of the memory will be increased. So basically
02:20 with this kind of big powerful mechanism, we can guarantee end
02:23 to end of transmission. More importantly, how to handle a
02:27 symmetric of traffic flow, you can do the micro-burst
02:30 absorption.
02:31 Our ACX, the latest generation of the metro product family is
02:36 based on the most advanced version of the silicon
02:39 technology. It gives you the 8G buffer for here.
02:43 This is our checkpoint for support the very first step to
02:47 support the Cloud Metro. Thank you.
02:52 What else do you want to learn about the next generation Metro
02:54 networks? Send us your question by email at metroAMA@juniper.net
03:00 and join us at our monthly live AMA session. You can find the
03:04 date and registration link at the description box below. See
03:09 you next time.